diff --git a/Samples/0_Introduction/UnifiedMemoryStreams/README.md b/Samples/0_Introduction/UnifiedMemoryStreams/README.md index f4d669ab..806277a9 100644 --- a/Samples/0_Introduction/UnifiedMemoryStreams/README.md +++ b/Samples/0_Introduction/UnifiedMemoryStreams/README.md @@ -31,42 +31,5 @@ cudaStreamDestroy, cudaFree, cudaMallocManaged, cudaStreamAttachMemAsync, cudaSe Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/0_Introduction/asyncAPI/README.md b/Samples/0_Introduction/asyncAPI/README.md index 0771cf13..f7b7d6a5 100644 --- a/Samples/0_Introduction/asyncAPI/README.md +++ b/Samples/0_Introduction/asyncAPI/README.md @@ -29,42 +29,5 @@ cudaProfilerStop, cudaMalloc, cudaMemcpyAsync, cudaFree, cudaMallocHost, cudaPro Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/0_Introduction/clock/README.md b/Samples/0_Introduction/clock/README.md index d7e1a47d..d21e7528 100644 --- a/Samples/0_Introduction/clock/README.md +++ b/Samples/0_Introduction/clock/README.md @@ -29,42 +29,5 @@ cudaMalloc, cudaMemcpy, cudaFree Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/0_Introduction/clock_nvrtc/README.md b/Samples/0_Introduction/clock_nvrtc/README.md index f42090ea..5f78f5fd 100644 --- a/Samples/0_Introduction/clock_nvrtc/README.md +++ b/Samples/0_Introduction/clock_nvrtc/README.md @@ -36,42 +36,5 @@ cudaBlockSize, cudaGridSize Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, aarch64. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=aarch64`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/0_Introduction/cudaOpenMP/README.md b/Samples/0_Introduction/cudaOpenMP/README.md index 63ac988c..09b77331 100644 --- a/Samples/0_Introduction/cudaOpenMP/README.md +++ b/Samples/0_Introduction/cudaOpenMP/README.md @@ -33,42 +33,5 @@ cudaMemcpy, cudaGetErrorString, cudaFree, cudaGetLastError, cudaSetDevice, cudaG Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/0_Introduction/fp16ScalarProduct/README.md b/Samples/0_Introduction/fp16ScalarProduct/README.md index b91345aa..6f642d9d 100644 --- a/Samples/0_Introduction/fp16ScalarProduct/README.md +++ b/Samples/0_Introduction/fp16ScalarProduct/README.md @@ -33,42 +33,5 @@ cudaMemcpy, cudaFree, cudaMallocHost, cudaFreeHost, cudaMalloc, cudaGetDevicePro Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/0_Introduction/matrixMul/README.md b/Samples/0_Introduction/matrixMul/README.md index d6c029b3..f473acc5 100644 --- a/Samples/0_Introduction/matrixMul/README.md +++ b/Samples/0_Introduction/matrixMul/README.md @@ -29,42 +29,5 @@ cudaStreamCreateWithFlags, cudaProfilerStop, cudaMalloc, cudaFree, cudaMallocHos Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l, aarch64. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
`$ make TARGET_ARCH=aarch64`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/0_Introduction/matrixMulDrv/README.md b/Samples/0_Introduction/matrixMulDrv/README.md index 0b141e1a..3b4d48e3 100644 --- a/Samples/0_Introduction/matrixMulDrv/README.md +++ b/Samples/0_Introduction/matrixMulDrv/README.md @@ -29,42 +29,5 @@ cuMemcpyDtoH, cuLaunchKernel, cuMemcpyHtoD, cuDeviceGetName, cuDeviceTotalMem, c Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l, aarch64. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
`$ make TARGET_ARCH=aarch64`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/0_Introduction/matrixMulDynlinkJIT/README.md b/Samples/0_Introduction/matrixMulDynlinkJIT/README.md index 9428da91..7b8fc55b 100644 --- a/Samples/0_Introduction/matrixMulDynlinkJIT/README.md +++ b/Samples/0_Introduction/matrixMulDynlinkJIT/README.md @@ -29,42 +29,5 @@ cuMemcpyDtoH, cuDeviceGetName, cuParamSeti, cuModuleLoadDataEx, cuModuleGetFunct Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/0_Introduction/matrixMul_nvrtc/README.md b/Samples/0_Introduction/matrixMul_nvrtc/README.md index 00dc0ae7..7d5d7b9e 100644 --- a/Samples/0_Introduction/matrixMul_nvrtc/README.md +++ b/Samples/0_Introduction/matrixMul_nvrtc/README.md @@ -33,42 +33,5 @@ cuMemcpyDtoH, cuLaunchKernel, cuMemcpyHtoD, cuCtxSynchronize, cuMemAlloc, cuMemF Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, aarch64. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=aarch64`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/0_Introduction/mergeSort/README.md b/Samples/0_Introduction/mergeSort/README.md index 0865f1ae..e75462a2 100644 --- a/Samples/0_Introduction/mergeSort/README.md +++ b/Samples/0_Introduction/mergeSort/README.md @@ -29,42 +29,5 @@ cudaMalloc, cudaDeviceSynchronize, cudaMemcpy, cudaFree Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/0_Introduction/simpleAWBarrier/README.md b/Samples/0_Introduction/simpleAWBarrier/README.md index 699b2d05..a36faa92 100644 --- a/Samples/0_Introduction/simpleAWBarrier/README.md +++ b/Samples/0_Introduction/simpleAWBarrier/README.md @@ -33,42 +33,5 @@ cudaStreamCreateWithFlags, cudaFree, cudaDeviceGetAttribute, cudaMallocHost, cud Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l, aarch64. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
`$ make TARGET_ARCH=aarch64`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/0_Introduction/simpleAssert/README.md b/Samples/0_Introduction/simpleAssert/README.md index 9db26069..1cae293d 100644 --- a/Samples/0_Introduction/simpleAssert/README.md +++ b/Samples/0_Introduction/simpleAssert/README.md @@ -29,42 +29,5 @@ cudaDeviceSynchronize, cudaGetErrorString Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/0_Introduction/simpleAssert_nvrtc/README.md b/Samples/0_Introduction/simpleAssert_nvrtc/README.md index 5a633d76..3f15c89e 100644 --- a/Samples/0_Introduction/simpleAssert_nvrtc/README.md +++ b/Samples/0_Introduction/simpleAssert_nvrtc/README.md @@ -33,42 +33,5 @@ cuModuleGetFunction, cuLaunchKernel, cuCtxSynchronize Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, aarch64. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=aarch64`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/0_Introduction/simpleAtomicIntrinsics/README.md b/Samples/0_Introduction/simpleAtomicIntrinsics/README.md index f584d3de..e3ce6f66 100644 --- a/Samples/0_Introduction/simpleAtomicIntrinsics/README.md +++ b/Samples/0_Introduction/simpleAtomicIntrinsics/README.md @@ -29,42 +29,5 @@ cudaStreamCreateWithFlags, cudaFree, cudaMallocHost, cudaFreeHost, cudaStreamSyn Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l, aarch64. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
`$ make TARGET_ARCH=aarch64`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/0_Introduction/simpleAtomicIntrinsics_nvrtc/README.md b/Samples/0_Introduction/simpleAtomicIntrinsics_nvrtc/README.md index 753c5c9d..edc3dc66 100644 --- a/Samples/0_Introduction/simpleAtomicIntrinsics_nvrtc/README.md +++ b/Samples/0_Introduction/simpleAtomicIntrinsics_nvrtc/README.md @@ -36,42 +36,5 @@ cudaBlockSize, cudaGridSize Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, aarch64. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=aarch64`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/0_Introduction/simpleAttributes/README.md b/Samples/0_Introduction/simpleAttributes/README.md index cba3752e..50f76eb9 100644 --- a/Samples/0_Introduction/simpleAttributes/README.md +++ b/Samples/0_Introduction/simpleAttributes/README.md @@ -29,42 +29,5 @@ cudaFree, cudaMallocHost, cudaFreeHost, cudaStreamSynchronize, cudaStreamSetAttr Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/0_Introduction/simpleCUDA2GL/README.md b/Samples/0_Introduction/simpleCUDA2GL/README.md index 6c17427f..bd22edcf 100644 --- a/Samples/0_Introduction/simpleCUDA2GL/README.md +++ b/Samples/0_Introduction/simpleCUDA2GL/README.md @@ -33,42 +33,5 @@ cudaHostAlloc, cudaGraphicsUnmapResources, cudaMalloc, cudaFree, cudaGraphicsRes Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/0_Introduction/simpleCallback/README.md b/Samples/0_Introduction/simpleCallback/README.md index 2dd97e18..8b9189b3 100644 --- a/Samples/0_Introduction/simpleCallback/README.md +++ b/Samples/0_Introduction/simpleCallback/README.md @@ -29,42 +29,5 @@ cudaHostAlloc, cudaStreamDestroy, cudaFree, cudaSetDevice, cudaGetDeviceCount, c Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/0_Introduction/simpleCooperativeGroups/README.md b/Samples/0_Introduction/simpleCooperativeGroups/README.md index ebb1a9a5..d29a5782 100644 --- a/Samples/0_Introduction/simpleCooperativeGroups/README.md +++ b/Samples/0_Introduction/simpleCooperativeGroups/README.md @@ -29,42 +29,5 @@ cudaDeviceSynchronize, cudaGetErrorString Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/0_Introduction/simpleCubemapTexture/README.md b/Samples/0_Introduction/simpleCubemapTexture/README.md index 87e4a5cb..ba80ec08 100644 --- a/Samples/0_Introduction/simpleCubemapTexture/README.md +++ b/Samples/0_Introduction/simpleCubemapTexture/README.md @@ -29,42 +29,5 @@ cudaMemcpy, cudaCreateChannelDesc, cudaFreeArray, cudaFree, cudaPitchedPtr, cuda Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/0_Introduction/simpleDrvRuntime/README.md b/Samples/0_Introduction/simpleDrvRuntime/README.md index 52a2ed1e..47c70421 100644 --- a/Samples/0_Introduction/simpleDrvRuntime/README.md +++ b/Samples/0_Introduction/simpleDrvRuntime/README.md @@ -32,42 +32,5 @@ cudaStreamCreateWithFlags, cudaFree, cudaMallocHost, cudaFreeHost, cudaStreamSyn Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/0_Introduction/simpleHyperQ/README.md b/Samples/0_Introduction/simpleHyperQ/README.md index fcbe2e98..71f7da2b 100644 --- a/Samples/0_Introduction/simpleHyperQ/README.md +++ b/Samples/0_Introduction/simpleHyperQ/README.md @@ -29,43 +29,6 @@ cudaMemcpy, cudaStreamDestroy, cudaMalloc, cudaFree, cudaMallocHost, cudaEventSy Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) [whitepaper](./doc/HyperQ.pdf) diff --git a/Samples/0_Introduction/simpleIPC/README.md b/Samples/0_Introduction/simpleIPC/README.md index bd80ca7e..8c3640f4 100644 --- a/Samples/0_Introduction/simpleIPC/README.md +++ b/Samples/0_Introduction/simpleIPC/README.md @@ -33,42 +33,5 @@ cudaSetDevice, cudaIpcCloseMemHandle, cudaEventDestroy, cudaGetDeviceCount, cuda Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/0_Introduction/simpleLayeredTexture/README.md b/Samples/0_Introduction/simpleLayeredTexture/README.md index f13fae26..59925e48 100644 --- a/Samples/0_Introduction/simpleLayeredTexture/README.md +++ b/Samples/0_Introduction/simpleLayeredTexture/README.md @@ -29,42 +29,5 @@ cudaMemcpy, cudaCreateChannelDesc, cudaFreeArray, cudaFree, cudaPitchedPtr, cuda Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/0_Introduction/simpleMPI/README.md b/Samples/0_Introduction/simpleMPI/README.md index 1a8a152a..d47d39b0 100644 --- a/Samples/0_Introduction/simpleMPI/README.md +++ b/Samples/0_Introduction/simpleMPI/README.md @@ -33,42 +33,5 @@ cudaMalloc, cudaGetLastError, cudaMemcpy, cudaFree Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, aarch64. