[PATCH] D86864: [MachineSinking] sink more profitable loads

ChenZheng via Phabricator via llvm-commits llvm-commits at lists.llvm.org
Mon Aug 31 02:11:08 PDT 2020


shchenz created this revision.
shchenz added reviewers: MatzeB, echristo, efriedma, qcolombet, PowerPC.
Herald added subscribers: llvm-commits, danielkiss, luismarques, apazos, sameer.abuasal, pzheng, s.egerton, lenary, Jim, jocewei, PkmX, arphaman, the_o, brucehoult, MartinMosbeck, rogfer01, edward-jones, zzheng, jrtc27, niosHD, sabuasal, simoncook, johnrusso, rbar, asb, hiraditya.
Herald added a project: LLVM.
shchenz requested review of this revision.
Herald added a subscriber: MaskRay.

Instead of setting Store to true blindly, handle some cases where Store can be determined by some analysis if `MI.parent()` dom `SuccToSinkTo` and `SuccToSinkTo` post dom `MI.parent()`.

Thus we can get more profitable loads to be sunk.


Repository:
  rG LLVM Github Monorepo

https://reviews.llvm.org/D86864

Files:
  llvm/lib/CodeGen/MachineSink.cpp
  llvm/test/CodeGen/RISCV/select-optimize-multiple.ll
  llvm/test/CodeGen/Thumb2/LowOverheadLoops/reductions.ll
  llvm/test/CodeGen/X86/2007-01-13-StackPtrIndex.ll
  llvm/test/CodeGen/X86/MachineSink-eflags.ll
  llvm/test/CodeGen/X86/avx2-masked-gather.ll
  llvm/test/CodeGen/X86/cmovcmov.ll
  llvm/test/CodeGen/X86/select.ll
  llvm/test/CodeGen/X86/vec_int_to_fp.ll

-------------- next part --------------
A non-text attachment was scrubbed...
Name: D86864.288903.patch
Type: text/x-patch
Size: 41715 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20200831/a3d6cb7f/attachment-0001.bin>


More information about the llvm-commits mailing list