[PATCH] D154678: [InstCombine] Fold IEEE `fmul`/`fdiv` by Pow2 to `add`/`sub` of exp

[Matt Arsenault via Phabricator via llvm-commits]

arsenm added a comment.

In D154678#4479850 <https://reviews.llvm.org/D154678#4479850>, @RKSimon wrote:

> Do we do much floatbits wrangling in the middle-end? This feels like something that should be done later, but I don't know if there are precedents.

I actually think we need to do the opposite and turn int-to-fp, not fp-to-int (e.g. D151939 <https://reviews.llvm.org/D151939> and parents).


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D154678/new/

https://reviews.llvm.org/D154678