[Openmp-commits] [PATCH] D41724: [OpenMP][libomptarget] Enable the compilation of multiple bc libraries for runtime inlining

[Jonas Hahnfeld via Phabricator via Openmp-commits]

Hahnfeld requested changes to this revision.
Hahnfeld added inline comments.
This revision now requires changes to proceed.


================
Comment at: libomptarget/deviceRTLs/nvptx/CMakeLists.txt:63-64
   # SM_35 is what clang uses by default.
-  set(LIBOMPTARGET_NVPTX_COMPUTE_CAPABILITY 35 CACHE STRING
-    "CUDA Compute Capability to be used to compile the NVPTX device RTL.")
-  set(CUDA_ARCH -arch sm_${LIBOMPTARGET_NVPTX_COMPUTE_CAPABILITY})
----------------
This should definitely stay a `CACHE` variable. And I still propose to rename it to `_CAPABILITIES`...


================
Comment at: libomptarget/deviceRTLs/nvptx/CMakeLists.txt:200
+        # Install device RTL under the lib destination folder.
+        install(FILES ${CMAKE_CURRENT_BINARY_DIR}/libomptarget-nvptx-${sm}.bc DESTINATION "lib")
+      endforeach()
----------------
Do we want to use eg `35` as suffix or full `sm_35`?


Repository:
  rOMP OpenMP

https://reviews.llvm.org/D41724