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=aarch64`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/0_Introduction/simpleMultiCopy/README.md b/Samples/0_Introduction/simpleMultiCopy/README.md index d531a50c..2a0ad058 100644 --- a/Samples/0_Introduction/simpleMultiCopy/README.md +++ b/Samples/0_Introduction/simpleMultiCopy/README.md @@ -29,42 +29,5 @@ cudaHostAlloc, cudaStreamDestroy, cudaMalloc, cudaMemcpyAsync, cudaFree, cudaSet Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/0_Introduction/simpleMultiGPU/README.md b/Samples/0_Introduction/simpleMultiGPU/README.md index 3fc5e0b3..2eec927e 100644 --- a/Samples/0_Introduction/simpleMultiGPU/README.md +++ b/Samples/0_Introduction/simpleMultiGPU/README.md @@ -29,42 +29,5 @@ cudaStreamDestroy, cudaFree, cudaMallocHost, cudaSetDevice, cudaFreeHost, cudaSt Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/0_Introduction/simpleOccupancy/README.md b/Samples/0_Introduction/simpleOccupancy/README.md index 2f327a6e..5ce91a94 100644 --- a/Samples/0_Introduction/simpleOccupancy/README.md +++ b/Samples/0_Introduction/simpleOccupancy/README.md @@ -29,42 +29,5 @@ cudaMemcpy, cudaFree, cudaDeviceSynchronize, cudaEventRecord, cudaGetDevice, cud Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/0_Introduction/simpleP2P/README.md b/Samples/0_Introduction/simpleP2P/README.md index 15179b35..8d9b3770 100644 --- a/Samples/0_Introduction/simpleP2P/README.md +++ b/Samples/0_Introduction/simpleP2P/README.md @@ -33,42 +33,5 @@ cudaMemcpy, cudaMalloc, cudaFree, cudaMallocHost, cudaEventCreateWithFlags, cuda Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/0_Introduction/simplePitchLinearTexture/README.md b/Samples/0_Introduction/simplePitchLinearTexture/README.md index 921f3891..2f2198fc 100644 --- a/Samples/0_Introduction/simplePitchLinearTexture/README.md +++ b/Samples/0_Introduction/simplePitchLinearTexture/README.md @@ -29,42 +29,5 @@ cudaMallocArray, cudaFreeArray, cudaFree, cudaMallocPitch, cudaDestroyTextureObj Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/0_Introduction/simplePrintf/README.md b/Samples/0_Introduction/simplePrintf/README.md index 69dba7a2..c6925f8d 100644 --- a/Samples/0_Introduction/simplePrintf/README.md +++ b/Samples/0_Introduction/simplePrintf/README.md @@ -29,42 +29,5 @@ cudaGetDeviceProperties, cudaDeviceSynchronize, cudaGetDevice Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/0_Introduction/simpleStreams/README.md b/Samples/0_Introduction/simpleStreams/README.md index 439f733a..3fdb9e46 100644 --- a/Samples/0_Introduction/simpleStreams/README.md +++ b/Samples/0_Introduction/simpleStreams/README.md @@ -29,42 +29,5 @@ cudaMemcpy, cudaSetDeviceFlags, cudaSetDevice, cudaEventDestroy, cudaStreamCreat Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/0_Introduction/simpleSurfaceWrite/README.md b/Samples/0_Introduction/simpleSurfaceWrite/README.md index a61c1d2e..3d81c682 100644 --- a/Samples/0_Introduction/simpleSurfaceWrite/README.md +++ b/Samples/0_Introduction/simpleSurfaceWrite/README.md @@ -29,42 +29,5 @@ cudaMemcpy, cudaCreateChannelDesc, cudaMallocArray, cudaFreeArray, cudaFree, cud Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/0_Introduction/simpleTemplates/README.md b/Samples/0_Introduction/simpleTemplates/README.md index 5ab565a0..1e2a6e85 100644 --- a/Samples/0_Introduction/simpleTemplates/README.md +++ b/Samples/0_Introduction/simpleTemplates/README.md @@ -29,42 +29,5 @@ cudaMalloc, cudaMemcpy, cudaGetDeviceProperties, cudaFree Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/0_Introduction/simpleTexture/README.md b/Samples/0_Introduction/simpleTexture/README.md index fdca8911..b3e2535b 100644 --- a/Samples/0_Introduction/simpleTexture/README.md +++ b/Samples/0_Introduction/simpleTexture/README.md @@ -29,42 +29,5 @@ cudaMemcpy, cudaCreateChannelDesc, cudaMallocArray, cudaFreeArray, cudaFree, cud Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/0_Introduction/simpleTexture3D/README.md b/Samples/0_Introduction/simpleTexture3D/README.md index 48f9961c..235f1276 100644 --- a/Samples/0_Introduction/simpleTexture3D/README.md +++ b/Samples/0_Introduction/simpleTexture3D/README.md @@ -33,42 +33,5 @@ cudaGraphicsUnmapResources, cudaMemcpy, cudaFreeArray, cudaFree, cudaPitchedPtr, Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/0_Introduction/simpleTextureDrv/README.md b/Samples/0_Introduction/simpleTextureDrv/README.md index 461d408c..604cee8e 100644 --- a/Samples/0_Introduction/simpleTextureDrv/README.md +++ b/Samples/0_Introduction/simpleTextureDrv/README.md @@ -29,42 +29,5 @@ cuMemcpyDtoH, cuLaunchKernel, cuModuleLoadData, cuDeviceGetName, cuDeviceGetAttr Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/0_Introduction/simpleVoteIntrinsics/README.md b/Samples/0_Introduction/simpleVoteIntrinsics/README.md index 7558f259..a44a2abf 100644 --- a/Samples/0_Introduction/simpleVoteIntrinsics/README.md +++ b/Samples/0_Introduction/simpleVoteIntrinsics/README.md @@ -29,42 +29,5 @@ cudaMemcpy, cudaFree, cudaDeviceSynchronize, cudaMalloc, cudaGetDeviceProperties Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l, aarch64. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
`$ make TARGET_ARCH=aarch64`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/0_Introduction/simpleZeroCopy/README.md b/Samples/0_Introduction/simpleZeroCopy/README.md index faf5b407..78ff257d 100644 --- a/Samples/0_Introduction/simpleZeroCopy/README.md +++ b/Samples/0_Introduction/simpleZeroCopy/README.md @@ -29,42 +29,5 @@ cudaHostAlloc, cudaSetDeviceFlags, cudaHostRegister, cudaSetDevice, cudaGetDevic Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/0_Introduction/systemWideAtomics/README.md b/Samples/0_Introduction/systemWideAtomics/README.md index 35d4886e..4a539cd6 100644 --- a/Samples/0_Introduction/systemWideAtomics/README.md +++ b/Samples/0_Introduction/systemWideAtomics/README.md @@ -33,32 +33,5 @@ cudaDeviceSynchronize, cudaMallocManaged, cudaGetDeviceProperties, cudaFree Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/0_Introduction/template/README.md b/Samples/0_Introduction/template/README.md index af7fa757..670f172b 100644 --- a/Samples/0_Introduction/template/README.md +++ b/Samples/0_Introduction/template/README.md @@ -29,42 +29,5 @@ cudaMalloc, cudaMemcpy, cudaFree Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/0_Introduction/vectorAdd/README.md b/Samples/0_Introduction/vectorAdd/README.md index bcafb84a..8db6365b 100644 --- a/Samples/0_Introduction/vectorAdd/README.md +++ b/Samples/0_Introduction/vectorAdd/README.md @@ -29,42 +29,5 @@ cudaMemcpy, cudaGetErrorString, cudaFree, cudaGetLastError, cudaMalloc Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/0_Introduction/vectorAddDrv/README.md b/Samples/0_Introduction/vectorAddDrv/README.md index ededa2ac..a23138a1 100644 --- a/Samples/0_Introduction/vectorAddDrv/README.md +++ b/Samples/0_Introduction/vectorAddDrv/README.md @@ -29,42 +29,5 @@ cuMemcpyDtoH, cuLaunchKernel, cuMemcpyHtoD, cuModuleLoadData, cuCtxSynchronize, Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/0_Introduction/vectorAddMMAP/README.md b/Samples/0_Introduction/vectorAddMMAP/README.md index 701e08e3..512ae5f2 100644 --- a/Samples/0_Introduction/vectorAddMMAP/README.md +++ b/Samples/0_Introduction/vectorAddMMAP/README.md @@ -29,42 +29,5 @@ cuMemcpyDtoH, cuDeviceCanAccessPeer, cuModuleGetFunction, cuMemSetAccess, cuMemR Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/0_Introduction/vectorAdd_nvrtc/README.md b/Samples/0_Introduction/vectorAdd_nvrtc/README.md index 863f4c5a..0fa763dd 100644 --- a/Samples/0_Introduction/vectorAdd_nvrtc/README.md +++ b/Samples/0_Introduction/vectorAdd_nvrtc/README.md @@ -36,42 +36,5 @@ cudaBlockSize, cudaGridSize Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, aarch64. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=aarch64`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/1_Utilities/bandwidthTest/README.md b/Samples/1_Utilities/bandwidthTest/README.md index 392e4897..88fe8b0a 100644 --- a/Samples/1_Utilities/bandwidthTest/README.md +++ b/Samples/1_Utilities/bandwidthTest/README.md @@ -29,42 +29,5 @@ cudaHostAlloc, cudaMemcpy, cudaMalloc, cudaMemcpyAsync, cudaFree, cudaGetErrorSt Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/1_Utilities/deviceQuery/README.md b/Samples/1_Utilities/deviceQuery/README.md index cce64760..812483aa 100644 --- a/Samples/1_Utilities/deviceQuery/README.md +++ b/Samples/1_Utilities/deviceQuery/README.md @@ -32,42 +32,5 @@ cudaRuntimeGetVersion, cudaGetErrorString, cudaDeviceCanAccessPeer, cudaSetDevic Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l, aarch64. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
`$ make TARGET_ARCH=aarch64`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/1_Utilities/deviceQueryDrv/README.md b/Samples/1_Utilities/deviceQueryDrv/README.md index e71a2c74..a8d09163 100644 --- a/Samples/1_Utilities/deviceQueryDrv/README.md +++ b/Samples/1_Utilities/deviceQueryDrv/README.md @@ -32,42 +32,5 @@ cudaSetDevice Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/1_Utilities/topologyQuery/README.md b/Samples/1_Utilities/topologyQuery/README.md index 6f0c48e8..0ee74818 100644 --- a/Samples/1_Utilities/topologyQuery/README.md +++ b/Samples/1_Utilities/topologyQuery/README.md @@ -29,42 +29,5 @@ cudaGetDeviceCount, cudaDeviceGetAttribute Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/2_Concepts_and_Techniques/EGLStream_CUDA_CrossGPU/README.md b/Samples/2_Concepts_and_Techniques/EGLStream_CUDA_CrossGPU/README.md index d753a314..93c703e8 100644 --- a/Samples/2_Concepts_and_Techniques/EGLStream_CUDA_CrossGPU/README.md +++ b/Samples/2_Concepts_and_Techniques/EGLStream_CUDA_CrossGPU/README.md @@ -36,32 +36,5 @@ cudaMemcpy, cudaMalloc, cudaProducerPresentFrame, cudaFree, cudaGetErrorString, Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/2_Concepts_and_Techniques/EGLStream_CUDA_Interop/README.md b/Samples/2_Concepts_and_Techniques/EGLStream_CUDA_Interop/README.md index 3dda94de..2cc244e5 100644 --- a/Samples/2_Concepts_and_Techniques/EGLStream_CUDA_Interop/README.md +++ b/Samples/2_Concepts_and_Techniques/EGLStream_CUDA_Interop/README.md @@ -36,32 +36,5 @@ cudaProducerReadYUVFrame, cudaProducerTest, cudaProducerDeinit, cudaDeviceCreate Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, aarch64. