[PATCH] D145170: [AMDGPU] Vectorize misaligned global loads & stores

[Matt Arsenault via Phabricator via llvm-commits]

arsenm added inline comments.


================
Comment at: llvm/lib/Target/AMDGPU/SIISelLowering.cpp:1551
+  // than multiple smaller memory ops -- even when misaligned
+  if (AddrSpace == AMDGPUAS::GLOBAL_ADDRESS ||
+      AddrSpace == AMDGPUAS::CONSTANT_ADDRESS ||
----------------
Should also cover "other" address space. We have isFlatGlobalAddrSpace but probably need one that excludes flat


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D145170/new/

https://reviews.llvm.org/D145170