[libc-commits] [libc] [libc][string] Optimize inline_memmove routine for Aarch64 (PR #232494)

[Victor Campos via libc-commits]

vhscampos wrote:

The regression is concerning, specially if these are with -O3 (are they?).

I don't think it's worth aligning aarch64 with x86 just because if it doesn't come with improvements across the board. I prefer to wait for your next experimentation part before we make a decision for this one here.

By the way, we at Arm might be able to help with benchmarking as we do have some infra set up. Let me know what you think.

https://github.com/llvm/llvm-project/pull/232494