[PATCH] D72865: [AMDGPU] allow multi-dword flat scratch access since GFX9

Matt Arsenault via Phabricator via llvm-commits llvm-commits at lists.llvm.org
Thu Jan 16 12:57:03 PST 2020


arsenm added a comment.

This also needs to be reproduce for globalisel



================
Comment at: llvm/lib/Target/AMDGPU/SIISelLowering.cpp:7385
+  if (AS == AMDGPUAS::FLAT_ADDRESS &&
+      getSubtarget()->getGeneration() < AMDGPUSubtarget::GFX9)
     AS = MFI->hasFlatScratchInit() ?
----------------
Should move this into a predicate function inside the subtarget


================
Comment at: llvm/test/CodeGen/AMDGPU/flat-address-space.ll:160
 }
 
 ; CHECK-LABEL: flat_scratch_multidword_store:
----------------
What is the alignment situation for this? I thought unaligned scratch access was introduced but we never started using it?


CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D72865/new/

https://reviews.llvm.org/D72865





More information about the llvm-commits mailing list