mirror of
https://github.com/NVIDIA/cuda-samples.git
synced 2026-10-13 00:08:26 +08:00
CUDA 13.2 samples update (#432)
- Added Python samples for CUDA Python 1.0 release - Renamed top-level `Samples` directory to `cpp` to accommodate Python samples.
This commit is contained in:
Dheemanth
GitHub
18
cpp/4_CUDA_Libraries/jitLto/.vscode/c_cpp_properties.json
vendored
Normal file
18
cpp/4_CUDA_Libraries/jitLto/.vscode/c_cpp_properties.json
vendored
Normal file
@@ -0,0 +1,18 @@
|
||||
{
|
||||
"configurations": [
|
||||
{
|
||||
"name": "Linux",
|
||||
"includePath": [
|
||||
"${workspaceFolder}/**",
|
||||
"${workspaceFolder}/../../../Common"
|
||||
],
|
||||
"defines": [],
|
||||
"compilerPath": "/usr/local/cuda/bin/nvcc",
|
||||
"cStandard": "gnu17",
|
||||
"cppStandard": "gnu++14",
|
||||
"intelliSenseMode": "linux-gcc-x64",
|
||||
"configurationProvider": "ms-vscode.makefile-tools"
|
||||
}
|
||||
],
|
||||
"version": 4
|
||||
}
|
||||
7
cpp/4_CUDA_Libraries/jitLto/.vscode/extensions.json
vendored
Normal file
7
cpp/4_CUDA_Libraries/jitLto/.vscode/extensions.json
vendored
Normal file
@@ -0,0 +1,7 @@
|
||||
{
|
||||
"recommendations": [
|
||||
"nvidia.nsight-vscode-edition",
|
||||
"ms-vscode.cpptools",
|
||||
"ms-vscode.makefile-tools"
|
||||
]
|
||||
}
|
||||
54
cpp/4_CUDA_Libraries/jitLto/CMakeLists.txt
Normal file
54
cpp/4_CUDA_Libraries/jitLto/CMakeLists.txt
Normal file
@@ -0,0 +1,54 @@
|
||||
cmake_minimum_required(VERSION 3.20)
|
||||
|
||||
list(APPEND CMAKE_MODULE_PATH "${CMAKE_CURRENT_SOURCE_DIR}/../../../cmake/Modules")
|
||||
|
||||
project(jitLto LANGUAGES CXX)
|
||||
|
||||
find_package(CUDAToolkit REQUIRED)
|
||||
|
||||
set(CMAKE_POSITION_INDEPENDENT_CODE ON)
|
||||
|
||||
set(CMAKE_CUDA_ARCHITECTURES 75 80 86 87 89 90 100 110 120)
|
||||
set(CMAKE_CUDA_FLAGS "${CMAKE_CUDA_FLAGS} -Wno-deprecated-gpu-targets")
|
||||
if(ENABLE_CUDA_DEBUG)
|
||||
set(CMAKE_CUDA_FLAGS "${CMAKE_CUDA_FLAGS} -G") # enable cuda-gdb (may significantly affect performance on some targets)
|
||||
else()
|
||||
set(CMAKE_CUDA_FLAGS "${CMAKE_CUDA_FLAGS} -lineinfo") # add line information to all builds for debug tools (exclusive to -G option)
|
||||
endif()
|
||||
|
||||
# Include directories and libraries
|
||||
include_directories(../../../Common)
|
||||
|
||||
# This sample is not supported on QNX
|
||||
if(CMAKE_SYSTEM_NAME STREQUAL "QNX")
|
||||
message(STATUS "Will not build sample ${PROJECT_NAME} - not supported on QNX")
|
||||
return()
|
||||
endif()
|
||||
|
||||
# Source file
|
||||
# Add target for jitLto
|
||||
add_executable(jitLto jitLto.cpp)
|
||||
|
||||
target_compile_options(jitLto PRIVATE $<$<COMPILE_LANGUAGE:CUDA>:--extended-lambda>)
|
||||
|
||||
target_compile_features(jitLto PRIVATE cxx_std_17 cuda_std_17)
|
||||
|
||||
set_target_properties(jitLto PROPERTIES CUDA_SEPARABLE_COMPILATION ON)
|
||||
|
||||
find_library(NVJITLINK_LIB
|
||||
NAME nvJitLink
|
||||
PATHS "${CUDAToolkit_LIBRARY_DIR}" "${CMAKE_LIBRARY_PATH}")
|
||||
|
||||
target_include_directories(jitLto PRIVATE
|
||||
${CUDAToolkit_INCLUDE_DIRS}
|
||||
)
|
||||
|
||||
target_link_libraries(jitLto PRIVATE
|
||||
CUDA::cuda_driver
|
||||
CUDA::nvrtc
|
||||
${NVJITLINK_LIB}
|
||||
)
|
||||
|
||||
# Include installation configuration
|
||||
include(${CMAKE_CURRENT_SOURCE_DIR}/../../../cmake/InstallSamples.cmake)
|
||||
setup_samples_install()
|
||||
36
cpp/4_CUDA_Libraries/jitLto/README.md
Normal file
36
cpp/4_CUDA_Libraries/jitLto/README.md
Normal file
@@ -0,0 +1,36 @@
|
||||
# jitLto - Saxpy with libnvJitLink
|
||||
|
||||
## Description
|
||||
|
||||
This sample does a simple saxpy multiply and add using nvrtc and nvJitLink with LTO (Link Time Optimization). It has been written for clarity of exposition to illustrate various CUDA programming principles, not with the goal of providing the most performant generic kernel for saxpy.
