[Openmp-commits] [openmp] ca3545f - [Libomptarget] Bump up PTX version from +ptx61 to +ptx63 (#70227)

[via Openmp-commits]

Author: Joseph Huber
Date: 2023-10-25T13:28:02-04:00
New Revision: ca3545f0efe17d7cb3b80de1fdcae80e3ce75b0f

URL: https://github.com/llvm/llvm-project/commit/ca3545f0efe17d7cb3b80de1fdcae80e3ce75b0f
DIFF: https://github.com/llvm/llvm-project/commit/ca3545f0efe17d7cb3b80de1fdcae80e3ce75b0f.diff

LOG: [Libomptarget] Bump up PTX version from +ptx61 to +ptx63 (#70227)

Summary:
This version is required to support the 'activemask' feature which is
used for certain features, such as reductions. This ties the
implementation of the DeviceRTL roughly to the features provided by the
CUDA 9.0 release, which should be sufficienly old as to not cause
problems since this is a minor version jump that corresponds to the
release of `sm_53`.

Added: 
    

Modified: 
    openmp/libomptarget/DeviceRTL/CMakeLists.txt

Removed: 
    


################################################################################
diff  --git a/openmp/libomptarget/DeviceRTL/CMakeLists.txt b/openmp/libomptarget/DeviceRTL/CMakeLists.txt
index f71bdeae3d7f091..df8e4a5126fd443 100644
--- a/openmp/libomptarget/DeviceRTL/CMakeLists.txt
+++ b/openmp/libomptarget/DeviceRTL/CMakeLists.txt
@@ -232,7 +232,7 @@ function(compileDeviceRTLLibrary target_cpu target_name target_triple)
 
   set(target_feature "")
   if("${target_triple}" STREQUAL "nvptx64-nvidia-cuda")
-    set(target_feature "feature=+ptx61")
+    set(target_feature "feature=+ptx63")
   endif()
 
   # Package the bitcode in the bitcode and embed it in an ELF for the static library