[PATCH] D145614: [AARCH64] Enable STORE of v4i8 to help more vectorization opportunities

[Guozhi Wei via Phabricator via llvm-commits]

Carrot created this revision.
Herald added subscribers: pengfei, hiraditya, kristof.beyls.
Herald added a project: All.
Carrot requested review of this revision.
Herald added subscribers: llvm-commits, pcwang-thead.
Herald added a project: LLVM.

For the attached test case, currently llvm generates instructions to load/or/store the bytes one by one. Although NEON doesn't support v4i8 natively, we can mix v8i8 alu instructions with 32 bit load/store instructions to handle them, just like x86 backend. So this patch does the same thing as in x86, enable custom STORE of v4i8, and finally I can get vectorized instructions.


Repository:
  rG LLVM Github Monorepo

https://reviews.llvm.org/D145614

Files:
  llvm/lib/Target/AArch64/AArch64ISelLowering.cpp
  llvm/test/Transforms/SLPVectorizer/AArch64/32-bit.ll
  llvm/test/Transforms/SLPVectorizer/AArch64/vectorizable-selects-uniform-cmps.ll

-------------- next part --------------
A non-text attachment was scrubbed...
Name: D145614.503505.patch
Type: text/x-patch
Size: 16828 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20230308/a9c6da59/attachment.bin>