[PATCH] D103695: [WIP][RFC][Utils] Helper script to check sanity of cost tables vs scheduler models

[Simon Pilgrim via Phabricator via llvm-commits]

RKSimon added a comment.

As I said in the summary - when the script reports a diff you then need to start looking into manually to see where the problem is, its most likely the cost tables but the models aren't always great.

At the moment I'm mainly using osaca (inside godbolt) when I need an alternative analysis to llvm-mca

If you have access to particular CPUs that we have models for - PLEASE run llvm-exegesis and report any mismatches on bugzilla - ideally just attach analysis.html report - https://llvm.org/docs/CommandGuide/llvm-exegesis.html


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D103695/new/

https://reviews.llvm.org/D103695