[PATCH] D110950: [AMDGPU] Enable divergence-driven BFE selection

[Jay Foad via Phabricator via llvm-commits]

foad added a comment.

There are also patterns in SIInstructions.td that generate S_BFE without checking that the node is uniform.



================
Comment at: llvm/lib/Target/AMDGPU/AMDGPUISelDAGToDAG.cpp:272-275
   SDNode *getS_BFE(unsigned Opcode, const SDLoc &DL, SDValue Val,
                    uint32_t Offset, uint32_t Width);
+  SDNode *getV_BFE(unsigned Opcode, const SDLoc &DL, SDValue Val,
+                   uint32_t Offset, uint32_t Width);
----------------
How about replacing this with a single getBFE32 function, which takes "bool isSigned" and "bool isDivergent" arguments instead of "unsigned Opcode"?


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D110950/new/

https://reviews.llvm.org/D110950