diff options
| author | Daniil Seredkin <vdsered@gmail.com> | 2021-06-07 07:01:29 -0400 |
|---|---|---|
| committer | Sanjay Patel <spatel@rotateright.com> | 2021-06-07 08:08:05 -0400 |
| commit | 7736c1936a93d7effecf32ef5193dfc00b732cd2 (patch) | |
| tree | 4ac5e80435c8637e9aee3d2d6eb9e8ebf9d3b84b /flang/lib/Frontend/CompilerInvocation.cpp | |
| parent | 2def12ebc6cc904cddb4bc608df63014ec2bfe86 (diff) | |
| download | llvm-7736c1936a93d7effecf32ef5193dfc00b732cd2.zip llvm-7736c1936a93d7effecf32ef5193dfc00b732cd2.tar.gz llvm-7736c1936a93d7effecf32ef5193dfc00b732cd2.tar.bz2 | |
[InstCombine] Missed optimization for pow(x, y) * pow(x, z) with fast-math
If FP reassociation (fast-math) is allowed, then LLVM is free to do the
following transformation pow(x, y) * pow(x, z) -> pow(x, y + z).
This patch adds this transformation and tests for it.
See more https://bugs.llvm.org/show_bug.cgi?id=47205
It handles two cases
1. When operands of fmul are different instructions
%4 = call reassoc float @llvm.pow.f32(float %0, float %1)
%5 = call reassoc float @llvm.pow.f32(float %0, float %2)
%6 = fmul reassoc float %5, %4
-->
%3 = fadd reassoc float %1, %2
%4 = call reassoc float @llvm.pow.f32(float %0, float %3)
2. When operands of fmul are the same instruction
%4 = call reassoc float @llvm.pow.f32(float %0, float %1)
%5 = fmul reassoc float %4, %4
-->
%3 = fadd reassoc float %1, %1
%4 = call reassoc float @llvm.pow.f32(float %0, float %3)
Differential Revision: https://reviews.llvm.org/D102574
Diffstat (limited to 'flang/lib/Frontend/CompilerInvocation.cpp')
0 files changed, 0 insertions, 0 deletions