|
||||
|
||||
## Key Concepts
|
||||
|
||||
CUDA Runtime API, Runtime Compilation
|
||||
|
||||
## Supported SM Architectures
|
||||
|
||||
[SM 5.0 ](https://developer.nvidia.com/cuda-gpus) [SM 5.2 ](https://developer.nvidia.com/cuda-gpus) [SM 5.3 ](https://developer.nvidia.com/cuda-gpus) [SM 6.0 ](https://developer.nvidia.com/cuda-gpus) [SM 6.1 ](https://developer.nvidia.com/cuda-gpus) [SM 7.0 ](https://developer.nvidia.com/cuda-gpus) [SM 7.2 ](https://developer.nvidia.com/cuda-gpus) [SM 7.5 ](https://developer.nvidia.com/cuda-gpus) [SM 8.0 ](https://developer.nvidia.com/cuda-gpus) [SM 8.6 ](https://developer.nvidia.com/cuda-gpus) [SM 8.7 ](https://developer.nvidia.com/cuda-gpus) [SM 8.9 ](https://developer.nvidia.com/cuda-gpus) [SM 9.0 ](https://developer.nvidia.com/cuda-gpus)
|
||||
|
||||
## Supported OSes
|
||||
|
||||
Linux, Windows
|
||||
|
||||
## Supported CPU Architecture
|
||||
|
||||
x86_64, aarch64
|
||||
|
||||
## CUDA APIs involved
|
||||
|
||||
### [CUDA Driver API](http://docs.nvidia.com/cuda/cuda-driver-api/index.html)
|
||||
cuModuleLoad, cuModuleLoadDataEx, cuModuleGetFunction, cuMemAlloc, cuMemFree, cuMemcpyHtoD, cuMemcpyDtoH, cuLaunchKernel
|
||||
|
||||
## Dependencies needed to build/run
|
||||
[NVRTC](../../../README.md#nvrtc), [NVJITLINK](../../../README.md#nvjitlink)
|
||||
|
||||
## Prerequisites
|
||||
|
||||
Download and install the [CUDA Toolkit](https://developer.nvidia.com/cuda-downloads) for your corresponding platform.
|
||||
Make sure the dependencies mentioned in [Dependencies]() section above are installed.
|
||||
|
||||
## References (for more details)
|
||||
260
cpp/4_CUDA_Libraries/jitLto/jitLto.cpp
Normal file
260
cpp/4_CUDA_Libraries/jitLto/jitLto.cpp
Normal file
@@ -0,0 +1,260 @@
|
||||
/* Copyright (c) 2022, NVIDIA CORPORATION. All rights reserved.
|
||||
*
|
||||
* Redistribution and use in source and binary forms, with or without
|
||||
* modification, are permitted provided that the following conditions
|
||||
* are met:
|
||||
* * Redistributions of source code must retain the above copyright
|
||||
* notice, this list of conditions and the following disclaimer.
|
||||
* * Redistributions in binary form must reproduce the above copyright
|
||||
* notice, this list of conditions and the following disclaimer in the
|
||||
* documentation and/or other materials provided with the distribution.
