Yes, run https://github.com/antonblanchard/will-it-scale on operating systems with hilarious grandstanders like OpenBSD and Illumos, then run it on Linux and compare the results. Run a highly parallel make on a modern server and tell me which OS remains usable for other work. OpenBSD runs acceptably on 2-4 core laptops with HT disabled because a desktop workload doesn't typically invoke a lot of the kernel implementation during steady state, but you can do it there with a moderate disk and network load. That said, syscalls are more expensive in OpenBSD than contemporary operating systems so even as a uniprocessor OS it is funny.

Fair points. I’ll wager that most users are willing to accept the trade-offs in favor of security, given that it likely won’t affect most day-to-day usage.

(Also, I’m missing what’s so “funny” and “hilarious” about it.)

I use OpenBSD for fun on a couple systems and it is a cute OS, there are things to love about it. But the grandstanding is funny.

Will-it-scale has a dependency on hwloc which is not supported on OpenBSD.

Do you have a better synthetic benchmark to suggest?

And "speed" does depend on expectations. Compute tasks should be approximately equivalent, right? And os-limited tasks with many syscalls less so. You don't specifically state your expectations. Just that you are disappointed.

This is really the benchmark you want. There are only a dozen lines to patch. AFAICT OpenBSD neither has a cpuset(3) like API nor any intention of growing one, so you can just delete all the affinity code.

I don't think anyone that uses OpenBSD will claim that it is fast.