cc @matthiasgoergens

I will wait for @ned-deily 's review as the final reviewer since he is the expert on macOS releases.

Should we do a performance benchmark of the generated executable?

@matthiasgoergens
My last attempt did not give a significant impact but it was measured by VM, checking the benchmark result at this moment would be valuable.
#88506 (comment)

Thanks!

Ha, I should learn how this pyperformance module works for my own experiments. Thanks for linking to that comment.

@ned-deily @methane cc @matthiasgoergens

https://gist.github.com/corona10/40d48caebb498ade456d2332c4fa430e

Here is the latest benchmark with my local macOS environment(not fully isolated) when applying only LTO optimization.

Here is the PGO + LTO benchmark (consistently 3-4% faster :))

https://gist.github.com/corona10/60d73e610f14661e131ed9037eab4d38

Thank you @ronaldoussoren and @methane