NASP/cuda-samples
85
0
Fork 0
mirror of https://github.com/NVIDIA/cuda-samples.git synced 2026-10-25 22:28:17 +08:00
Code Issues Packages Projects Releases Wiki Activity

Changes for building for aarch64, specifically jetson orin nano

Browse Source
This commit is contained in:
Jonathan Bentz
2025-01-08 17:05:45 -06:00
parent 1e0c660f22
commit a1cf9e4183
15 changed files with 189 additions and 120 deletions
Download Patch File Download Diff File Expand all files Collapse all files

26
Samples/0_Introduction/simpleCUDA2GL/CMakeLists.txt
View File

@@ -20,25 +20,29 @@ find_package(GLUT)
# Source file
if(${OpenGL_FOUND})
if (${GLUT_FOUND})
# Add target for simpleCUDA2GL
add_executable(simpleCUDA2GL simpleCUDA2GL.cu main.cpp)
if(CMAKE_SYSTEM_PROCESSOR STREQUAL "aarch64")
message(STATUS "Will not build sample simpleCUDA2GL - not supported on aarch64")
else()
# Add target for simpleCUDA2GL
add_executable(simpleCUDA2GL simpleCUDA2GL.cu main.cpp)
target_compile_options(simpleCUDA2GL PRIVATE $<$<COMPILE_LANGUAGE:CUDA>:--extended-lambda>)
target_compile_options(simpleCUDA2GL PRIVATE $<$<COMPILE_LANGUAGE:CUDA>:--extended-lambda>)
target_compile_features(simpleCUDA2GL PRIVATE cxx_std_17 cuda_std_17)
target_compile_features(simpleCUDA2GL PRIVATE cxx_std_17 cuda_std_17)
set_target_properties(simpleCUDA2GL PROPERTIES CUDA_SEPARABLE_COMPILATION ON)
target_include_directories(simpleCUDA2GL PUBLIC
${OPENGL_INCLUDE_DIR}
${CUDAToolkit_INCLUDE_DIRS}
${GLUT_INCLUDE_DIRS}
)
set_target_properties(simpleCUDA2GL PROPERTIES CUDA_SEPARABLE_COMPILATION ON)
target_include_directories(simpleCUDA2GL PUBLIC
${OPENGL_INCLUDE_DIR}
${CUDAToolkit_INCLUDE_DIRS}
${GLUT_INCLUDE_DIRS}
)
target_link_libraries(simpleCUDA2GL
${OPENGL_LIBRARIES}
${GLUT_LIBRARIES}
)
endif()
else()
message(STATUS "GLUT not found - will not build sample 'simpleCUDA2GL'")
endif()

16
Samples/0_Introduction/simpleIPC/CMakeLists.txt
View File

@@ -14,12 +14,16 @@ endif()
# Include directories and libraries
include_directories(../../../Common)
# Source file
# Add target for simpleIPC
add_executable(simpleIPC simpleIPC.cu ../../../Common/helper_multiprocess.cpp)
if(CMAKE_SYSTEM_PROCESSOR STREQUAL "aarch64")
message(STATUS "Will not build sample simpleIPC - not supported on aarch64")
else()
# Source file
# Add target for simpleIPC
add_executable(simpleIPC simpleIPC.cu ../../../Common/helper_multiprocess.cpp)
target_compile_options(simpleIPC PRIVATE $<$<COMPILE_LANGUAGE:CUDA>:--extended-lambda>)
target_compile_options(simpleIPC PRIVATE $<$<COMPILE_LANGUAGE:CUDA>:--extended-lambda>)
target_compile_features(simpleIPC PRIVATE cxx_std_17 cuda_std_17)
target_compile_features(simpleIPC PRIVATE cxx_std_17 cuda_std_17)
set_target_properties(simpleIPC PROPERTIES CUDA_SEPARABLE_COMPILATION ON)
set_target_properties(simpleIPC PROPERTIES CUDA_SEPARABLE_COMPILATION ON)
endif()

12
Samples/0_Introduction/simpleP2P/CMakeLists.txt
View File

@@ -16,10 +16,14 @@ include_directories(../../../Common)
# Source file
# Add target for simpleP2P
add_executable(simpleP2P simpleP2P.cu)
if(CMAKE_SYSTEM_PROCESSOR STREQUAL "aarch64")
message(STATUS "Will not build sample simpleP2P - not supported on aarch64")
else()
add_executable(simpleP2P simpleP2P.cu)
target_compile_options(simpleP2P PRIVATE $<$<COMPILE_LANGUAGE:CUDA>:--extended-lambda>)
target_compile_options(simpleP2P PRIVATE $<$<COMPILE_LANGUAGE:CUDA>:--extended-lambda>)
target_compile_features(simpleP2P PRIVATE cxx_std_17 cuda_std_17)
target_compile_features(simpleP2P PRIVATE cxx_std_17 cuda_std_17)
set_target_properties(simpleP2P PROPERTIES CUDA_SEPARABLE_COMPILATION ON)
set_target_properties(simpleP2P PROPERTIES CUDA_SEPARABLE_COMPILATION ON)
endif()

