[PATCH] D149463: [X86] LowerFunnelShift: prefer unpack-based algorithm

[Ivan Chikish via Phabricator via llvm-commits]

Nekotekina created this revision.
Nekotekina added reviewers: RKSimon, LuoYuanke, pengfei.
Nekotekina added a project: LLVM.
Herald added a subscriber: hiraditya.
Herald added a project: All.
Nekotekina requested review of this revision.

Continuation of the https://reviews.llvm.org/D149071

Using unpack for splitting and using double-bitwidth shifts may improve performance according to https://uica.uops.info tests.

1. No cross-lane shuffles
2. No dirtying double-width registers


Repository:
  rG LLVM Github Monorepo

https://reviews.llvm.org/D149463

Files:
  llvm/lib/Target/X86/X86ISelLowering.cpp
  llvm/test/CodeGen/X86/vector-fshl-128.ll
  llvm/test/CodeGen/X86/vector-fshl-256.ll
  llvm/test/CodeGen/X86/vector-fshl-512.ll
  llvm/test/CodeGen/X86/vector-fshr-128.ll
  llvm/test/CodeGen/X86/vector-fshr-256.ll
  llvm/test/CodeGen/X86/vector-fshr-512.ll

-------------- next part --------------
A non-text attachment was scrubbed...
Name: D149463.517965.patch
Type: text/x-patch
Size: 116554 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20230428/91b58993/attachment-0001.bin>