[all-commits] [llvm/llvm-project] 2e8733: [InstCombine] convert mul by negative-pow2 to nega...

[Sanjay Patel via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: 2e87333bfe7e226533fb66c84015c4523e352d3f
      https://github.com/llvm/llvm-project/commit/2e87333bfe7e226533fb66c84015c4523e352d3f
  Author: Sanjay Patel <spatel at rotateright.com>
  Date:   2022-10-02 (Sun, 02 Oct 2022)

  Changed paths:
    M llvm/lib/Transforms/InstCombine/InstCombineMulDivRem.cpp
    M llvm/test/Transforms/InstCombine/add.ll
    M llvm/test/Transforms/InstCombine/mul.ll

  Log Message:
  -----------
  [InstCombine] convert mul by negative-pow2 to negate and shift

This is an unusual canonicalization because we create an extra instruction,
but it's likely better for analysis and codegen (similar reasoning as D133399).

InstCombine::Negator may create this kind of multiply from negate and shift,
but this should not conflict because of the narrow negation.

I don't know how to create a fully general proof for this kind of transform in
Alive2, but here's an example with bitwidths similar to one of the regression
tests:
https://alive2.llvm.org/ce/z/J3jTjR

Differential Revision: https://reviews.llvm.org/D133667