[PATCH] D105709: [AMDGPU][GlobalISel] Insert an and with exec before s_cbranch_vccnz if necessary
Jay Foad via Phabricator via llvm-commits
llvm-commits at lists.llvm.org
Wed Jul 28 07:52:32 PDT 2021
foad added inline comments.
================
Comment at: llvm/lib/Target/AMDGPU/AMDGPUInstructionSelector.cpp:2485-2486
+ return IntrinsicID == Intrinsic::amdgcn_class ||
+ IntrinsicID == Intrinsic::amdgcn_icmp ||
+ IntrinsicID == Intrinsic::amdgcn_fcmp;
+ }
----------------
I don't think you should handle amdgcn_icmp and amdgcn_fcmp here. They are strange beasts that return the full 32/64-bit results of executing a divergent comparison in all lanes, and they should be deprecated in favour of amdgcn_ballot. Remove the corresponding test case as well.
CHANGES SINCE LAST ACTION
https://reviews.llvm.org/D105709/new/
https://reviews.llvm.org/D105709
More information about the llvm-commits
mailing list