[libcxx-commits] [libcxx] [libc++] Vectorize std::find (PR #156431)

[Nikolas Klauser via libcxx-commits]

philnik777 wrote:

> Looking at it (and I might have missed something, would be easier to see if the table would report the % change), we only get speedups for `long long`, but not `int`?

Correct. We forward to `wmemchr` with `int`, but unfortunately that's currently quite slow on macOS. We plan to disable that "optimization" on macOS in a follow-up to get the vectorization for 4 byte integral types as well.

https://github.com/llvm/llvm-project/pull/156431