|
||||
* * Neither the name of NVIDIA CORPORATION nor the names of its
|
||||
* contributors may be used to endorse or promote products derived
|
||||
* from this software without specific prior written permission.
|
||||
*
|
||||
* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS ``AS IS'' AND ANY
|
||||
* EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
|
||||
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
|
||||
* PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR
|
||||
* CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
|
||||
* EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
|
||||
* PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
|
||||
* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY
|
||||
* OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
|
||||
* (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
|
||||
* OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
|
||||
*/
|
||||
|
||||
#include <cstring>
|
||||
#include <cuda.h>
|
||||
#include <iostream>
|
||||
#include <nvJitLink.h>
|
||||
#include <nvrtc.h>
|
||||
|
||||
#define NUM_THREADS 128
|
||||
#define NUM_BLOCKS 32
|
||||
|
||||
#define NVRTC_SAFE_CALL(x) \
|
||||
do { \
|
||||
nvrtcResult result = x; \
|
||||
if (result != NVRTC_SUCCESS) { \
|
||||
std::cerr << "\nerror: " #x " failed with error " << nvrtcGetErrorString(result) << '\n'; \
|
||||
exit(1); \
|
||||
} \
|
||||
} while (0)
|
||||
#define CUDA_SAFE_CALL(x) \
|
||||
do { \
|
||||
CUresult result = x; \
|
||||
if (result != CUDA_SUCCESS) { \
|
||||
const char *msg; \
|
||||
cuGetErrorName(result, &msg); \
|
||||
std::cerr << "\nerror: " #x " failed with error " << msg << '\n'; \
|
||||
exit(1); \
|
||||
} \
|
||||
} while (0)
|
||||
#define NVJITLINK_SAFE_CALL(h, x) \
|
||||
do { \
|
||||
nvJitLinkResult result = x; \
|
||||
if (result != NVJITLINK_SUCCESS) { \
|
||||
std::cerr << "\nerror: " #x " failed with error " << result << '\n'; \
|
||||
size_t lsize; \
|
||||
result = nvJitLinkGetErrorLogSize(h, &lsize); \
|
||||
if (result == NVJITLINK_SUCCESS && lsize > 0) { \
|
||||
char *log = (char *)malloc(lsize); \
|
||||
result = nvJitLinkGetErrorLog(h, log); \
|
||||
if (result == NVJITLINK_SUCCESS) { \
|
||||
std::cerr << "error log: " << log << '\n'; \
|
||||
free(log); \
|
||||
} \
|
||||
} \
|
||||
exit(1); \
|
||||
} \
|
||||
} while (0)
|
||||
|
||||
const char *lto_saxpy = " \n\
|
||||
extern __device__ float compute(float a, float x, float y); \n\
|
||||
\n\
|
||||
extern \"C\" __global__ \n\
|
||||
void saxpy(float a, float *x, float *y, float *out, size_t n) \n\
|
||||
{ \n\
|
||||
size_t tid = blockIdx.x * blockDim.x + threadIdx.x; \n\
|
||||
if (tid < n) { \n\
|
||||
out[tid] = compute(a, x[tid], y[tid]); \n\
|
||||
} \n\
|
||||
} \n";
|
||||
|
||||
const char *lto_compute = " \n\
|
||||
__device__ float compute(float a, float x, float y) { \n\
|
||||
return a * x + y; \n\
|
||||
} \n";
|
||||
|
||||
// compile code into LTOIR, returning the IR and its size
|
||||
static void getLTOIR(const char *code, const char *name, char **ltoIR, size_t *ltoIRSize)
|
||||
{
|
||||
// Create an instance of nvrtcProgram with the code string.
|
||||
nvrtcProgram prog;
|
||||
NVRTC_SAFE_CALL(nvrtcCreateProgram(&prog, // prog
|
||||
code, // buffer
|
||||
name, // name
|
||||
0, // numHeaders
|
||||
NULL, // headers
|
||||
NULL)); // includeNames
|
||||
|
||||
// specify that LTO IR should be generated for LTO operation
|
||||
const char *opts[] = {"-dlto", "--relocatable-device-code=true"};
|
||||
nvrtcResult compileResult = nvrtcCompileProgram(prog, // prog
|
||||
2, // numOptions
|
||||
opts); // options
|
||||
// Obtain compilation log from the program.
