[PATCH] D101095: [clang][amdgpu] Use implicit code object version

Jon Chesterfield via Phabricator via cfe-commits cfe-commits at lists.llvm.org
Thu Apr 22 11:26:17 PDT 2021


JonChesterfield added a comment.

Diff created relative to D101077 <https://reviews.llvm.org/D101077>, not relative to main. Iiuc that will work out fine if the NFC lands first.



================
Comment at: clang/test/Driver/hip-code-object-version.hip:55
 // RUN:   %s 2>&1 | FileCheck -check-prefix=VD %s
 
 // VD: "-targets=host-x86_64-unknown-linux,hipv4-amdgcn-amd-amdhsa--gfx906"
----------------
This is the interesting test change - we still use 'hipv4', but no longer pass amdhsa-code-object-version=4 explicitly.

The other tests don't depend on code object version so removing the regex from those seems a strict improvement.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D101095/new/

https://reviews.llvm.org/D101095



More information about the cfe-commits mailing list