[Scuds]

Oh yes, apple hardware continues to improve and M4 pro still the single threaded champion of anything under 300 w.

FWIW - last stage where the binary is produced takes the longest and is single threaded and that's the largest difference between release and debug.

[hu3]

Sounds suspicious to me.

At a quick glance, most benchmarks show only 50% to 60% single threaded improvement between M1 and M4.

That wouldn't explain compilation times going from 6 minutes to 2 minutes.