On Mon, Oct 14, 2024 at 6:53 AM Bruno Haible via Gnulib discussion list <bug-gnulib@gnu.org> wrote: > Hi Sam, > > Thanks for the contribution offer! > > > I've noticed that GZIP trails behind zlib in performance and part of this > > is down to the fact that zlib is using a more efficient CRC32 > > implementation. > > How much of a speedup do you obtain in gzip overall (not in CRC32 alone) > for large files, through these slice-by-4 and slice-by-8 techniques?
Thanks for the offer, indeed! Lasse Collin suggested precisely this (improve gzip's CRC code) a couple months ago, which is what prompted my early-September patches in gzip. I measured up to 39%(!) overall speedup using a i7-4770S on large, "real" files when with zlib-ng-linked gzip on Fedora 40, but I am leery of making a tool like gzip depend on such a shared library. If it is easy to make gnulib's CRC32 code competitive (across common arches), that would be best. Actually, with pathologically uniform (all-NUL) files, (as done by one of gzip's own tests), the speedup was more than 7x.