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=aarch64`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/2_Concepts_and_Techniques/EGLSync_CUDAEvent_Interop/README.md b/Samples/2_Concepts_and_Techniques/EGLSync_CUDAEvent_Interop/README.md index 9668be56..d659b8f8 100644 --- a/Samples/2_Concepts_and_Techniques/EGLSync_CUDAEvent_Interop/README.md +++ b/Samples/2_Concepts_and_Techniques/EGLSync_CUDAEvent_Interop/README.md @@ -36,32 +36,5 @@ cudaMemcpy, cudaGetErrorString, cudaFree, cudaDeviceSynchronize, cudaGetValueMis Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/2_Concepts_and_Techniques/FunctionPointers/README.md b/Samples/2_Concepts_and_Techniques/FunctionPointers/README.md index 883676df..76ecfefe 100644 --- a/Samples/2_Concepts_and_Techniques/FunctionPointers/README.md +++ b/Samples/2_Concepts_and_Techniques/FunctionPointers/README.md @@ -33,42 +33,5 @@ cudaGraphicsUnmapResources, cudaMemcpy, cudaMallocArray, cudaFreeArray, cudaFree Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/2_Concepts_and_Techniques/MC_EstimatePiInlineP/README.md b/Samples/2_Concepts_and_Techniques/MC_EstimatePiInlineP/README.md index e534a2af..8d4831d8 100644 --- a/Samples/2_Concepts_and_Techniques/MC_EstimatePiInlineP/README.md +++ b/Samples/2_Concepts_and_Techniques/MC_EstimatePiInlineP/README.md @@ -33,42 +33,5 @@ cudaMemcpy, cudaGetErrorString, cudaFree, cudaSetDevice, cudaGetDeviceCount, cud Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/2_Concepts_and_Techniques/MC_EstimatePiInlineQ/README.md b/Samples/2_Concepts_and_Techniques/MC_EstimatePiInlineQ/README.md index 7de22e0c..f5152d22 100644 --- a/Samples/2_Concepts_and_Techniques/MC_EstimatePiInlineQ/README.md +++ b/Samples/2_Concepts_and_Techniques/MC_EstimatePiInlineQ/README.md @@ -33,42 +33,5 @@ cudaMemcpy, cudaGetErrorString, cudaFree, cudaSetDevice, cudaGetDeviceCount, cud Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/2_Concepts_and_Techniques/MC_EstimatePiP/README.md b/Samples/2_Concepts_and_Techniques/MC_EstimatePiP/README.md index 43c16ba4..026e6da4 100644 --- a/Samples/2_Concepts_and_Techniques/MC_EstimatePiP/README.md +++ b/Samples/2_Concepts_and_Techniques/MC_EstimatePiP/README.md @@ -33,42 +33,5 @@ cudaMemcpy, cudaGetErrorString, cudaFree, cudaSetDevice, cudaGetDeviceCount, cud Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/2_Concepts_and_Techniques/MC_EstimatePiQ/README.md b/Samples/2_Concepts_and_Techniques/MC_EstimatePiQ/README.md index a26ae785..913f4978 100644 --- a/Samples/2_Concepts_and_Techniques/MC_EstimatePiQ/README.md +++ b/Samples/2_Concepts_and_Techniques/MC_EstimatePiQ/README.md @@ -33,42 +33,5 @@ cudaMemcpy, cudaGetErrorString, cudaFree, cudaSetDevice, cudaGetDeviceCount, cud Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/2_Concepts_and_Techniques/MC_SingleAsianOptionP/README.md b/Samples/2_Concepts_and_Techniques/MC_SingleAsianOptionP/README.md index 4984ce2a..2e93a566 100644 --- a/Samples/2_Concepts_and_Techniques/MC_SingleAsianOptionP/README.md +++ b/Samples/2_Concepts_and_Techniques/MC_SingleAsianOptionP/README.md @@ -33,42 +33,5 @@ cudaMemcpy, cudaGetErrorString, cudaFree, cudaSetDevice, cudaGetDeviceCount, cud Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/2_Concepts_and_Techniques/boxFilter/README.md b/Samples/2_Concepts_and_Techniques/boxFilter/README.md index 9ed5b0e4..98ee4901 100644 --- a/Samples/2_Concepts_and_Techniques/boxFilter/README.md +++ b/Samples/2_Concepts_and_Techniques/boxFilter/README.md @@ -33,42 +33,5 @@ cudaGraphicsUnmapResources, cudaCreateChannelDesc, cudaMallocArray, cudaFreeArra Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/2_Concepts_and_Techniques/convolutionSeparable/README.md b/Samples/2_Concepts_and_Techniques/convolutionSeparable/README.md index c79eb31a..02a3fe70 100644 --- a/Samples/2_Concepts_and_Techniques/convolutionSeparable/README.md +++ b/Samples/2_Concepts_and_Techniques/convolutionSeparable/README.md @@ -29,43 +29,6 @@ cudaMemcpy, cudaFree, cudaDeviceSynchronize, cudaMemcpyToSymbol, cudaMalloc Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) [whitepaper](./doc/convolutionSeparable.pdf) diff --git a/Samples/2_Concepts_and_Techniques/convolutionTexture/README.md b/Samples/2_Concepts_and_Techniques/convolutionTexture/README.md index 5e5ffb26..f96b1eab 100644 --- a/Samples/2_Concepts_and_Techniques/convolutionTexture/README.md +++ b/Samples/2_Concepts_and_Techniques/convolutionTexture/README.md @@ -29,42 +29,5 @@ cudaMemcpy, cudaMallocArray, cudaFreeArray, cudaFree, cudaMemcpyToArray, cudaDev Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/2_Concepts_and_Techniques/dct8x8/README.md b/Samples/2_Concepts_and_Techniques/dct8x8/README.md index 22ee2698..683a9bd1 100644 --- a/Samples/2_Concepts_and_Techniques/dct8x8/README.md +++ b/Samples/2_Concepts_and_Techniques/dct8x8/README.md @@ -29,43 +29,6 @@ cudaMallocArray, cudaFreeArray, cudaFree, cudaMallocPitch, cudaDestroyTextureObj Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) [whitepaper](./doc/dct8x8.pdf) diff --git a/Samples/2_Concepts_and_Techniques/eigenvalues/README.md b/Samples/2_Concepts_and_Techniques/eigenvalues/README.md index 235fac21..b44ca506 100644 --- a/Samples/2_Concepts_and_Techniques/eigenvalues/README.md +++ b/Samples/2_Concepts_and_Techniques/eigenvalues/README.md @@ -29,43 +29,6 @@ cudaMalloc, cudaDeviceSynchronize, cudaMemcpy, cudaFree Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) [whitepaper](./doc/eigenvalues.pdf) diff --git a/Samples/2_Concepts_and_Techniques/histogram/README.md b/Samples/2_Concepts_and_Techniques/histogram/README.md index a6088c65..27409eb3 100644 --- a/Samples/2_Concepts_and_Techniques/histogram/README.md +++ b/Samples/2_Concepts_and_Techniques/histogram/README.md @@ -29,43 +29,6 @@ cudaMemcpy, cudaFree, cudaDeviceSynchronize, cudaMalloc, cudaGetDeviceProperties Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) [whitepaper](./doc/histogram.pdf) diff --git a/Samples/2_Concepts_and_Techniques/imageDenoising/README.md b/Samples/2_Concepts_and_Techniques/imageDenoising/README.md index 9242b34f..05f25f5f 100644 --- a/Samples/2_Concepts_and_Techniques/imageDenoising/README.md +++ b/Samples/2_Concepts_and_Techniques/imageDenoising/README.md @@ -33,43 +33,6 @@ cudaGraphicsUnmapResources, cudaMemcpy, cudaMallocArray, cudaFreeArray, cudaFree Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) [whitepaper](./doc/imageDenoising.pdf) diff --git a/Samples/2_Concepts_and_Techniques/inlinePTX/README.md b/Samples/2_Concepts_and_Techniques/inlinePTX/README.md index 2760047c..5dbe521a 100644 --- a/Samples/2_Concepts_and_Techniques/inlinePTX/README.md +++ b/Samples/2_Concepts_and_Techniques/inlinePTX/README.md @@ -29,42 +29,5 @@ cudaMemcpy, cudaFree, cudaMallocHost, cudaGetLastError, cudaGridSize, cudaBlockS Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/2_Concepts_and_Techniques/inlinePTX_nvrtc/README.md b/Samples/2_Concepts_and_Techniques/inlinePTX_nvrtc/README.md index 6e535bd5..88b3552f 100644 --- a/Samples/2_Concepts_and_Techniques/inlinePTX_nvrtc/README.md +++ b/Samples/2_Concepts_and_Techniques/inlinePTX_nvrtc/README.md @@ -36,42 +36,5 @@ cudaBlockSize, cudaGridSize Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, aarch64. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=aarch64`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/2_Concepts_and_Techniques/interval/README.md b/Samples/2_Concepts_and_Techniques/interval/README.md index 9eafdaba..1b71b0d8 100644 --- a/Samples/2_Concepts_and_Techniques/interval/README.md +++ b/Samples/2_Concepts_and_Techniques/interval/README.md @@ -29,42 +29,5 @@ cudaMemcpy, cudaFuncSetCacheConfig, cudaMalloc, cudaFree, cudaGetLastError, cuda Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/2_Concepts_and_Techniques/particles/README.md b/Samples/2_Concepts_and_Techniques/particles/README.md index f86cac02..1cd0b89b 100644 --- a/Samples/2_Concepts_and_Techniques/particles/README.md +++ b/Samples/2_Concepts_and_Techniques/particles/README.md @@ -33,43 +33,6 @@ cudaGraphicsUnmapResources, cudaMemcpy, cudaFree, cudaGraphicsResourceGetMappedP Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) [whitepaper](./doc/particles.pdf) diff --git a/Samples/2_Concepts_and_Techniques/radixSortThrust/README.md b/Samples/2_Concepts_and_Techniques/radixSortThrust/README.md index 2ffa76f2..ac83400d 100644 --- a/Samples/2_Concepts_and_Techniques/radixSortThrust/README.md +++ b/Samples/2_Concepts_and_Techniques/radixSortThrust/README.md @@ -29,43 +29,6 @@ cudaEventSynchronize, cudaEventRecord, cudaGetDevice, cudaEventDestroy, cudaEven Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) [whitepaper](./doc/readme.txt) diff --git a/Samples/2_Concepts_and_Techniques/reduction/README.md b/Samples/2_Concepts_and_Techniques/reduction/README.md index ac916704..9ecb49ba 100644 --- a/Samples/2_Concepts_and_Techniques/reduction/README.md +++ b/Samples/2_Concepts_and_Techniques/reduction/README.md @@ -33,42 +33,5 @@ cudaMemcpy, cudaFree, cudaSetDevice, cudaDeviceSynchronize, cudaGetDevice, cudaM Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/2_Concepts_and_Techniques/reductionMultiBlockCG/README.md b/Samples/2_Concepts_and_Techniques/reductionMultiBlockCG/README.md index 706766aa..79285575 100644 --- a/Samples/2_Concepts_and_Techniques/reductionMultiBlockCG/README.md +++ b/Samples/2_Concepts_and_Techniques/reductionMultiBlockCG/README.md @@ -33,42 +33,5 @@ cudaMemcpy, cudaFree, cudaSetDevice, cudaDeviceSynchronize, cudaLaunchCooperativ Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, aarch64. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=aarch64`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/2_Concepts_and_Techniques/scalarProd/README.md b/Samples/2_Concepts_and_Techniques/scalarProd/README.md index 139d1ca0..99bc11bb 100644 --- a/Samples/2_Concepts_and_Techniques/scalarProd/README.md +++ b/Samples/2_Concepts_and_Techniques/scalarProd/README.md @@ -29,42 +29,5 @@ cudaMalloc, cudaDeviceSynchronize, cudaMemcpy, cudaFree Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/2_Concepts_and_Techniques/scan/README.md b/Samples/2_Concepts_and_Techniques/scan/README.md index 5db006d3..45cd8578 100644 --- a/Samples/2_Concepts_and_Techniques/scan/README.md +++ b/Samples/2_Concepts_and_Techniques/scan/README.md @@ -29,42 +29,5 @@ cudaMalloc, cudaDeviceSynchronize, cudaMemcpy, cudaFree Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/2_Concepts_and_Techniques/segmentationTreeThrust/README.md b/Samples/2_Concepts_and_Techniques/segmentationTreeThrust/README.md index 1959b9a8..60a7f985 100644 --- a/Samples/2_Concepts_and_Techniques/segmentationTreeThrust/README.md +++ b/Samples/2_Concepts_and_Techniques/segmentationTreeThrust/README.md @@ -29,42 +29,5 @@ cudaMemcpy, cudaMemGetInfo, cudaEventSynchronize, cudaEventRecord, cudaMemset, c Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/2_Concepts_and_Techniques/shfl_scan/README.md b/Samples/2_Concepts_and_Techniques/shfl_scan/README.md index 8d04c940..33f5c6d7 100644 --- a/Samples/2_Concepts_and_Techniques/shfl_scan/README.md +++ b/Samples/2_Concepts_and_Techniques/shfl_scan/README.md @@ -31,42 +31,5 @@ cudaMemcpy, cudaFree, cudaMallocHost, cudaEventSynchronize, cudaEventRecord, cud Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l, aarch64. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