|
||||
size_t logSize;
|
||||
NVRTC_SAFE_CALL(nvrtcGetProgramLogSize(prog, &logSize));
|
||||
char *log = new char[logSize];
|
||||
NVRTC_SAFE_CALL(nvrtcGetProgramLog(prog, log));
|
||||
std::cout << log << '\n';
|
||||
delete[] log;
|
||||
if (compileResult != NVRTC_SUCCESS) {
|
||||
exit(1);
|
||||
}
|
||||
// Obtain generated LTO IR from the program.
|
||||
NVRTC_SAFE_CALL(nvrtcGetLTOIRSize(prog, ltoIRSize));
|
||||
*ltoIR = new char[*ltoIRSize];
|
||||
NVRTC_SAFE_CALL(nvrtcGetLTOIR(prog, *ltoIR));
|
||||
// Destroy the program.
|
||||
NVRTC_SAFE_CALL(nvrtcDestroyProgram(&prog));
|
||||
}
|
||||
|
||||
int main(int argc, char *argv[])
|
||||
{
|
||||
unsigned int cuda_major = 0;
|
||||
unsigned int cuda_minor = 0;
|
||||
nvJitLinkResult res = nvJitLinkVersion(&cuda_major, &cuda_minor);
|
||||
if (res != NVJITLINK_SUCCESS) {
|
||||
std::cerr << "Version check failed" << '\n';
|
||||
}
|
||||
else {
|
||||
std::cout << "CUDA " << cuda_major << "." << cuda_minor << '\n';
|
||||
}
|
||||
|
||||
|
||||
char *ltoIR1;
|
||||
char *ltoIR2;
|
||||
size_t ltoIR1Size;
|
||||
size_t ltoIR2Size;
|
||||
// getLTOIR uses nvrtc to get the LTOIR.
|
||||
// We could also use nvcc offline with -dlto -fatbin
|
||||
// to generate the IR, but using nvrtc keeps the build simpler.
|
||||
getLTOIR(lto_saxpy, "lto_saxpy.cu", <oIR1, <oIR1Size);
|
||||
getLTOIR(lto_compute, "lto_compute.cu", <oIR2, <oIR2Size);
|
||||
|
||||
CUdevice cuDevice;
|
||||
CUcontext context;
|
||||
CUmodule module;
|
||||
CUfunction kernel;
|
||||
CUctxCreateParams ctxCreateParams = {};
|
||||
CUDA_SAFE_CALL(cuInit(0));
|
||||
CUDA_SAFE_CALL(cuDeviceGet(&cuDevice, 0));
|
||||
CUDA_SAFE_CALL(cuCtxCreate(&context, &ctxCreateParams, 0, cuDevice));
|
||||
|
||||
// Dynamically determine the arch to link for
|
||||
int major = 0;
|
||||
int minor = 0;
|
||||
CUDA_SAFE_CALL(cuDeviceGetAttribute(&major, CU_DEVICE_ATTRIBUTE_COMPUTE_CAPABILITY_MAJOR, cuDevice));
|
||||
CUDA_SAFE_CALL(cuDeviceGetAttribute(&minor, CU_DEVICE_ATTRIBUTE_COMPUTE_CAPABILITY_MINOR, cuDevice));
|
||||
int arch = major * 10 + minor;
|
||||
char smbuf[16];
|
||||
memset(smbuf, 0, 16);
|
||||
sprintf(smbuf, "-arch=sm_%d", arch);
|
||||
|
||||
// Load the generated LTO IR and link them together
|
||||
nvJitLinkHandle handle;
|
||||
const char *lopts[] = {"-lto", smbuf};
|
||||
NVJITLINK_SAFE_CALL(handle, nvJitLinkCreate(&handle, 2, lopts));
|
||||
|
||||
NVJITLINK_SAFE_CALL(handle,
|
||||
nvJitLinkAddData(handle, NVJITLINK_INPUT_LTOIR, (void *)ltoIR1, ltoIR1Size, "lto_saxpy"));
|
||||
NVJITLINK_SAFE_CALL(handle,
|
||||
nvJitLinkAddData(handle, NVJITLINK_INPUT_LTOIR, (void *)ltoIR2, ltoIR2Size, "lto_compute"));
|
||||
|
||||
// The call to nvJitLinkComplete causes linker to link together the two
|
||||
// LTO IR modules, do optimization on the linked LTO IR,
|
||||
// and generate cubin from it.
