[PATCH] D120736: [AMDGPU] Make more use of madmk/fmamk instructions

[Jay Foad via Phabricator via llvm-commits]

foad created this revision.
foad added reviewers: arsenm, rampitec, piotr.
Herald added subscribers: mstorsjo, kerbowa, hiraditya, t-tye, tpr, dstuttard, yaxunl, nhaehnle, jvesely, kzhuravl.
foad requested review of this revision.
Herald added subscribers: llvm-commits, wdng.
Herald added a project: LLVM.

In convertToThreeAddress handle VOP2 mac/fmac instructions with a
literal src0 operand, since these are prime candidates for
converting to madmk/fmamk.

Previously this would only happen if src0 (or src1) was a register
defined by a move-immediate instruction, but in many cases these
operands have already been folded because SIFoldOperands runs
before TwoAddressInstructionPass.


Repository:
  rG LLVM Github Monorepo

https://reviews.llvm.org/D120736

Files:
  llvm/lib/Target/AMDGPU/SIInstrInfo.cpp
  llvm/test/CodeGen/AMDGPU/GlobalISel/urem.i64.ll
  llvm/test/CodeGen/AMDGPU/twoaddr-fma.mir
  llvm/test/CodeGen/AMDGPU/twoaddr-mad.mir

-------------- next part --------------
A non-text attachment was scrubbed...
Name: D120736.412132.patch
Type: text/x-patch
Size: 22724 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20220301/9ad288d8/attachment.bin>