[llvm] [AArch64] Add @llvm.experimental.vector.match (PR #101974)

[Ricardo Jesus via llvm-commits]

rj-jesus wrote:

Hi @paulwalker-arm, No worries at all, thank you very much for having a look! Once #115189 lands I'll rebase this patch and simplify `LowerVectorMatch` accordingly. In the meantime, if there's anything you'd like to see done here, please let me know.

You're right, I _believe_ that when I tried with separate main and tail loops, I wasn't fully convinced that the performance difference was worth the extra code size and complexity. but I'll have a look again since I can't really recall the numbers.

https://github.com/llvm/llvm-project/pull/101974