[llvm] [AArch64] New pass for code layout optimizations. (PR #184434)

[Ahmad Yasin via llvm-commits]

ayasin-a wrote:

> Do you have any idea what percentage of the performance stabilization is coming from cmp/fcmp fusion, vs. other sources like branch prediction?

Let me dig out data that correlates pair dynamic execution to noise reduction we did on some internal workloads.

https://github.com/llvm/llvm-project/pull/184434