[PATCH] D83676: AMDGPUPerfHintAnalysis: don't treat constant loads as large stride

[Matt Arsenault via Phabricator via llvm-commits]

arsenm added inline comments.


================
Comment at: llvm/lib/Target/AMDGPU/AMDGPUPerfHintAnalysis.cpp:345-346
     unsigned As = PT->getAddressSpace();
     return As == AMDGPUAS::CONSTANT_ADDRESS ||
            As == AMDGPUAS::CONSTANT_ADDRESS_32BIT;
   }
----------------
This isn't really a good implementation. We use scalar loads in more cases and constant address space isn't a guarantee of SMEM loads


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D83676/new/

https://reviews.llvm.org/D83676