`$ make TARGET_ARCH=aarch64`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/2_Concepts_and_Techniques/sortingNetworks/README.md b/Samples/2_Concepts_and_Techniques/sortingNetworks/README.md index 1b100131..3605f5c4 100644 --- a/Samples/2_Concepts_and_Techniques/sortingNetworks/README.md +++ b/Samples/2_Concepts_and_Techniques/sortingNetworks/README.md @@ -29,42 +29,5 @@ cudaMalloc, cudaDeviceSynchronize, cudaMemcpy, cudaFree Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/2_Concepts_and_Techniques/streamOrderedAllocation/README.md b/Samples/2_Concepts_and_Techniques/streamOrderedAllocation/README.md index a93e1121..d108965f 100644 --- a/Samples/2_Concepts_and_Techniques/streamOrderedAllocation/README.md +++ b/Samples/2_Concepts_and_Techniques/streamOrderedAllocation/README.md @@ -29,42 +29,5 @@ cudaDeviceGetDefaultMemPool, cudaFreeAsync, cudaStreamCreateWithFlags, cudaStrea Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/2_Concepts_and_Techniques/streamOrderedAllocationIPC/README.md b/Samples/2_Concepts_and_Techniques/streamOrderedAllocationIPC/README.md index e6633880..8b4ee825 100644 --- a/Samples/2_Concepts_and_Techniques/streamOrderedAllocationIPC/README.md +++ b/Samples/2_Concepts_and_Techniques/streamOrderedAllocationIPC/README.md @@ -32,32 +32,5 @@ cudaDeviceGetAttribute, cudaMemPoolImportFromShareableHandle, cudaSetDevice, cud Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. -## Build and Run - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/2_Concepts_and_Techniques/streamOrderedAllocationP2P/README.md b/Samples/2_Concepts_and_Techniques/streamOrderedAllocationP2P/README.md index 775bc09b..9e1f80ac 100644 --- a/Samples/2_Concepts_and_Techniques/streamOrderedAllocationP2P/README.md +++ b/Samples/2_Concepts_and_Techniques/streamOrderedAllocationP2P/README.md @@ -29,42 +29,5 @@ cudaDeviceGetDefaultMemPool, cudaFreeAsync, cudaStreamCreateWithFlags, cudaMemPo Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/2_Concepts_and_Techniques/threadFenceReduction/README.md b/Samples/2_Concepts_and_Techniques/threadFenceReduction/README.md index 15c1f3e7..02acdb96 100644 --- a/Samples/2_Concepts_and_Techniques/threadFenceReduction/README.md +++ b/Samples/2_Concepts_and_Techniques/threadFenceReduction/README.md @@ -29,42 +29,5 @@ cudaMemcpy, cudaFree, cudaDeviceSynchronize, cudaMalloc, cudaGetDeviceProperties Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/2_Concepts_and_Techniques/threadMigration/README.md b/Samples/2_Concepts_and_Techniques/threadMigration/README.md index 89357423..9519bff3 100644 --- a/Samples/2_Concepts_and_Techniques/threadMigration/README.md +++ b/Samples/2_Concepts_and_Techniques/threadMigration/README.md @@ -29,42 +29,5 @@ cuMemcpyDtoH, cuLaunchKernel, cuModuleLoadData, cuDeviceGetName, cuDeviceGet, cu Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/3_CUDA_Features/StreamPriorities/README.md b/Samples/3_CUDA_Features/StreamPriorities/README.md index 1657bddf..9b9d056e 100644 --- a/Samples/3_CUDA_Features/StreamPriorities/README.md +++ b/Samples/3_CUDA_Features/StreamPriorities/README.md @@ -31,32 +31,5 @@ cudaMemcpy, cudaStreamCreateWithPriority, cudaDeviceGetStreamPriorityRange, cuda Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/3_CUDA_Features/bf16TensorCoreGemm/README.md b/Samples/3_CUDA_Features/bf16TensorCoreGemm/README.md index 9f72eebe..040faa92 100644 --- a/Samples/3_CUDA_Features/bf16TensorCoreGemm/README.md +++ b/Samples/3_CUDA_Features/bf16TensorCoreGemm/README.md @@ -33,42 +33,5 @@ cudaMemcpy, cudaFree, cudaGetErrorString, cudaGetLastError, cudaEventSynchronize Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, aarch64. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=aarch64`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/3_CUDA_Features/binaryPartitionCG/README.md b/Samples/3_CUDA_Features/binaryPartitionCG/README.md index 1b26aec5..cc4c1c6d 100644 --- a/Samples/3_CUDA_Features/binaryPartitionCG/README.md +++ b/Samples/3_CUDA_Features/binaryPartitionCG/README.md @@ -29,42 +29,5 @@ cudaStreamCreateWithFlags, cudaFree, cudaMallocHost, cudaFreeHost, cudaStreamSyn Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/3_CUDA_Features/bindlessTexture/README.md b/Samples/3_CUDA_Features/bindlessTexture/README.md index 18414d86..d9820b0b 100644 --- a/Samples/3_CUDA_Features/bindlessTexture/README.md +++ b/Samples/3_CUDA_Features/bindlessTexture/README.md @@ -31,42 +31,5 @@ cudaMemcpy, cudaGetMipmappedArrayLevel, cudaGraphicsMapResources, cudaDestroySur Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/3_CUDA_Features/cdpAdvancedQuicksort/README.md b/Samples/3_CUDA_Features/cdpAdvancedQuicksort/README.md index 843cff41..9e976076 100644 --- a/Samples/3_CUDA_Features/cdpAdvancedQuicksort/README.md +++ b/Samples/3_CUDA_Features/cdpAdvancedQuicksort/README.md @@ -31,42 +31,5 @@ cudaStreamCreateWithFlags, cudaMemcpy, cudaMemcpyAsync, cudaFree, cudaGetErrorSt Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/3_CUDA_Features/cdpBezierTessellation/README.md b/Samples/3_CUDA_Features/cdpBezierTessellation/README.md index 2b3cc6cb..9ce5d9b7 100644 --- a/Samples/3_CUDA_Features/cdpBezierTessellation/README.md +++ b/Samples/3_CUDA_Features/cdpBezierTessellation/README.md @@ -31,42 +31,5 @@ cudaMemcpy, cudaFree, cudaGetDeviceCount, cudaMalloc, cudaGetDeviceProperties Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/3_CUDA_Features/cdpQuadtree/README.md b/Samples/3_CUDA_Features/cdpQuadtree/README.md index 13fc0f3f..edf9b6cd 100644 --- a/Samples/3_CUDA_Features/cdpQuadtree/README.md +++ b/Samples/3_CUDA_Features/cdpQuadtree/README.md @@ -31,42 +31,5 @@ cudaMemcpy, cudaFree, cudaGetLastError, cudaDeviceSetLimit, cudaMalloc, cudaGetD Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/3_CUDA_Features/cdpSimplePrint/README.md b/Samples/3_CUDA_Features/cdpSimplePrint/README.md index e975d689..6b6d9fbb 100644 --- a/Samples/3_CUDA_Features/cdpSimplePrint/README.md +++ b/Samples/3_CUDA_Features/cdpSimplePrint/README.md @@ -31,42 +31,5 @@ cudaDeviceSynchronize, cudaGetLastError, cudaGetDeviceProperties, cudaDeviceSetL Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/3_CUDA_Features/cdpSimpleQuicksort/README.md b/Samples/3_CUDA_Features/cdpSimpleQuicksort/README.md index a4aa7761..76485b93 100644 --- a/Samples/3_CUDA_Features/cdpSimpleQuicksort/README.md +++ b/Samples/3_CUDA_Features/cdpSimpleQuicksort/README.md @@ -31,42 +31,5 @@ cudaStreamCreateWithFlags, cudaMemcpy, cudaStreamDestroy, cudaFree, cudaDeviceSy Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/3_CUDA_Features/cudaCompressibleMemory/README.md b/Samples/3_CUDA_Features/cudaCompressibleMemory/README.md index a223d3c5..e4ec9945 100644 --- a/Samples/3_CUDA_Features/cudaCompressibleMemory/README.md +++ b/Samples/3_CUDA_Features/cudaCompressibleMemory/README.md @@ -32,42 +32,5 @@ cudaMemcpy, cudaEventSynchronize, cudaEventRecord, cudaEventElapsedTime, cudaOcc Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, aarch64. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=aarch64`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/3_CUDA_Features/cudaTensorCoreGemm/README.md b/Samples/3_CUDA_Features/cudaTensorCoreGemm/README.md index e89c6d0a..fcb5a6c9 100644 --- a/Samples/3_CUDA_Features/cudaTensorCoreGemm/README.md +++ b/Samples/3_CUDA_Features/cudaTensorCoreGemm/README.md @@ -33,42 +33,5 @@ cudaMemcpy, cudaFree, cudaGetErrorString, cudaGetLastError, cudaEventSynchronize Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, aarch64. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=aarch64`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/3_CUDA_Features/dmmaTensorCoreGemm/README.md b/Samples/3_CUDA_Features/dmmaTensorCoreGemm/README.md index 2320afba..ab43c04f 100644 --- a/Samples/3_CUDA_Features/dmmaTensorCoreGemm/README.md +++ b/Samples/3_CUDA_Features/dmmaTensorCoreGemm/README.md @@ -33,42 +33,5 @@ cudaMemcpy, cudaFree, cudaGetErrorString, cudaGetLastError, cudaEventSynchronize Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, aarch64. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=aarch64`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/3_CUDA_Features/globalToShmemAsyncCopy/README.md b/Samples/3_CUDA_Features/globalToShmemAsyncCopy/README.md index 807d0e34..f2c9590b 100644 --- a/Samples/3_CUDA_Features/globalToShmemAsyncCopy/README.md +++ b/Samples/3_CUDA_Features/globalToShmemAsyncCopy/README.md @@ -33,42 +33,5 @@ cudaStreamCreateWithFlags, cudaMalloc, cudaDeviceGetAttribute, cudaFree, cudaMal Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l, aarch64. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
`$ make TARGET_ARCH=aarch64`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/3_CUDA_Features/graphConditionalNodes/README.md b/Samples/3_CUDA_Features/graphConditionalNodes/README.md index 8fd835cb..1ab9a410 100644 --- a/Samples/3_CUDA_Features/graphConditionalNodes/README.md +++ b/Samples/3_CUDA_Features/graphConditionalNodes/README.md @@ -29,42 +29,5 @@ cudaDeviceSynchronize, cudaDriverGetVersion, cudaFree, cudaGraphAddNode, cudaGra Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/3_CUDA_Features/graphMemoryFootprint/README.md b/Samples/3_CUDA_Features/graphMemoryFootprint/README.md index 7e0ffde4..bfe72e5b 100644 --- a/Samples/3_CUDA_Features/graphMemoryFootprint/README.md +++ b/Samples/3_CUDA_Features/graphMemoryFootprint/README.md @@ -29,42 +29,5 @@ cudaGraphAddMemAllocNode, cudaStreamCreateWithFlags, cudaGraphInstantiate, cudaS Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/3_CUDA_Features/graphMemoryNodes/README.md b/Samples/3_CUDA_Features/graphMemoryNodes/README.md index 1c993f6a..36236822 100644 --- a/Samples/3_CUDA_Features/graphMemoryNodes/README.md +++ b/Samples/3_CUDA_Features/graphMemoryNodes/README.md @@ -29,42 +29,5 @@ cudaMemcpy, cudaDeviceGetAttribute, cudaDriverGetVersion, cudaGraphLaunch, cudaE Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/3_CUDA_Features/immaTensorCoreGemm/README.md b/Samples/3_CUDA_Features/immaTensorCoreGemm/README.md index 88f21201..83f230ae 100644 --- a/Samples/3_CUDA_Features/immaTensorCoreGemm/README.md +++ b/Samples/3_CUDA_Features/immaTensorCoreGemm/README.md @@ -29,42 +29,5 @@ cudaMemcpy, cudaFree, cudaGetErrorString, cudaGetLastError, cudaEventSynchronize Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, aarch64. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=aarch64`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/3_CUDA_Features/jacobiCudaGraphs/README.md b/Samples/3_CUDA_Features/jacobiCudaGraphs/README.md index 0370d726..58476d2e 100644 --- a/Samples/3_CUDA_Features/jacobiCudaGraphs/README.md +++ b/Samples/3_CUDA_Features/jacobiCudaGraphs/README.md @@ -27,42 +27,5 @@ cudaExtent, cudaGraphLaunch, cudaGraphAddMemcpyNode, cudaMallocHost, cudaPitched Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/3_CUDA_Features/memMapIPCDrv/README.md b/Samples/3_CUDA_Features/memMapIPCDrv/README.md index d15b4c92..661a15a8 100644 --- a/Samples/3_CUDA_Features/memMapIPCDrv/README.md +++ b/Samples/3_CUDA_Features/memMapIPCDrv/README.md @@ -33,42 +33,5 @@ cuDeviceCanAccessPeer, cuMemImportFromShareableHandle, cuModuleLoadDataEx, cuMod Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l, aarch64. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
