[PATCH] D143731: [AMDGPU] Break-up large PHIs for DAGISel

Matt Arsenault via Phabricator via llvm-commits llvm-commits at lists.llvm.org
Tue Feb 14 05:23:51 PST 2023


arsenm added inline comments.


================
Comment at: llvm/lib/Target/AMDGPU/AMDGPUCodeGenPrepare.cpp:1415
+
+  return B.CreateExtractVector(SubVecTy, Val, B.getInt64(Idx));
+}
----------------
Pierre-vh wrote:
> arsenm wrote:
> > This must be a new intrinsic, I didn't know it existed. Usually you would do a shufflevector to get the reduced elements
> If shufflevector has better handling and gets folded out for sure then it might be simpler to use it.
> Do I just do a shufflevector with the same operand twice + the mask of the indexes I want?
You can use poison for the second operand 


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D143731/new/

https://reviews.llvm.org/D143731



More information about the llvm-commits mailing list