[PATCH] D86364: [ValueTracking] Interpret GEPs as a series of adds multiplied by the related scaling factor

[Quentin Colombet via Phabricator via llvm-commits]

qcolombet added a comment.

One thing we could do to help compile time is maybe add more caching to the Query API.
Right now it only caches `llvm.assume` AFAICT, whereas we could cache known bits for values. That's what we did in the Machine IR variant of computeKnownBits.

This would be a functional change though as in theory not having a cache could give you more information. E.g., imagine we hit a value at Depth 6 with a max depth of 6 and then later on, we hit it again at Depth 4. With a cache that value would have the information cached on the first hit, thus we would only have information for it with 0 depth worth of analysis, whereas with the current implementation the second call would have 2 depth worth of analysis.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D86364/new/

https://reviews.llvm.org/D86364