|
||||
NVJITLINK_SAFE_CALL(handle, nvJitLinkComplete(handle));
|
||||
|
||||
// check error log
|
||||
size_t logSize;
|
||||
NVJITLINK_SAFE_CALL(handle, nvJitLinkGetErrorLogSize(handle, &logSize));
|
||||
if (logSize > 0) {
|
||||
char *log = (char *)malloc(logSize + 1);
|
||||
NVJITLINK_SAFE_CALL(handle, nvJitLinkGetErrorLog(handle, log));
|
||||
std::cout << "Error log: " << log << std::endl;
|
||||
free(log);
|
||||
}
|
||||
|
||||
// get linked cubin
|
||||
size_t cubinSize;
|
||||
NVJITLINK_SAFE_CALL(handle, nvJitLinkGetLinkedCubinSize(handle, &cubinSize));
|
||||
void *cubin = malloc(cubinSize);
|
||||
NVJITLINK_SAFE_CALL(handle, nvJitLinkGetLinkedCubin(handle, cubin));
|
||||
|
||||
NVJITLINK_SAFE_CALL(handle, nvJitLinkDestroy(&handle));
|
||||
delete[] ltoIR1;
|
||||
delete[] ltoIR2;
|
||||
|
||||
// cubin is linked, so now load it
|
||||
CUDA_SAFE_CALL(cuModuleLoadData(&module, cubin));
|
||||
CUDA_SAFE_CALL(cuModuleGetFunction(&kernel, module, "saxpy"));
|
||||
|
||||
// Generate input for execution, and create output buffers.
|
||||
size_t n = NUM_THREADS * NUM_BLOCKS;
|
||||
size_t bufferSize = n * sizeof(float);
|
||||
float a = 5.1f;
|
||||
float *hX = new float[n], *hY = new float[n], *hOut = new float[n];
|
||||
for (size_t i = 0; i < n; ++i) {
|
||||
hX[i] = static_cast<float>(i);
|
||||
hY[i] = static_cast<float>(i * 2);
|
||||
}
|
||||
CUdeviceptr dX, dY, dOut;
|
||||
CUDA_SAFE_CALL(cuMemAlloc(&dX, bufferSize));
|
||||
CUDA_SAFE_CALL(cuMemAlloc(&dY, bufferSize));
|
||||
CUDA_SAFE_CALL(cuMemAlloc(&dOut, bufferSize));
|
||||
CUDA_SAFE_CALL(cuMemcpyHtoD(dX, hX, bufferSize));
|
||||
CUDA_SAFE_CALL(cuMemcpyHtoD(dY, hY, bufferSize));
|
||||
// Execute SAXPY.
|
||||
void *args[] = {&a, &dX, &dY, &dOut, &n};
|
||||
CUDA_SAFE_CALL(cuLaunchKernel(kernel,
|
||||
NUM_BLOCKS,
|
||||
1,
|
||||
1, // grid dim
|
||||
NUM_THREADS,
|
||||
1,
|
||||
1, // block dim
|
||||
0,
|
||||
NULL, // shared mem and stream
|
||||
args,
|
||||
0)); // arguments
|
||||
CUDA_SAFE_CALL(cuCtxSynchronize());
|
||||
// Retrieve and print output.
|
||||
CUDA_SAFE_CALL(cuMemcpyDtoH(hOut, dOut, bufferSize));
|
||||
|
||||
for (size_t i = 0; i < n; ++i) {
|
||||
std::cout << a << " * " << hX[i] << " + " << hY[i] << " = " << hOut[i] << '\n';
|
||||
}
|
||||
// check last value to verify
|
||||
if (hOut[n - 1] == 29074.5) {
|
||||
std::cout << "PASSED!\n";
|
||||
}
|
||||
else {
|
||||
std::cout << "values not expected?\n";
|
||||
}
|
||||
// Release resources.
|
||||
CUDA_SAFE_CALL(cuMemFree(dX));
|
||||
CUDA_SAFE_CALL(cuMemFree(dY));
|
||||
CUDA_SAFE_CALL(cuMemFree(dOut));
|
||||
CUDA_SAFE_CALL(cuModuleUnload(module));
|
||||
CUDA_SAFE_CALL(cuCtxDestroy(context));
|
||||
free(cubin);
|
||||
delete[] hX;
|
||||
delete[] hY;
|
||||
delete[] hOut;
|
||||
return 0;
|
||||
}
|
||||
Reference in New Issue
Block a user