[PATCH] D19558: Codegen: [X86] Set preferred loop alignment to 32 bytes.

[Eric Christopher via llvm-commits]

echristo added a comment.

Hi Quentin,

To be clear we're seeing multiple benchmarks speeding up 10% because of this change. That said, size numbers would be good just so we know what we're getting into - though I don't think they should be a blocking factor here. If that's an issue then we can make loop alignment in general dependent upon optimization level, but for cpu optimization we don't want that to be the guiding factor (i.e. the patch is still fine, but we might want to change where we align loops).

Thanks!

-eric


Repository:
  rL LLVM

http://reviews.llvm.org/D19558