[Pixman] [cairo] pixman: New ARM NEON optimizations
sandmann at daimi.au.dk
Tue Feb 16 13:52:13 PST 2010
Siarhei Siamashka <siarhei.siamashka at gmail.com> writes:
> The biggest surprise here is the pathologically bad performance of 'memset'
> function in 'image' backend tests, especially for 'evolution' benchmark. My
> only guess is that glibc could have probably messed up with the caches somehow
> (maybe by improperly using nontemporal memory writes or something).
Could it be page faulting? I have seen profiles where a lot of time
was spent in fill routines ending up in the kernel page fault
handler. I have never tracked that down exactly, and I can't reproduce
your results here on x86-32.
There used to be a bug, where cairo and X would allocate huge
temporary pixmaps instead of clipping them to the destination
size. When it initialized them to 0, tons of page faults were
generated, but as far as I know, that was fixed years ago.
More information about the Pixman