[PATCH] D44814: [CodeGenPrepare] Split huge basic blocks for faster compilation.

[Michael Zolotukhin via Phabricator via llvm-commits]

mzolotukhin added a comment.

Hi Eli,

Thanks for the input, please take a look at the updated patch. For now I didn't introduce a logic to skip debug intrinsics - is it needed from a correctness point of view? I've seen test-cases with thousands of `llvm.dbg.value` intrinsics that would also benefit from splitting (provided it's legal). Also, I wonder if we actually should only do that on -Os - maybe it would make sense to do that on -O3 too with a significantly high threshold. What do you think?

Thanks,
Michael


Repository:
  rL LLVM

https://reviews.llvm.org/D44814