On Wed, 2008-11-19 at 09:01 +0100, Jasper van de Gronde wrote:
Felipe Sanches wrote:
I tested it and it worked ok, but I have no idea how to benchmark it. I removed the logging code that was windows-only.
Yeah, I'm afraid I'm just not familiar enough with Linux programming to know how to do accurate timing (let alone cross-platform timing). You could try profiling though. For me this doesn't work (for some reason I get really crazy profiling results on Vista), but it will probably work just fine for you, see: http://wiki.inkscape.org/wiki/index.php/Profiling
I don't know exactly how well it works with multi-threading though.
Not the most accurate profiling but what I used is this:
time inkscape --verb=FileQuit myfile.svg
This will cause the file to open, render once and the Inkscape quits. Doing that on Bulia's gradient mesh example (which uses blur) it took 5.3s without OpenMP and 4.8s with. So there is about a 10% speed up. Not sure how that breaks down overall, but it seems like a nice improvement.
here is my version of the patch, I hope it helps.
If a few people test the code and no fundamental problems arise I'll probably commit it, so yes, then it would definitely help :) (To prevent the build from breaking.)
I changed the autotools stuff a bit to be more flexible with other compilers pulling an macro from the autotools macro archive. I'll commit it once SF starts talking to me again...
--Ted