[PATCH] D71706: [SimplifyLibCalls] require fast-math-flags for pow(X, -0.5) transforms

[Sanjay Patel via Phabricator via llvm-commits]

spatel created this revision.
spatel added reviewers: efriedma, scanon, takahiro.miyoshi, evandro.
Herald added subscribers: hiraditya, mcrosier.
Herald added a project: LLVM.

As discussed in PR44330:
https://bugs.llvm.org/show_bug.cgi?id=44330
...the transform from pow(X, -0.5) libcall/intrinsic to reciprocal square root can result in small deviations from the expected result due to differences in the pow() implementation and/or the extra rounding step from the division.

This patch proposes to allow that difference with either the 'approximate functions' or 'reassociate' FMF:
http://llvm.org/docs/LangRef.html#fast-math-flags

In practice, this likely means that the code is compiled with all of 'fast' (-ffast-math), but I have preserved the existing specializations for -0.0/-INF that enable generating safe code if those special values are allowed simultaneously with allowing approximation/reassociation.

The question about whether a similar restriction is needed for the non-reciprocal case -- pow(X, 0.5) -- is deferred. That transform is allowed without FMF currently, and this patch does not change that behavior.


https://reviews.llvm.org/D71706

Files:
  llvm/lib/Transforms/Utils/SimplifyLibCalls.cpp
  llvm/test/Transforms/InstCombine/pow-sqrt.ll

-------------- next part --------------
A non-text attachment was scrubbed...
Name: D71706.234716.patch
Type: text/x-patch
Size: 4545 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20191219/a71412db/attachment.bin>