[PATCH] D74345: AMDGPU: Don't report 2-byte alignment as fast

[Matt Arsenault via Phabricator via llvm-commits]

arsenm created this revision.
arsenm added reviewers: nhaehnle, rampitec.
Herald added subscribers: kerbowa, hiraditya, t-tye, tpr, dstuttard, yaxunl, wdng, jvesely, kzhuravl.
Herald added a project: LLVM.

This is apparently worse than 1-byte alignment. This does not attempt
to decompose 2-byte aligned wide stores, but will stop trying to
produce them.

      

Also fix bug in LoadStoreVectorizer which was decreasing the alignment
and vectorizing stack accesses. It was assuming a stack object was an
alloca that could have its base alignment changed, which is not true
if the pointer is derived from a function argument.


https://reviews.llvm.org/D74345

Files:
  llvm/lib/Target/AMDGPU/SIISelLowering.cpp
  llvm/lib/Transforms/Vectorize/LoadStoreVectorizer.cpp
  llvm/test/CodeGen/AMDGPU/chain-hi-to-lo.ll
  llvm/test/CodeGen/AMDGPU/fast-unaligned-load-store.global.ll
  llvm/test/CodeGen/AMDGPU/fast-unaligned-load-store.private.ll
  llvm/test/CodeGen/AMDGPU/unaligned-load-store.ll
  llvm/test/Transforms/LoadStoreVectorizer/AMDGPU/adjust-alloca-alignment.ll
  llvm/test/Transforms/LoadStoreVectorizer/AMDGPU/merge-stores-private.ll
  llvm/test/Transforms/LoadStoreVectorizer/AMDGPU/merge-stores.ll

-------------- next part --------------
A non-text attachment was scrubbed...
Name: D74345.243625.patch
Type: text/x-patch
Size: 34662 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20200210/b575b877/attachment.bin>