`$ make TARGET_ARCH=aarch64`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/3_CUDA_Features/newdelete/README.md b/Samples/3_CUDA_Features/newdelete/README.md index b28d236a..42aacd3b 100644 --- a/Samples/3_CUDA_Features/newdelete/README.md +++ b/Samples/3_CUDA_Features/newdelete/README.md @@ -29,42 +29,5 @@ cudaMemcpy, cudaFree, cudaDeviceSynchronize, cudaDeviceSetLimit, cudaMalloc Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/3_CUDA_Features/ptxjit/README.md b/Samples/3_CUDA_Features/ptxjit/README.md index 9f76f157..daef8fdf 100644 --- a/Samples/3_CUDA_Features/ptxjit/README.md +++ b/Samples/3_CUDA_Features/ptxjit/README.md @@ -32,42 +32,5 @@ cudaMalloc, cudaDriverGetVersion, cudaMemcpy, cudaFree Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/3_CUDA_Features/simpleCudaGraphs/README.md b/Samples/3_CUDA_Features/simpleCudaGraphs/README.md index f0f4c86d..8d37abb4 100644 --- a/Samples/3_CUDA_Features/simpleCudaGraphs/README.md +++ b/Samples/3_CUDA_Features/simpleCudaGraphs/README.md @@ -27,42 +27,5 @@ cudaGraphClone, cudaExtent, cudaGraphLaunch, cudaStreamCreate, cudaLaunchHostFun Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/3_CUDA_Features/tf32TensorCoreGemm/README.md b/Samples/3_CUDA_Features/tf32TensorCoreGemm/README.md index d7f38915..8d76f9a4 100644 --- a/Samples/3_CUDA_Features/tf32TensorCoreGemm/README.md +++ b/Samples/3_CUDA_Features/tf32TensorCoreGemm/README.md @@ -33,42 +33,5 @@ cudaMemcpy, cudaFree, cudaGetErrorString, cudaGetLastError, cudaEventSynchronize Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, aarch64. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=aarch64`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/3_CUDA_Features/warpAggregatedAtomicsCG/README.md b/Samples/3_CUDA_Features/warpAggregatedAtomicsCG/README.md index c177226d..10a77acd 100644 --- a/Samples/3_CUDA_Features/warpAggregatedAtomicsCG/README.md +++ b/Samples/3_CUDA_Features/warpAggregatedAtomicsCG/README.md @@ -27,42 +27,5 @@ cudaMemcpy, cudaFree, cudaDeviceGetAttribute, cudaMemset, cudaMalloc Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l, aarch64. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
`$ make TARGET_ARCH=aarch64`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/4_CUDA_Libraries/FilterBorderControlNPP/README.md b/Samples/4_CUDA_Libraries/FilterBorderControlNPP/README.md index 3cd4a91b..b7ff5323 100644 --- a/Samples/4_CUDA_Libraries/FilterBorderControlNPP/README.md +++ b/Samples/4_CUDA_Libraries/FilterBorderControlNPP/README.md @@ -33,42 +33,5 @@ cudaRuntimeGetVersion, cudaDeviceReset, cudaSetDevice, cudaGetDeviceCount, cudaD Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/4_CUDA_Libraries/MersenneTwisterGP11213/README.md b/Samples/4_CUDA_Libraries/MersenneTwisterGP11213/README.md index d64f4f70..52a3efff 100644 --- a/Samples/4_CUDA_Libraries/MersenneTwisterGP11213/README.md +++ b/Samples/4_CUDA_Libraries/MersenneTwisterGP11213/README.md @@ -33,42 +33,5 @@ cudaStreamCreateWithFlags, cudaStreamDestroy, cudaFree, cudaMallocHost, cudaFree Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/4_CUDA_Libraries/batchCUBLAS/README.md b/Samples/4_CUDA_Libraries/batchCUBLAS/README.md index 05325ca7..d9f4bead 100644 --- a/Samples/4_CUDA_Libraries/batchCUBLAS/README.md +++ b/Samples/4_CUDA_Libraries/batchCUBLAS/README.md @@ -36,42 +36,5 @@ cudaMemcpy, cudaGetErrorString, cudaFree, cudaGetLastError, cudaDeviceSynchroniz Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/4_CUDA_Libraries/boxFilterNPP/README.md b/Samples/4_CUDA_Libraries/boxFilterNPP/README.md index 32133125..28562192 100644 --- a/Samples/4_CUDA_Libraries/boxFilterNPP/README.md +++ b/Samples/4_CUDA_Libraries/boxFilterNPP/README.md @@ -33,42 +33,5 @@ cudaRuntimeGetVersion, cudaDriverGetVersion Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/4_CUDA_Libraries/cannyEdgeDetectorNPP/README.md b/Samples/4_CUDA_Libraries/cannyEdgeDetectorNPP/README.md index 32f33996..4ed590bd 100644 --- a/Samples/4_CUDA_Libraries/cannyEdgeDetectorNPP/README.md +++ b/Samples/4_CUDA_Libraries/cannyEdgeDetectorNPP/README.md @@ -33,42 +33,5 @@ cudaRuntimeGetVersion, cudaFree, cudaSetDevice, cudaGetDeviceCount, cudaDeviceIn Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/4_CUDA_Libraries/conjugateGradient/README.md b/Samples/4_CUDA_Libraries/conjugateGradient/README.md index 13fbaa97..c744ae49 100644 --- a/Samples/4_CUDA_Libraries/conjugateGradient/README.md +++ b/Samples/4_CUDA_Libraries/conjugateGradient/README.md @@ -33,42 +33,5 @@ cudaMemcpy, cudaFree, cudaDeviceSynchronize, cudaMalloc, cudaGetDeviceProperties Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/4_CUDA_Libraries/conjugateGradientCudaGraphs/README.md b/Samples/4_CUDA_Libraries/conjugateGradientCudaGraphs/README.md index d7dc1bc2..0af4da12 100644 --- a/Samples/4_CUDA_Libraries/conjugateGradientCudaGraphs/README.md +++ b/Samples/4_CUDA_Libraries/conjugateGradientCudaGraphs/README.md @@ -33,42 +33,5 @@ cudaGraphInstantiate, cudaStreamDestroy, cudaStreamBeginCapture, cudaFree, cudaM Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/4_CUDA_Libraries/conjugateGradientMultiBlockCG/README.md b/Samples/4_CUDA_Libraries/conjugateGradientMultiBlockCG/README.md index 6667faf0..5f7321cf 100644 --- a/Samples/4_CUDA_Libraries/conjugateGradientMultiBlockCG/README.md +++ b/Samples/4_CUDA_Libraries/conjugateGradientMultiBlockCG/README.md @@ -33,42 +33,5 @@ cudaFree, cudaMallocManaged, cudaDeviceSynchronize, cudaEventRecord, cudaLaunchC Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, aarch64. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=aarch64`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/4_CUDA_Libraries/conjugateGradientMultiDeviceCG/README.md b/Samples/4_CUDA_Libraries/conjugateGradientMultiDeviceCG/README.md index 5553a7ef..ac0e8f86 100644 --- a/Samples/4_CUDA_Libraries/conjugateGradientMultiDeviceCG/README.md +++ b/Samples/4_CUDA_Libraries/conjugateGradientMultiDeviceCG/README.md @@ -33,42 +33,5 @@ cudaHostAlloc, cudaMemPrefetchAsync, cudaFree, cudaLaunchCooperativeKernel, cuda Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, aarch64. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=aarch64`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/4_CUDA_Libraries/conjugateGradientPrecond/README.md b/Samples/4_CUDA_Libraries/conjugateGradientPrecond/README.md index a7b0f0ab..e4a3bf59 100644 --- a/Samples/4_CUDA_Libraries/conjugateGradientPrecond/README.md +++ b/Samples/4_CUDA_Libraries/conjugateGradientPrecond/README.md @@ -33,42 +33,5 @@ cudaMemcpy, cudaFree, cudaMemset, cudaMalloc, cudaGetDeviceProperties Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/4_CUDA_Libraries/conjugateGradientUM/README.md b/Samples/4_CUDA_Libraries/conjugateGradientUM/README.md index ac93a42f..de41245b 100644 --- a/Samples/4_CUDA_Libraries/conjugateGradientUM/README.md +++ b/Samples/4_CUDA_Libraries/conjugateGradientUM/README.md @@ -31,42 +31,5 @@ cudaFree, cudaMallocManaged, cudaDeviceSynchronize, cudaMalloc, cudaGetDevicePro Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/4_CUDA_Libraries/cuSolverDn_LinearSolver/README.md b/Samples/4_CUDA_Libraries/cuSolverDn_LinearSolver/README.md index c21dedd4..93b77a28 100644 --- a/Samples/4_CUDA_Libraries/cuSolverDn_LinearSolver/README.md +++ b/Samples/4_CUDA_Libraries/cuSolverDn_LinearSolver/README.md @@ -36,42 +36,5 @@ cudaMemcpy, cudaStreamDestroy, cudaFree, cudaDeviceSynchronize, cudaMemset, cuda Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, aarch64. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=aarch64`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/4_CUDA_Libraries/cuSolverRf/README.md b/Samples/4_CUDA_Libraries/cuSolverRf/README.md index 15dbc164..218232bd 100644 --- a/Samples/4_CUDA_Libraries/cuSolverRf/README.md +++ b/Samples/4_CUDA_Libraries/cuSolverRf/README.md @@ -36,42 +36,5 @@ cudaMemcpy, cudaStreamDestroy, cudaFree, cudaDeviceSynchronize, cudaMalloc, cuda Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/4_CUDA_Libraries/cuSolverSp_LinearSolver/README.md b/Samples/4_CUDA_Libraries/cuSolverSp_LinearSolver/README.md index 5f46bd1f..a4992857 100644 --- a/Samples/4_CUDA_Libraries/cuSolverSp_LinearSolver/README.md +++ b/Samples/4_CUDA_Libraries/cuSolverSp_LinearSolver/README.md @@ -36,42 +36,5 @@ cudaStreamDestroy, cudaFree, cudaDeviceSynchronize, cudaMalloc, cudaStreamCreate Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/4_CUDA_Libraries/cuSolverSp_LowlevelCholesky/README.md b/Samples/4_CUDA_Libraries/cuSolverSp_LowlevelCholesky/README.md index 7273ac70..b15f2fbe 100644 --- a/Samples/4_CUDA_Libraries/cuSolverSp_LowlevelCholesky/README.md +++ b/Samples/4_CUDA_Libraries/cuSolverSp_LowlevelCholesky/README.md @@ -36,42 +36,5 @@ cudaMemcpy, cudaStreamDestroy, cudaFree, cudaMalloc, cudaStreamCreate Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/4_CUDA_Libraries/cuSolverSp_LowlevelQR/README.md b/Samples/4_CUDA_Libraries/cuSolverSp_LowlevelQR/README.md index a3c2ad5d..f6d63010 100644 --- a/Samples/4_CUDA_Libraries/cuSolverSp_LowlevelQR/README.md +++ b/Samples/4_CUDA_Libraries/cuSolverSp_LowlevelQR/README.md @@ -36,42 +36,5 @@ cudaMemcpy, cudaStreamDestroy, cudaFree, cudaMalloc, cudaStreamCreate Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/4_CUDA_Libraries/freeImageInteropNPP/README.md b/Samples/4_CUDA_Libraries/freeImageInteropNPP/README.md index 1c467a99..50eb6dd3 100644 --- a/Samples/4_CUDA_Libraries/freeImageInteropNPP/README.md +++ b/Samples/4_CUDA_Libraries/freeImageInteropNPP/README.md @@ -33,42 +33,5 @@ cudaRuntimeGetVersion, cudaSetDevice, cudaGetDeviceCount, cudaDeviceInit, cudaDr Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/4_CUDA_Libraries/histEqualizationNPP/README.md b/Samples/4_CUDA_Libraries/histEqualizationNPP/README.md index 0f19949b..fc73c129 100644 --- a/Samples/4_CUDA_Libraries/histEqualizationNPP/README.md +++ b/Samples/4_CUDA_Libraries/histEqualizationNPP/README.md @@ -33,42 +33,5 @@ cudaRuntimeGetVersion, cudaMemcpy, cudaFree, cudaSetDevice, cudaGetDeviceCount, Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/4_CUDA_Libraries/jitLto/README.md b/Samples/4_CUDA_Libraries/jitLto/README.md index d1bf0e4e..afda50be 100644 --- a/Samples/4_CUDA_Libraries/jitLto/README.md +++ b/Samples/4_CUDA_Libraries/jitLto/README.md @@ -33,42 +33,5 @@ cuModuleLoad, cuModuleLoadDataEx, cuModuleGetFunction, cuMemAlloc, cuMemFree, cu Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, aarch64. