A 100LOC C impl of memset, that is faster than glibc's

Re:A 100LOC C impl of memset, that is faster than glibc's by josefx on Friday November 12, 2021 @09:21AM (#29199486)

> on x86, probably "rep movsb" or whatever)

Sadly I don't have a link, but as far as I remember rep movsb was always hilariously slow. So memcpy implementations tried to optimize copies using half a page of vector instructions with size and alignment tests, which of course killed the CPUs instruction cache.

[ Parent | on HN ]

Re:A 100LOC C impl of memset, that is faster than glibc's by sedatk on Saturday November 13, 2021 @02:11AM (#29208030)

Yes, a compiler would at least add a combo of rep movsq + rep movsd + rep movsw to the mix before finishing the final remainder with rep movsb. Vector instructions might help tremendously too.

[ Parent | on HN ]

Re:A 100LOC C impl of memset, that is faster than glibc's by benttoothpaste on Friday November 12, 2021 @02:26PM (#29203509)

Always hilariously slow? That must have been before Ivy Bridge.

[ Parent | on HN ]

Re:A 100LOC C impl of memset, that is faster than glibc's by bjourne on Saturday November 13, 2021 @06:16AM (#29209024)

Hilariously slow before Intel engineers decided to optimize the shit out of the construction. :)

[ Parent | on HN ]