[all-commits] [llvm/llvm-project] 1e1b60: [X86] Improve uniform funnelshift/rotation amount ...

[Simon Pilgrim via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: 1e1b60138c2b48b9212c4101de7b684f6652fdd5
      https://github.com/llvm/llvm-project/commit/1e1b60138c2b48b9212c4101de7b684f6652fdd5
  Author: Simon Pilgrim <llvm-dev at redking.me.uk>
  Date:   2022-02-12 (Sat, 12 Feb 2022)

  Changed paths:
    M llvm/lib/Target/X86/X86ISelLowering.cpp
    M llvm/test/CodeGen/X86/combine-rotates.ll
    M llvm/test/CodeGen/X86/min-legal-vector-width.ll
    M llvm/test/CodeGen/X86/vector-fshl-128.ll
    M llvm/test/CodeGen/X86/vector-fshl-256.ll
    M llvm/test/CodeGen/X86/vector-fshl-512.ll
    M llvm/test/CodeGen/X86/vector-fshl-rot-128.ll
    M llvm/test/CodeGen/X86/vector-fshl-rot-256.ll
    M llvm/test/CodeGen/X86/vector-fshl-rot-512.ll
    M llvm/test/CodeGen/X86/vector-fshl-rot-sub128.ll
    M llvm/test/CodeGen/X86/vector-fshr-128.ll
    M llvm/test/CodeGen/X86/vector-fshr-256.ll
    M llvm/test/CodeGen/X86/vector-fshr-512.ll
    M llvm/test/CodeGen/X86/vector-fshr-rot-128.ll
    M llvm/test/CodeGen/X86/vector-fshr-rot-256.ll
    M llvm/test/CodeGen/X86/vector-fshr-rot-512.ll
    M llvm/test/CodeGen/X86/vector-fshr-rot-sub128.ll
    M llvm/test/CodeGen/X86/vector-rotate-128.ll
    M llvm/test/CodeGen/X86/vector-rotate-256.ll
    M llvm/test/CodeGen/X86/vector-rotate-512.ll

  Log Message:
  -----------
  [X86] Improve uniform funnelshift/rotation amount handling

To find uniform shift/rotation amounts, we currently use SelectionDAG::getSplatValue which creates a node that extracts the scalar value from the source vector, this makes it more difficult for later combines to remove the extraction and stay on the SIMD unit, and can be a problem when the scalar type is illegal (i.e. i64 vs v2i64 on 32-bit targets).

This patch begins to use SelectionDAG::getSplatSourceVector (which SelectionDAG::getSplatValue uses internally) and adds a new variant of getTargetVShiftNode that takes the source vector and the splat index, and adjusts the vector in place to create the zero-extended value suitable for the SSE PSLL/PSRL/PSRA uniform instructions.

I'm still addressing a number of regressions when used for normal vector shifts, so I've just handled the funnelshift/rotation lowering for this first patch. I can then focus on the yak shaving (SimplifyDemandedBits/Elts in particular) necessary to always use SelectionDAG::getSplatSourceVector.

Differential Revision: https://reviews.llvm.org/D119090