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=aarch64`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/4_CUDA_Libraries/lineOfSight/README.md b/Samples/4_CUDA_Libraries/lineOfSight/README.md index 41215bd8..20044ade 100644 --- a/Samples/4_CUDA_Libraries/lineOfSight/README.md +++ b/Samples/4_CUDA_Libraries/lineOfSight/README.md @@ -29,42 +29,5 @@ cudaCreateChannelDesc, cudaMallocArray, cudaFreeArray, cudaDeviceSynchronize, cu Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/4_CUDA_Libraries/matrixMulCUBLAS/README.md b/Samples/4_CUDA_Libraries/matrixMulCUBLAS/README.md index b21ed988..f6ab6e30 100644 --- a/Samples/4_CUDA_Libraries/matrixMulCUBLAS/README.md +++ b/Samples/4_CUDA_Libraries/matrixMulCUBLAS/README.md @@ -33,42 +33,5 @@ cudaMemcpy, cudaFree, cudaEventSynchronize, cudaEventRecord, cudaMalloc, cudaEve Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/4_CUDA_Libraries/nvJPEG/README.md b/Samples/4_CUDA_Libraries/nvJPEG/README.md index 08f1be00..ad375ea0 100644 --- a/Samples/4_CUDA_Libraries/nvJPEG/README.md +++ b/Samples/4_CUDA_Libraries/nvJPEG/README.md @@ -31,42 +31,5 @@ cudaHostAlloc, cudaStreamCreateWithFlags, cudaStreamDestroy, cudaFree, cudaEvent Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, aarch64. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=aarch64`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/4_CUDA_Libraries/nvJPEG_encoder/README.md b/Samples/4_CUDA_Libraries/nvJPEG_encoder/README.md index cd3fbb21..7526b29b 100644 --- a/Samples/4_CUDA_Libraries/nvJPEG_encoder/README.md +++ b/Samples/4_CUDA_Libraries/nvJPEG_encoder/README.md @@ -31,42 +31,5 @@ cudaFree, cudaGetErrorString, cudaEventSynchronize, cudaDeviceSynchronize, cudaE Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, aarch64. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=aarch64`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/4_CUDA_Libraries/oceanFFT/README.md b/Samples/4_CUDA_Libraries/oceanFFT/README.md index fe6f0713..de66ebbe 100644 --- a/Samples/4_CUDA_Libraries/oceanFFT/README.md +++ b/Samples/4_CUDA_Libraries/oceanFFT/README.md @@ -33,42 +33,5 @@ cudaGraphicsUnmapResources, cudaMemcpy, cudaMalloc, cudaFree, cudaGraphicsResour Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/4_CUDA_Libraries/randomFog/README.md b/Samples/4_CUDA_Libraries/randomFog/README.md index 1181ed59..9291a64d 100644 --- a/Samples/4_CUDA_Libraries/randomFog/README.md +++ b/Samples/4_CUDA_Libraries/randomFog/README.md @@ -33,42 +33,5 @@ cudaMalloc, cudaGetErrorString, cudaMemcpy, cudaFree Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/4_CUDA_Libraries/simpleCUBLAS/README.md b/Samples/4_CUDA_Libraries/simpleCUBLAS/README.md index 64495d61..6794be67 100644 --- a/Samples/4_CUDA_Libraries/simpleCUBLAS/README.md +++ b/Samples/4_CUDA_Libraries/simpleCUBLAS/README.md @@ -33,42 +33,5 @@ cudaMalloc, cudaFree Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l, aarch64. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
`$ make TARGET_ARCH=aarch64`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/4_CUDA_Libraries/simpleCUBLASXT/README.md b/Samples/4_CUDA_Libraries/simpleCUBLASXT/README.md index 443a3218..8c11171f 100644 --- a/Samples/4_CUDA_Libraries/simpleCUBLASXT/README.md +++ b/Samples/4_CUDA_Libraries/simpleCUBLASXT/README.md @@ -33,42 +33,5 @@ cudaGetDeviceProperties, cudaGetDeviceCount, cudaFree Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, aarch64. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=aarch64`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/4_CUDA_Libraries/simpleCUBLAS_LU/README.md b/Samples/4_CUDA_Libraries/simpleCUBLAS_LU/README.md index 4605545a..b2d2a5eb 100644 --- a/Samples/4_CUDA_Libraries/simpleCUBLAS_LU/README.md +++ b/Samples/4_CUDA_Libraries/simpleCUBLAS_LU/README.md @@ -33,42 +33,5 @@ cudaGetErrorEnum, cudaMalloc, cudaMemcpy, cudaFree Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, aarch64. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=aarch64`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/4_CUDA_Libraries/simpleCUFFT/README.md b/Samples/4_CUDA_Libraries/simpleCUFFT/README.md index cddeb1d5..7c390b87 100644 --- a/Samples/4_CUDA_Libraries/simpleCUFFT/README.md +++ b/Samples/4_CUDA_Libraries/simpleCUFFT/README.md @@ -33,42 +33,5 @@ cudaMalloc, cudaMemcpy, cudaFree Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l, aarch64. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
`$ make TARGET_ARCH=aarch64`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/4_CUDA_Libraries/simpleCUFFT_2d_MGPU/README.md b/Samples/4_CUDA_Libraries/simpleCUFFT_2d_MGPU/README.md index bbf62e0e..ca8c2ba4 100644 --- a/Samples/4_CUDA_Libraries/simpleCUFFT_2d_MGPU/README.md +++ b/Samples/4_CUDA_Libraries/simpleCUFFT_2d_MGPU/README.md @@ -33,42 +33,5 @@ cudaXtFree, cudaMemcpy, cudaFree, cudaSetDevice, cudaGetDeviceCount, cudaDeviceS Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/4_CUDA_Libraries/simpleCUFFT_MGPU/README.md b/Samples/4_CUDA_Libraries/simpleCUFFT_MGPU/README.md index 32795573..f5a33254 100644 --- a/Samples/4_CUDA_Libraries/simpleCUFFT_MGPU/README.md +++ b/Samples/4_CUDA_Libraries/simpleCUFFT_MGPU/README.md @@ -33,42 +33,5 @@ cudaXtFree, cudaSetDevice, cudaGetDeviceCount, cudaDeviceSynchronize, cudaGetDev Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/4_CUDA_Libraries/simpleCUFFT_callback/README.md b/Samples/4_CUDA_Libraries/simpleCUFFT_callback/README.md index 7e3d8a46..155498af 100644 --- a/Samples/4_CUDA_Libraries/simpleCUFFT_callback/README.md +++ b/Samples/4_CUDA_Libraries/simpleCUFFT_callback/README.md @@ -33,32 +33,5 @@ cudaMemcpy, cudaFree, cudaMemcpyFromSymbol, cudaGetDevice, cudaMalloc, cudaGetDe Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, aarch64. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=aarch64`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/4_CUDA_Libraries/watershedSegmentationNPP/README.md b/Samples/4_CUDA_Libraries/watershedSegmentationNPP/README.md index ab548cab..2747ffb3 100644 --- a/Samples/4_CUDA_Libraries/watershedSegmentationNPP/README.md +++ b/Samples/4_CUDA_Libraries/watershedSegmentationNPP/README.md @@ -33,42 +33,5 @@ cudaRuntimeGetVersion, cudaFree, cudaDeviceGetAttribute, cudaDriverGetVersion, c Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/5_Domain_Specific/BlackScholes/README.md b/Samples/5_Domain_Specific/BlackScholes/README.md index 3ea0baab..01f1aff4 100644 --- a/Samples/5_Domain_Specific/BlackScholes/README.md +++ b/Samples/5_Domain_Specific/BlackScholes/README.md @@ -29,43 +29,6 @@ cudaMalloc, cudaDeviceSynchronize, cudaMemcpy, cudaFree Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) [whitepaper](./doc/BlackScholes.pdf) diff --git a/Samples/5_Domain_Specific/BlackScholes_nvrtc/README.md b/Samples/5_Domain_Specific/BlackScholes_nvrtc/README.md index d2cb2092..61e3bb63 100644 --- a/Samples/5_Domain_Specific/BlackScholes_nvrtc/README.md +++ b/Samples/5_Domain_Specific/BlackScholes_nvrtc/README.md @@ -36,42 +36,5 @@ cudaBlockSize, cudaGridSize Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, aarch64. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=aarch64`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/5_Domain_Specific/FDTD3d/README.md b/Samples/5_Domain_Specific/FDTD3d/README.md index f4cf1037..3ed51f9a 100644 --- a/Samples/5_Domain_Specific/FDTD3d/README.md +++ b/Samples/5_Domain_Specific/FDTD3d/README.md @@ -29,42 +29,5 @@ cudaMemcpy, cudaMalloc, cudaFree, cudaFuncGetAttributes, cudaSetDevice, cudaGetD Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/5_Domain_Specific/HSOpticalFlow/README.md b/Samples/5_Domain_Specific/HSOpticalFlow/README.md index c48b998d..5651f8e9 100644 --- a/Samples/5_Domain_Specific/HSOpticalFlow/README.md +++ b/Samples/5_Domain_Specific/HSOpticalFlow/README.md @@ -29,43 +29,6 @@ cudaMalloc, cudaMemcpy, cudaMemset, cudaFree Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) [whitepaper](./doc/OpticalFlow.pdf) diff --git a/Samples/5_Domain_Specific/Mandelbrot/README.md b/Samples/5_Domain_Specific/Mandelbrot/README.md index 49605ec3..cf973bae 100644 --- a/Samples/5_Domain_Specific/Mandelbrot/README.md +++ b/Samples/5_Domain_Specific/Mandelbrot/README.md @@ -33,42 +33,5 @@ cudaGLUnmapBufferObject, cudaGraphicsUnmapResources, cudaMemcpy, cudaFree, cudaG Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/5_Domain_Specific/MonteCarloMultiGPU/README.md b/Samples/5_Domain_Specific/MonteCarloMultiGPU/README.md index dc1429ae..f1e06c62 100644 --- a/Samples/5_Domain_Specific/MonteCarloMultiGPU/README.md +++ b/Samples/5_Domain_Specific/MonteCarloMultiGPU/README.md @@ -33,43 +33,6 @@ cudaStreamDestroy, cudaMalloc, cudaFree, cudaMallocHost, cudaSetDevice, cudaEven Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) [whitepaper](./doc/MonteCarlo.pdf) diff --git a/Samples/5_Domain_Specific/NV12toBGRandResize/README.md b/Samples/5_Domain_Specific/NV12toBGRandResize/README.md index f6c3eee6..82600663 100644 --- a/Samples/5_Domain_Specific/NV12toBGRandResize/README.md +++ b/Samples/5_Domain_Specific/NV12toBGRandResize/README.md @@ -29,42 +29,5 @@ cudaMemcpy, cudaStreamDestroy, cudaMalloc, cudaFree, cudaMallocManaged, cudaStre Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, aarch64. