mirrors/cmake
2
0
Fork 0
mirror of https://gitlab.kitware.com/cmake/cmake.git synced 2026-09-25 04:09:36 +03:00
Code Activity

FindCUDAToolkit: Detect target processor even when not cross-compiling

Browse Source
This commit is contained in:
Robert Maynard
2026-06-23 14:31:18 -04:00
committed by Brad King
parent a733b2c1ce
commit afb9e1eab8
1 changed files with 18 additions and 12 deletions
Download Patch File Download Diff File Expand all files Collapse all files
Modules/FindCUDAToolkit.cmake
+18 -12
View File
@@ -1053,19 +1053,24 @@ else()
endif()
endif()
# Figure out the target directory when either crosscompiling
# Figure out the target processor which is required to determine
# our target dir ( linux ) or our lib dir ( windows )
#
# When a language is enabled we can use its compiler's target architecture.
if(CMAKE_CUDA_COMPILER_LOADED AND CMAKE_CUDA_COMPILER_ARCHITECTURE_ID)
set(_CUDA_TARGET_PROCESSOR "${CMAKE_CUDA_COMPILER_ARCHITECTURE_ID}")
elseif(CMAKE_CXX_COMPILER_LOADED AND CMAKE_CXX_COMPILER_ARCHITECTURE_ID)
set(_CUDA_TARGET_PROCESSOR "${CMAKE_CXX_COMPILER_ARCHITECTURE_ID}")
elseif(CMAKE_C_COMPILER_LOADED AND CMAKE_C_COMPILER_ARCHITECTURE_ID)
set(_CUDA_TARGET_PROCESSOR "${CMAKE_C_COMPILER_ARCHITECTURE_ID}")
elseif(CMAKE_SYSTEM_PROCESSOR)
set(_CUDA_TARGET_PROCESSOR "${CMAKE_SYSTEM_PROCESSOR}")
endif()
# Figure out our crosscompiling targets dir
# or if we don't have `nvcc` and need to deduce the target arch
if(CMAKE_CROSSCOMPILING OR NOT CUDAToolkit_NVCC_EXECUTABLE)
# When a language is enabled we can use its compiler's target architecture.
if(CMAKE_CUDA_COMPILER_LOADED AND CMAKE_CUDA_COMPILER_ARCHITECTURE_ID)
set(_CUDA_TARGET_PROCESSOR "${CMAKE_CUDA_COMPILER_ARCHITECTURE_ID}")
elseif(CMAKE_CXX_COMPILER_LOADED AND CMAKE_CXX_COMPILER_ARCHITECTURE_ID)
set(_CUDA_TARGET_PROCESSOR "${CMAKE_CXX_COMPILER_ARCHITECTURE_ID}")
elseif(CMAKE_C_COMPILER_LOADED AND CMAKE_C_COMPILER_ARCHITECTURE_ID)
set(_CUDA_TARGET_PROCESSOR "${CMAKE_C_COMPILER_ARCHITECTURE_ID}")
elseif(CMAKE_SYSTEM_PROCESSOR)
set(_CUDA_TARGET_PROCESSOR "${CMAKE_SYSTEM_PROCESSOR}")
elseif(CMAKE_CROSSCOMPILING)
if(CMAKE_CROSSCOMPILING AND NOT _CUDA_TARGET_PROCESSOR)
message(FATAL_ERROR "Cross-compiling with the CUDA toolkit requires CMAKE_SYSTEM_PROCESSOR to be set.")
endif()
# Keep in sync with equivalent table in CMakeDetermineCUDACompiler and FindCUDA!
@@ -1085,7 +1090,6 @@ if(CMAKE_CROSSCOMPILING OR NOT CUDAToolkit_NVCC_EXECUTABLE)
elseif(_CUDA_TARGET_PROCESSOR STREQUAL "x86_64")
set(CUDAToolkit_TARGET_NAMES "x86_64-linux")
endif()
unset(_CUDA_TARGET_PROCESSOR)
foreach(CUDAToolkit_TARGET_NAME IN LISTS CUDAToolkit_TARGET_NAMES)
if(EXISTS "${CUDAToolkit_ROOT_DIR}/targets/${CUDAToolkit_TARGET_NAME}")
@@ -1116,6 +1120,8 @@ if(CMAKE_HOST_SYSTEM_NAME STREQUAL "Windows")
endif()
endif()
unset(_CUDA_TARGET_PROCESSOR)
# We don't need to verify the cuda_runtime header when we are using `nvcc` include paths
# as the compiler being enabled means the header was found
if(NOT CUDAToolkit_INCLUDE_DIRECTORIES)