-
-
-
This also means
@musllibc posix_spawn will fill the fork gap on NOMMU systems cleanly/safely (unlike vfork) once we get NOMMU working.
End of conversation
New conversation -
-
-
@RichFelker@musllibc@mpixorg how is the perf on fork() without exec? cygwin is 20x slower v linux native, ours is 7x slower v linux -
@richinseattle@mpixorg Are you asking about midipix? My numbers are using@musllibc on linux. -
@RichFelker@mpixorg@musllibc yes, sorry, was referring to midipix -
@richinseattle@mpixorg While I'm excited about the midipix project, I don't work on it myself. I'm the@musllibc maintainer. -
@richinseattle@mpixorg Just posted the posix_spawn timings link on HN because I think it's relevant to misplaced concerns about fork speed. -
@RichFelker@mpixorg thanks for the clarification, I'm blind or the midipix fork implementation hasnt been released so I cant test yet -
@richinseattle@mpixorg Indeed, the midipix runtime components are not released yet. Hopefully very soon now.
End of conversation
New conversation -
-
-
@RichFelker@musllibc posix_spawn (CLONE_VM) speedup is prop to mem used right? What parent size for 25x speedup? Do you have a progression? -
@pixelbeat_ Yes. posix_spawn wins even with trivial static binary and no allocations, but the ~25x is around 40mb in small mappings. -
@pixelbeat_ For trivial static binary, it's 280µs vs 450µs on the system I test on.
End of conversation
New conversation -
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.