caio/clang-p2996
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

[Offload] Rework handling for loading vendor runtimes (#93073)

Browse Source 
Summary:
We previously had multiple options for this, this patch replaces them
with `LIBOMPTARGET_DLOPEN_PLUGINS=` to be a list of plugins to
dynamically use. It defaults to everything right now. This ignores the
`host` plugin because the `libffi` dependency is going to be removed
soon hopefully in https://github.com/llvm/llvm-project/pull/91264.
This commit is contained in:
Joseph Huber
2024-05-22 13:04:52 -05:00
committed by GitHub
parent 96378b3da8
commit c618ae1734
5 changed files with 15 additions and 12 deletions
Download Patch File Download Diff File Expand all files Collapse all files

3
offload/CMakeLists.txt
View File

@@ -146,6 +146,9 @@ endif()
message(STATUS "Building the offload library with support for "
"the \"${LIBOMPTARGET_PLUGINS_TO_BUILD}\" plugins")
set(LIBOMPTARGET_DLOPEN_PLUGINS "${LIBOMPTARGET_PLUGINS_TO_BUILD}" CACHE STRING
"Semicolon-separated list of plugins to use 'dlopen' for runtime linking")
set(LIBOMPTARGET_ENUM_PLUGIN_TARGETS "")
foreach(plugin IN LISTS LIBOMPTARGET_PLUGINS_TO_BUILD)
set(LIBOMPTARGET_ENUM_PLUGIN_TARGETS

8
offload/cmake/Modules/LibomptargetGetDependencies.cmake
View File

@@ -3,7 +3,6 @@
#
# libffi : required to launch target kernels given function and argument
# pointers.
# CUDA : required to control offloading to NVIDIA GPUs.
include (FindPackageHandleStandardArgs)
@@ -43,13 +42,6 @@ endif()
find_package(FFI QUIET)
set(LIBOMPTARGET_DEP_LIBFFI_FOUND ${FFI_FOUND})
################################################################################
# Looking for CUDA...
################################################################################
find_package(CUDAToolkit QUIET)
set(LIBOMPTARGET_DEP_CUDA_FOUND ${CUDAToolkit_FOUND})
################################################################################
# Looking for NVIDIA GPUs...
################################################################################

3
offload/plugins-nextgen/amdgpu/CMakeLists.txt
View File

@@ -13,8 +13,7 @@ target_sources(omptarget.rtl.amdgpu PRIVATE src/rtl.cpp)
target_include_directories(omptarget.rtl.amdgpu PRIVATE
${CMAKE_CURRENT_SOURCE_DIR}/utils)
option(LIBOMPTARGET_FORCE_DLOPEN_LIBHSA "Build with dlopened libhsa" ON)
if(hsa-runtime64_FOUND AND NOT LIBOMPTARGET_FORCE_DLOPEN_LIBHSA)
if(hsa-runtime64_FOUND AND NOT "amdgpu" IN_LIST LIBOMPTARGET_DLOPEN_PLUGINS)
message(STATUS "Building AMDGPU plugin linked against libhsa")
target_link_libraries(omptarget.rtl.amdgpu PRIVATE hsa-runtime64::hsa-runtime64)
else()

4
offload/plugins-nextgen/cuda/CMakeLists.txt
View File

@@ -10,8 +10,8 @@ add_target_library(omptarget.rtl.cuda CUDA)
target_sources(omptarget.rtl.cuda PRIVATE src/rtl.cpp)
option(LIBOMPTARGET_FORCE_DLOPEN_LIBCUDA "Build with dlopened libcuda" ON)
if(LIBOMPTARGET_DEP_CUDA_FOUND AND NOT LIBOMPTARGET_FORCE_DLOPEN_LIBCUDA)
find_package(CUDAToolkit QUIET)
if(CUDAToolkit_FOUND AND NOT "cuda" IN_LIST LIBOMPTARGET_DLOPEN_PLUGINS)
message(STATUS "Building CUDA plugin linked against libcuda")
target_link_libraries(omptarget.rtl.cuda PRIVATE CUDA::cuda_driver)
else()

9
openmp/docs/SupportAndFAQ.rst
View File

@@ -454,6 +454,15 @@ Q: What command line options can I use for OpenMP?
We recommend taking a look at the OpenMP
:doc:`command line argument reference <CommandLineArgumentReference>` page.
Q: Can I build the offloading runtimes without CUDA or HSA?
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
By default, the offloading runtime will load the associated vendor runtime
during initialization rather than directly linking against them. This allows the
program to be built and run on many machine. If you wish to directly link
against these libraries, use the ``LIBOMPTARGET_DLOPEN_PLUGINS=""`` option to
suppress it for each plugin. The default value is every plugin enabled with
``LIBOMPTARGET_PLUGINS_TO_BUILD``.
Q: Why is my build taking a long time?
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
When installing OpenMP and other LLVM components, the build time on multicore