[all-commits] [llvm/llvm-project] 059050: [InstCombine][test] add tests for fadd reductions; ...

[RotateRight via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: 059050226532d633199f33c5df19b7e9eeccbf22
      https://github.com/llvm/llvm-project/commit/059050226532d633199f33c5df19b7e9eeccbf22
  Author: Sanjay Patel <spatel at rotateright.com>
  Date:   2021-07-18 (Sun, 18 Jul 2021)

  Changed paths:
    M llvm/test/Transforms/InstCombine/fadd.ll

  Log Message:
  -----------
  [InstCombine][test] add tests for fadd reductions; NFC


  Commit: 0e15de2d0c6976acecb25b90e1e27b278404b0cf
      https://github.com/llvm/llvm-project/commit/0e15de2d0c6976acecb25b90e1e27b278404b0cf
  Author: Sanjay Patel <spatel at rotateright.com>
  Date:   2021-07-18 (Sun, 18 Jul 2021)

  Changed paths:
    M llvm/lib/Transforms/InstCombine/InstCombineAddSub.cpp
    M llvm/test/Transforms/InstCombine/fadd.ll
    M llvm/test/Transforms/PhaseOrdering/X86/vector-reductions-expanded.ll

  Log Message:
  -----------
  [InstCombine] fold reassociative FP add into start value of fadd reduction

This pattern is visible in unrolled and vectorized loops.
Although the backend seems to be able to reassociate to
ideal form in the examples I looked at, we might as well
do that in IR for efficiency.


Compare: https://github.com/llvm/llvm-project/compare/b5a6ad8c893a...0e15de2d0c69