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=aarch64`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/5_Domain_Specific/SobelFilter/README.md b/Samples/5_Domain_Specific/SobelFilter/README.md index 7691119e..7d129afb 100644 --- a/Samples/5_Domain_Specific/SobelFilter/README.md +++ b/Samples/5_Domain_Specific/SobelFilter/README.md @@ -33,42 +33,5 @@ cudaGraphicsUnmapResources, cudaMemcpy, cudaMallocArray, cudaFreeArray, cudaFree Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/5_Domain_Specific/SobolQRNG/README.md b/Samples/5_Domain_Specific/SobolQRNG/README.md index 5ead5d1b..880d76e0 100644 --- a/Samples/5_Domain_Specific/SobolQRNG/README.md +++ b/Samples/5_Domain_Specific/SobolQRNG/README.md @@ -29,42 +29,5 @@ cudaMemcpy, cudaGetErrorString, cudaFree, cudaDeviceSynchronize, cudaGetDevice, Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/5_Domain_Specific/bicubicTexture/README.md b/Samples/5_Domain_Specific/bicubicTexture/README.md index c3f31ce9..b4824803 100644 --- a/Samples/5_Domain_Specific/bicubicTexture/README.md +++ b/Samples/5_Domain_Specific/bicubicTexture/README.md @@ -33,42 +33,5 @@ cudaGraphicsUnmapResources, cudaCreateChannelDesc, cudaMallocArray, cudaFreeArra Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/5_Domain_Specific/bilateralFilter/README.md b/Samples/5_Domain_Specific/bilateralFilter/README.md index 12aa41b8..e0bffcad 100644 --- a/Samples/5_Domain_Specific/bilateralFilter/README.md +++ b/Samples/5_Domain_Specific/bilateralFilter/README.md @@ -33,42 +33,5 @@ cudaRuntimeGetVersion, cudaGraphicsUnmapResources, cudaMallocPitch, cudaFree, cu Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/5_Domain_Specific/binomialOptions/README.md b/Samples/5_Domain_Specific/binomialOptions/README.md index d0da7371..27d2cb3c 100644 --- a/Samples/5_Domain_Specific/binomialOptions/README.md +++ b/Samples/5_Domain_Specific/binomialOptions/README.md @@ -29,43 +29,6 @@ cudaDeviceSynchronize, cudaMemcpyToSymbol, cudaMemcpyFromSymbol Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) [whitepaper](./doc/binomialOptions.pdf) diff --git a/Samples/5_Domain_Specific/binomialOptions_nvrtc/README.md b/Samples/5_Domain_Specific/binomialOptions_nvrtc/README.md index 6e81d3f4..6235bab7 100644 --- a/Samples/5_Domain_Specific/binomialOptions_nvrtc/README.md +++ b/Samples/5_Domain_Specific/binomialOptions_nvrtc/README.md @@ -36,42 +36,5 @@ cudaBlockSize, cudaGridSize Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, aarch64. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=aarch64`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/5_Domain_Specific/convolutionFFT2D/README.md b/Samples/5_Domain_Specific/convolutionFFT2D/README.md index 7a3c3740..f3984f54 100644 --- a/Samples/5_Domain_Specific/convolutionFFT2D/README.md +++ b/Samples/5_Domain_Specific/convolutionFFT2D/README.md @@ -33,42 +33,5 @@ cudaMemcpy, cudaFree, cudaDestroyTextureObject, cudaDeviceSynchronize, cudaCreat Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/5_Domain_Specific/dwtHaar1D/README.md b/Samples/5_Domain_Specific/dwtHaar1D/README.md index 23a55b77..035660dd 100644 --- a/Samples/5_Domain_Specific/dwtHaar1D/README.md +++ b/Samples/5_Domain_Specific/dwtHaar1D/README.md @@ -29,42 +29,5 @@ cudaMalloc, cudaMemcpy, cudaFree Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/5_Domain_Specific/dxtc/README.md b/Samples/5_Domain_Specific/dxtc/README.md index c3ff39eb..7d444be7 100644 --- a/Samples/5_Domain_Specific/dxtc/README.md +++ b/Samples/5_Domain_Specific/dxtc/README.md @@ -29,43 +29,6 @@ cudaMemcpy, cudaFree, cudaDeviceSynchronize, cudaGetDevice, cudaMalloc, cudaGetD Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) [whitepaper](./doc/cuda_dxtc.pdf) diff --git a/Samples/5_Domain_Specific/fastWalshTransform/README.md b/Samples/5_Domain_Specific/fastWalshTransform/README.md index 46236c1c..d7aad60e 100644 --- a/Samples/5_Domain_Specific/fastWalshTransform/README.md +++ b/Samples/5_Domain_Specific/fastWalshTransform/README.md @@ -29,42 +29,5 @@ cudaMemcpy, cudaFree, cudaDeviceSynchronize, cudaMemset, cudaMalloc Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/5_Domain_Specific/fluidsGL/README.md b/Samples/5_Domain_Specific/fluidsGL/README.md index fa66f2f0..7f3b3a4d 100644 --- a/Samples/5_Domain_Specific/fluidsGL/README.md +++ b/Samples/5_Domain_Specific/fluidsGL/README.md @@ -33,43 +33,6 @@ cudaGraphicsUnmapResources, cudaMemcpy, cudaMallocArray, cudaFreeArray, cudaFree Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) [whitepaper](./doc/fluidsGL.pdf) diff --git a/Samples/5_Domain_Specific/fluidsGLES/README.md b/Samples/5_Domain_Specific/fluidsGLES/README.md index 1abac27e..84a5ce8b 100644 --- a/Samples/5_Domain_Specific/fluidsGLES/README.md +++ b/Samples/5_Domain_Specific/fluidsGLES/README.md @@ -33,32 +33,5 @@ cudaGraphicsUnmapResources, cudaMemcpy, cudaMallocArray, cudaFreeArray, cudaFree Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/5_Domain_Specific/marchingCubes/README.md b/Samples/5_Domain_Specific/marchingCubes/README.md index ae1a2f4d..e9095565 100644 --- a/Samples/5_Domain_Specific/marchingCubes/README.md +++ b/Samples/5_Domain_Specific/marchingCubes/README.md @@ -33,42 +33,5 @@ cudaGLUnmapBufferObject, cudaGraphicsUnmapResources, cudaCreateChannelDesc, cuda Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/5_Domain_Specific/nbody/README.md b/Samples/5_Domain_Specific/nbody/README.md index 35a76ed8..bb4036bb 100644 --- a/Samples/5_Domain_Specific/nbody/README.md +++ b/Samples/5_Domain_Specific/nbody/README.md @@ -33,43 +33,6 @@ cudaGraphicsUnmapResources, cudaSetDeviceFlags, cudaGraphicsResourceSetMapFlags, Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) [whitepaper](./doc/nbody_gems3_ch31.pdf) diff --git a/Samples/5_Domain_Specific/nbody_opengles/README.md b/Samples/5_Domain_Specific/nbody_opengles/README.md index f8c93485..2dda4e15 100644 --- a/Samples/5_Domain_Specific/nbody_opengles/README.md +++ b/Samples/5_Domain_Specific/nbody_opengles/README.md @@ -33,32 +33,5 @@ cudaGraphicsUnmapResources, cudaSetDeviceFlags, cudaGraphicsResourceSetMapFlags, Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/5_Domain_Specific/nbody_screen/README.md b/Samples/5_Domain_Specific/nbody_screen/README.md index 834bdef8..3d8a057a 100644 --- a/Samples/5_Domain_Specific/nbody_screen/README.md +++ b/Samples/5_Domain_Specific/nbody_screen/README.md @@ -33,32 +33,5 @@ cudaGraphicsUnmapResources, cudaSetDeviceFlags, cudaGraphicsResourceSetMapFlags, Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are aarch64. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=aarch64`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/5_Domain_Specific/p2pBandwidthLatencyTest/README.md b/Samples/5_Domain_Specific/p2pBandwidthLatencyTest/README.md index d1be7ce9..c9492fad 100644 --- a/Samples/5_Domain_Specific/p2pBandwidthLatencyTest/README.md +++ b/Samples/5_Domain_Specific/p2pBandwidthLatencyTest/README.md @@ -29,42 +29,5 @@ cudaSetDevice, cudaEventDestroy, cudaOccupancyMaxPotentialBlockSize, cudaCheckEr Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/5_Domain_Specific/postProcessGL/README.md b/Samples/5_Domain_Specific/postProcessGL/README.md index b89f0349..4abab5d2 100644 --- a/Samples/5_Domain_Specific/postProcessGL/README.md +++ b/Samples/5_Domain_Specific/postProcessGL/README.md @@ -33,42 +33,5 @@ cudaHostAlloc, cudaGraphicsUnmapResources, cudaMalloc, cudaFree, cudaGetChannelD Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/5_Domain_Specific/quasirandomGenerator/README.md b/Samples/5_Domain_Specific/quasirandomGenerator/README.md index d2edaea2..d49f9039 100644 --- a/Samples/5_Domain_Specific/quasirandomGenerator/README.md +++ b/Samples/5_Domain_Specific/quasirandomGenerator/README.md @@ -29,42 +29,5 @@ cudaMemcpy, cudaFree, cudaDeviceSynchronize, cudaMemset, cudaMemcpyToSymbol, cud Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/5_Domain_Specific/quasirandomGenerator_nvrtc/README.md b/Samples/5_Domain_Specific/quasirandomGenerator_nvrtc/README.md index 9d169869..32b4c370 100644 --- a/Samples/5_Domain_Specific/quasirandomGenerator_nvrtc/README.md +++ b/Samples/5_Domain_Specific/quasirandomGenerator_nvrtc/README.md @@ -33,42 +33,5 @@ cuMemcpyDtoH, cuMemAlloc, cuMemFree Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, aarch64. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=aarch64`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/5_Domain_Specific/recursiveGaussian/README.md b/Samples/5_Domain_Specific/recursiveGaussian/README.md index 090f676f..6c9dc6dc 100644 --- a/Samples/5_Domain_Specific/recursiveGaussian/README.md +++ b/Samples/5_Domain_Specific/recursiveGaussian/README.md @@ -33,42 +33,5 @@ cudaGraphicsUnmapResources, cudaMemcpy, cudaFree, cudaGraphicsResourceGetMappedP Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/5_Domain_Specific/simpleD3D11/README.md b/Samples/5_Domain_Specific/simpleD3D11/README.md index 8b9f5e01..a3c7b2c6 100644 --- a/Samples/5_Domain_Specific/simpleD3D11/README.md +++ b/Samples/5_Domain_Specific/simpleD3D11/README.md @@ -33,17 +33,5 @@ cudaImportKeyedMutex, cudaExternalMemoryGetMappedBuffer, cudaStreamCreateWithFla Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - ## References (for more details) diff --git a/Samples/5_Domain_Specific/simpleD3D11Texture/README.md b/Samples/5_Domain_Specific/simpleD3D11Texture/README.md index 19aaf8a5..526e1e44 100644 --- a/Samples/5_Domain_Specific/simpleD3D11Texture/README.md +++ b/Samples/5_Domain_Specific/simpleD3D11Texture/README.md @@ -33,17 +33,5 @@ cudaGraphicsUnmapResources, cudaMalloc, cudaMallocPitch, cudaGetErrorString, cud Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - ## References (for more details) diff --git a/Samples/5_Domain_Specific/simpleD3D12/README.md b/Samples/5_Domain_Specific/simpleD3D12/README.md index ad39407f..4b3e6413 100644 --- a/Samples/5_Domain_Specific/simpleD3D12/README.md +++ b/Samples/5_Domain_Specific/simpleD3D12/README.md @@ -33,17 +33,5 @@ cudaWaitExternalSemaphoresAsync, cudaExternalMemoryGetMappedBuffer, cudaImportEx Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - ## References (for more details) diff --git a/Samples/5_Domain_Specific/simpleGL/README.md b/Samples/5_Domain_Specific/simpleGL/README.md index 2419f043..0019a36b 100644 --- a/Samples/5_Domain_Specific/simpleGL/README.md +++ b/Samples/5_Domain_Specific/simpleGL/README.md @@ -33,42 +33,5 @@ cudaGraphicsUnmapResources, cudaMemcpy, cudaFree, cudaGraphicsResourceGetMappedP Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/5_Domain_Specific/simpleGLES/README.md b/Samples/5_Domain_Specific/simpleGLES/README.md index f22b105f..f90fbc58 100644 --- a/Samples/5_Domain_Specific/simpleGLES/README.md +++ b/Samples/5_Domain_Specific/simpleGLES/README.md @@ -33,32 +33,5 @@ cudaGraphicsUnmapResources, cudaMemcpy, cudaFree, cudaGraphicsResourceGetMappedP Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/5_Domain_Specific/simpleGLES_EGLOutput/README.md b/Samples/5_Domain_Specific/simpleGLES_EGLOutput/README.md index 9af216b4..d79ce6e6 100644 --- a/Samples/5_Domain_Specific/simpleGLES_EGLOutput/README.md +++ b/Samples/5_Domain_Specific/simpleGLES_EGLOutput/README.md @@ -38,32 +38,5 @@ cudaGraphicsUnmapResources, cudaMemcpy, cudaFree, cudaGraphicsResourceGetMappedP Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/5_Domain_Specific/simpleGLES_screen/README.md b/Samples/5_Domain_Specific/simpleGLES_screen/README.md index 067bdaf4..b1f72fa1 100644 --- a/Samples/5_Domain_Specific/simpleGLES_screen/README.md +++ b/Samples/5_Domain_Specific/simpleGLES_screen/README.md @@ -33,32 +33,5 @@ cudaGraphicsUnmapResources, cudaMemcpy, cudaFree, cudaGraphicsResourceGetMappedP Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are aarch64. