[Pixman] [cairo] pixman: New ARM NEON optimizations
siarhei.siamashka at gmail.com
Tue Feb 16 08:53:15 PST 2010
On Tuesday 16 February 2010, Benjamin Otte wrote:
> On Tue, 2010-02-16 at 17:56 +0200, Siarhei Siamashka wrote:
> > The problem of cairo-perf-trace is that it does not cover 16bpp desktop
> > color depth well which is still used on a lot of ARM devices. Also a lot
> > of the standard tests run too long and are very memory hungry
> > ('ocitysmap' test is even practically impossible to run on this hardware
> > due to excessive swapping).
> I think we'd all be happy if there were some real-life tests recorded on
> embedded hardware so we're able to measure that workload, too.
> The hard part here is of course figuring out the proper things to test,
> the rest is just running cairo-trace (I'm assuming that one just works
> on ARM?) and uploading the results.
When I tried it last time, cairo-trace just does not work right with 16bpp
desktop. Well, it runs, but it records (and later plays back) something
different from what was really executed. Basically it looked like the
operations with 16bpp destination were changed into operations with 32bpp
destination in a trace playback.
Maybe it's not too difficult to address.
More information about the Pixman