[clang] [llvm] Disable pre-link LTO vectorisation and loop unroll for at O2 and O3 optimisation levels (PR #195886)

[Florian Hahn via cfe-commits]

https://github.com/fhahn commented:

Looking at the latest runtime data, it looks like there are also a number of workloads that regress slightly on AArch64 (https://github.com/llvm/llvm-project/pull/195886#issuecomment-4830998464). E.g. SPEC2017 looks worse overall. 

Do we already understand the root causes for some of the larger regressions, e.g. `736.ocio_r | 1.91%` or `714.cpython_r | 1.80%`.

It would probably also be good to include the latest performance data + compile-time impact in the PR description so it is easy to find




https://github.com/llvm/llvm-project/pull/195886