On Tuesday 14 September 2010 15:24:40 Dmitri Vorobiev wrote: > From: Jonathan Morton <[email protected]> > > This test is a modified version of Siarhei's compositor throughput > benchmark. It's expanded with explicit reporting of memory bandwidth > consumption for the M-test, and with an additional 8x8-random test > intended to determine peak ops/sec capability. There are also quite a > lot more operations tested for.
Thanks for pushing this benchmarking stuff forward. It's a bit embarrassing and I made this benchmark program as a quick hack to get some performance numbers when tuning ARM NEON optimizations. It worked for me, but it was not good enough for adding into pixman tree the way it was. The biggest problem (but easy to solve) is the use of 'gettimeofday' and 'memalign' functions, which are available not on all supported systems. I listed some of the issues/todo items here: http://lists.freedesktop.org/archives/pixman/2010-September/000470.html I will not go into further details, because reviewing mostly my own code is kind of weird :) I would wait for Søren's comment(s) to see what he considers to be real blockers. But thanks again for picking up this code. That's much appreciated. -- Best regards, Siarhei Siamashka
signature.asc
Description: This is a digitally signed message part.
_______________________________________________ Pixman mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/pixman