28
Samples/0_Introduction/systemWideAtomics/CMakeLists.txt
View File

@@ -14,17 +14,21 @@ endif()
# Include directories and libraries
include_directories(../../../Common)
if(CMAKE_SYSTEM_NAME STREQUAL "Linux")
# Source file
# Add target for systemWideAtomics
add_executable(systemWideAtomics systemWideAtomics.cu)
target_compile_options(systemWideAtomics PRIVATE $<$<COMPILE_LANGUAGE:CUDA>:--extended-lambda>)
target_compile_features(systemWideAtomics PRIVATE cxx_std_17 cuda_std_17)
set_target_properties(systemWideAtomics PROPERTIES CUDA_SEPARABLE_COMPILATION ON)
if(CMAKE_SYSTEM_PROCESSOR STREQUAL "aarch64")
message(STATUS "Will not build sample systemWideAtomics - not supported on aarch64")
else()
message(STATUS "Will not build sample systemWideAtomics - requires Linux OS")
if(CMAKE_SYSTEM_NAME STREQUAL "Linux")
# Source file
# Add target for systemWideAtomics
add_executable(systemWideAtomics systemWideAtomics.cu)
target_compile_options(systemWideAtomics PRIVATE $<$<COMPILE_LANGUAGE:CUDA>:--extended-lambda>)
target_compile_features(systemWideAtomics PRIVATE cxx_std_17 cuda_std_17)
set_target_properties(systemWideAtomics PROPERTIES CUDA_SEPARABLE_COMPILATION ON)
else()
message(STATUS "Will not build sample systemWideAtomics - requires Linux OS")
endif()
endif()

48
Samples/0_Introduction/vectorAddMMAP/CMakeLists.txt
View File

@@ -16,33 +16,37 @@ include_directories(../../../Common)
# Source file
# Add target for vectorAddMMAP
add_executable(vectorAddMMAP vectorAddMMAP.cpp multidevicealloc_memmap.cpp)
if(CMAKE_SYSTEM_PROCESSOR STREQUAL "aarch64")
message(STATUS "Will not build sample vectorAddMMAP - not supported on aarch64")
else()
add_executable(vectorAddMMAP vectorAddMMAP.cpp multidevicealloc_memmap.cpp)
target_compile_options(vectorAddMMAP PRIVATE $<$<COMPILE_LANGUAGE:CUDA>:--extended-lambda>)
target_compile_options(vectorAddMMAP PRIVATE $<$<COMPILE_LANGUAGE:CUDA>:--extended-lambda>)
target_compile_features(vectorAddMMAP PRIVATE cxx_std_17 cuda_std_17)
target_compile_features(vectorAddMMAP PRIVATE cxx_std_17 cuda_std_17)
set_target_properties(vectorAddMMAP PROPERTIES CUDA_SEPARABLE_COMPILATION ON)
target_include_directories(vectorAddMMAP PRIVATE
${CUDAToolkit_INCLUDE_DIRS}
)
set_target_properties(vectorAddMMAP PROPERTIES CUDA_SEPARABLE_COMPILATION ON)
target_include_directories(vectorAddMMAP PRIVATE
${CUDAToolkit_INCLUDE_DIRS}
)
target_link_libraries(vectorAddMMAP PUBLIC
CUDA::cuda_driver
)
target_link_libraries(vectorAddMMAP PUBLIC
CUDA::cuda_driver
)
set(CUDA_FATBIN_FILE "${CMAKE_CURRENT_BINARY_DIR}/vectorAdd_kernel64.fatbin")
set(CUDA_KERNEL_SOURCE "${CMAKE_CURRENT_SOURCE_DIR}/vectorAdd_kernel.cu")
set(CUDA_FATBIN_FILE "${CMAKE_CURRENT_BINARY_DIR}/vectorAdd_kernel64.fatbin")
set(CUDA_KERNEL_SOURCE "${CMAKE_CURRENT_SOURCE_DIR}/vectorAdd_kernel.cu")
add_custom_command(
OUTPUT ${CUDA_FATBIN_FILE}
COMMAND ${CMAKE_CUDA_COMPILER} ${INCLUDES} ${ALL_CCFLAGS} ${GENCODE_FLAGS} -o ${CUDA_FATBIN_FILE} -fatbin ${CUDA_KERNEL_SOURCE}
DEPENDS ${CUDA_KERNEL_SOURCE}
COMMENT "Building CUDA fatbin: ${CUDA_FATBIN_FILE}"
)
add_custom_command(
OUTPUT ${CUDA_FATBIN_FILE}
COMMAND ${CMAKE_CUDA_COMPILER} ${INCLUDES} ${ALL_CCFLAGS} ${GENCODE_FLAGS} -o ${CUDA_FATBIN_FILE} -fatbin ${CUDA_KERNEL_SOURCE}
DEPENDS ${CUDA_KERNEL_SOURCE}
COMMENT "Building CUDA fatbin: ${CUDA_FATBIN_FILE}"
)
# Create a dummy target for fatbin generation
add_custom_target(generate_fatbin_vectorAddMMAP ALL DEPENDS ${CUDA_FATBIN_FILE})
# Create a dummy target for fatbin generation
add_custom_target(generate_fatbin_vectorAddMMAP ALL DEPENDS ${CUDA_FATBIN_FILE})
# Ensure matrixMulDrv depends on the fatbin
add_dependencies(vectorAddMMAP generate_fatbin_vectorAddMMAP)
# Ensure matrixMulDrv depends on the fatbin
add_dependencies(vectorAddMMAP generate_fatbin_vectorAddMMAP)
endif()