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=aarch64`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/5_Domain_Specific/simpleVulkan/README.md b/Samples/5_Domain_Specific/simpleVulkan/README.md index 6d1a3282..d664cdd1 100644 --- a/Samples/5_Domain_Specific/simpleVulkan/README.md +++ b/Samples/5_Domain_Specific/simpleVulkan/README.md @@ -33,42 +33,5 @@ cudaStreamCreateWithFlags, cudaExternalMemoryGetMappedBuffer, cudaSignalSemaphor Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, aarch64. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=aarch64`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/5_Domain_Specific/simpleVulkanMMAP/README.md b/Samples/5_Domain_Specific/simpleVulkanMMAP/README.md index ba2bd918..9603b06e 100644 --- a/Samples/5_Domain_Specific/simpleVulkanMMAP/README.md +++ b/Samples/5_Domain_Specific/simpleVulkanMMAP/README.md @@ -36,42 +36,5 @@ cudaWaitExternalSemaphoresAsync, cudaImportExternalSemaphore, cudaDeviceGetAttri Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, aarch64. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=aarch64`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/5_Domain_Specific/smokeParticles/README.md b/Samples/5_Domain_Specific/smokeParticles/README.md index 366dea3d..e2bee3f4 100644 --- a/Samples/5_Domain_Specific/smokeParticles/README.md +++ b/Samples/5_Domain_Specific/smokeParticles/README.md @@ -33,43 +33,6 @@ cudaExtent, cudaPitchedPtr, cudaCreateTextureObject, cudaMemcpyToSymbol Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) [whitepaper](./doc/smokeParticles.pdf) diff --git a/Samples/5_Domain_Specific/stereoDisparity/README.md b/Samples/5_Domain_Specific/stereoDisparity/README.md index 182b109c..5bf7814d 100644 --- a/Samples/5_Domain_Specific/stereoDisparity/README.md +++ b/Samples/5_Domain_Specific/stereoDisparity/README.md @@ -29,42 +29,5 @@ cudaMemcpy, cudaFree, cudaEventSynchronize, cudaDeviceSynchronize, cudaCreateTex Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/5_Domain_Specific/volumeFiltering/README.md b/Samples/5_Domain_Specific/volumeFiltering/README.md index f07c7b27..10bd00e0 100644 --- a/Samples/5_Domain_Specific/volumeFiltering/README.md +++ b/Samples/5_Domain_Specific/volumeFiltering/README.md @@ -33,42 +33,5 @@ cudaMemcpy, cudaGraphicsMapResources, cudaDestroySurfaceObject, cudaExtent, cuda Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/5_Domain_Specific/volumeRender/README.md b/Samples/5_Domain_Specific/volumeRender/README.md index 57e40b0f..3a452362 100644 --- a/Samples/5_Domain_Specific/volumeRender/README.md +++ b/Samples/5_Domain_Specific/volumeRender/README.md @@ -33,42 +33,5 @@ cudaProfilerStop, cudaGraphicsUnmapResources, cudaMemcpy, cudaMallocArray, cudaF Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/5_Domain_Specific/vulkanImageCUDA/README.md b/Samples/5_Domain_Specific/vulkanImageCUDA/README.md index c0cfa7f8..44ef3f62 100644 --- a/Samples/5_Domain_Specific/vulkanImageCUDA/README.md +++ b/Samples/5_Domain_Specific/vulkanImageCUDA/README.md @@ -33,42 +33,5 @@ cudaVkSemaphoreSignal, cudaWaitExternalSemaphoresAsync, cudaMemcpy, cudaVkImport Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, aarch64. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=aarch64`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/6_Performance/LargeKernelParameter/README.md b/Samples/6_Performance/LargeKernelParameter/README.md index e91e7c51..9c5e1a72 100644 --- a/Samples/6_Performance/LargeKernelParameter/README.md +++ b/Samples/6_Performance/LargeKernelParameter/README.md @@ -29,42 +29,5 @@ cudaMemcpyToSymbol, cudaMemcpy, cudaFree, cudaDeviceSynchronize, cudaMalloc Download and install the [CUDA Toolkit 12.1](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/6_Performance/UnifiedMemoryPerf/README.md b/Samples/6_Performance/UnifiedMemoryPerf/README.md index dd9a9236..85bdb317 100644 --- a/Samples/6_Performance/UnifiedMemoryPerf/README.md +++ b/Samples/6_Performance/UnifiedMemoryPerf/README.md @@ -31,42 +31,5 @@ cudaMemcpy, cudaStreamDestroy, cudaMemPrefetchAsync, cudaFree, cudaMallocHost, c Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l, aarch64. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
`$ make TARGET_ARCH=aarch64`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/6_Performance/alignedTypes/README.md b/Samples/6_Performance/alignedTypes/README.md index 8ac983e6..8f3e3d3c 100644 --- a/Samples/6_Performance/alignedTypes/README.md +++ b/Samples/6_Performance/alignedTypes/README.md @@ -29,42 +29,5 @@ cudaMemcpy, cudaFree, cudaDeviceSynchronize, cudaMemset, cudaMalloc, cudaGetDevi Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/6_Performance/cudaGraphsPerfScaling/README.md b/Samples/6_Performance/cudaGraphsPerfScaling/README.md index dce334be..23cee264 100644 --- a/Samples/6_Performance/cudaGraphsPerfScaling/README.md +++ b/Samples/6_Performance/cudaGraphsPerfScaling/README.md @@ -29,42 +29,5 @@ cudaStreamBeginCapture, cudaGraphInstantiate, cudaGraphLaunch, cudaGraphUpload Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/6_Performance/transpose/README.md b/Samples/6_Performance/transpose/README.md index 32c10bb1..dc9401f9 100644 --- a/Samples/6_Performance/transpose/README.md +++ b/Samples/6_Performance/transpose/README.md @@ -29,43 +29,6 @@ cudaMemcpy, cudaMalloc, cudaFree, cudaGetLastError, cudaEventSynchronize, cudaEv Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. -## Build and Run - -### Windows -The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format: -``` -*_vs.sln - for Visual Studio -``` -Each individual sample has its own set of solution files in its directory: - -To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used. -> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details." - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
`$ make TARGET_ARCH=armv7l`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) [whitepaper](./doc/MatrixTranspose.pdf) diff --git a/Samples/8_Platform_Specific/Tegra/cuDLAErrorReporting/README.md b/Samples/8_Platform_Specific/Tegra/cuDLAErrorReporting/README.md index 38db6c24..156c6d60 100644 --- a/Samples/8_Platform_Specific/Tegra/cuDLAErrorReporting/README.md +++ b/Samples/8_Platform_Specific/Tegra/cuDLAErrorReporting/README.md @@ -29,32 +29,5 @@ cudaStreamCreateWithFlags, cudaStreamDestroy, cudaFree, cudaGetErrorName, cudaSe Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. -## Build and Run - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are aarch64. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=aarch64`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/8_Platform_Specific/Tegra/cuDLAHybridMode/README.md b/Samples/8_Platform_Specific/Tegra/cuDLAHybridMode/README.md index 057f7943..286a52e8 100644 --- a/Samples/8_Platform_Specific/Tegra/cuDLAHybridMode/README.md +++ b/Samples/8_Platform_Specific/Tegra/cuDLAHybridMode/README.md @@ -29,32 +29,5 @@ cudaStreamCreateWithFlags, cudaStreamDestroy, cudaFree, cudaGetErrorName, cudaSe Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. -## Build and Run - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are aarch64. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=aarch64`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/8_Platform_Specific/Tegra/cuDLALayerwiseStatsHybrid/README.md b/Samples/8_Platform_Specific/Tegra/cuDLALayerwiseStatsHybrid/README.md index 0a765e8b..330a8513 100644 --- a/Samples/8_Platform_Specific/Tegra/cuDLALayerwiseStatsHybrid/README.md +++ b/Samples/8_Platform_Specific/Tegra/cuDLALayerwiseStatsHybrid/README.md @@ -29,32 +29,5 @@ cudaStreamCreateWithFlags, cudaStreamDestroy, cudaFree, cudaGetErrorName, cudaSe Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. -## Build and Run - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are aarch64. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=aarch64`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/8_Platform_Specific/Tegra/cuDLALayerwiseStatsStandalone/README.md b/Samples/8_Platform_Specific/Tegra/cuDLALayerwiseStatsStandalone/README.md index fc3e8504..403afbfe 100644 --- a/Samples/8_Platform_Specific/Tegra/cuDLALayerwiseStatsStandalone/README.md +++ b/Samples/8_Platform_Specific/Tegra/cuDLALayerwiseStatsStandalone/README.md @@ -30,32 +30,5 @@ aarch64 Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are aarch64. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=aarch64`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/8_Platform_Specific/Tegra/cuDLAStandaloneMode/README.md b/Samples/8_Platform_Specific/Tegra/cuDLAStandaloneMode/README.md index 8f724901..099ace0b 100644 --- a/Samples/8_Platform_Specific/Tegra/cuDLAStandaloneMode/README.md +++ b/Samples/8_Platform_Specific/Tegra/cuDLAStandaloneMode/README.md @@ -30,32 +30,5 @@ aarch64 Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are aarch64. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=aarch64`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/8_Platform_Specific/Tegra/cudaNvSci/README.md b/Samples/8_Platform_Specific/Tegra/cudaNvSci/README.md index 06dd3e56..b6338aa5 100644 --- a/Samples/8_Platform_Specific/Tegra/cudaNvSci/README.md +++ b/Samples/8_Platform_Specific/Tegra/cudaNvSci/README.md @@ -36,32 +36,5 @@ cudaExternalMemoryGetMappedBuffer, cudaImportExternalSemaphore, cudaDeviceGetAtt Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, aarch64. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=aarch64`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details) diff --git a/Samples/8_Platform_Specific/Tegra/cudaNvSciNvMedia/README.md b/Samples/8_Platform_Specific/Tegra/cudaNvSciNvMedia/README.md index 8106f9c1..40ca89ed 100644 --- a/Samples/8_Platform_Specific/Tegra/cudaNvSciNvMedia/README.md +++ b/Samples/8_Platform_Specific/Tegra/cudaNvSciNvMedia/README.md @@ -36,32 +36,5 @@ cudaImportExternalSemaphore, cudaGetMipmappedArrayLevel, cudaSetDevice, cudaDest Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. Make sure the dependencies mentioned in [Dependencies]() section above are installed. -## Build and Run - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are aarch64. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=aarch64`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - ## References (for more details)