[PATCH] D87479: [InstCombine] Don't sink the fdiv from (fmul (fdiv 1.0, %x), %y) if the fdiv isn't in the same basic block as the fmul

[Sanjay Patel via Phabricator via llvm-commits]

spatel added a comment.

In D87479#2268474 <https://reviews.llvm.org/D87479#2268474>, @lebedev.ri wrote:

> I personally really think this should be viewed as phase ordering issue, that should be addressed by adding/moving pases in pipeline.

I agree. We have 3 likely problems already identified in this discussion: instcombine vs. LICM in LTO, instcombine vs. (no-)unroll, instcombine vs. loop-vectorizer.
The common opponent is instcombine of course because it does sinking transforms without cost models.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D87479/new/

https://reviews.llvm.org/D87479