couple of questions

i have self writen dynamic effects system (self colliding particles), but a count over 200 objects is reducing the perfomance dramatical. i fixed the perfomance a bit, by using double pass 2x2 rastering at a 10x10 grid and timedelays, the rastering has not much effect, it seems it takes to much of time to figure out in which segment my particles are.

if im using threading, the perfomance stays similar. is there a way to fixate a cpu for a thread? (at my computer the second cpu is starting, but its only running with max 10%, while the first cpu is using 100%, so how i can fixate a thread only to one cpu without getting split up?)

if not, i have to wait for stronger cpus, or i have to reduce the count.

or wait, maybe it would help if im reducing the the double raster size 1. 2x2 2. 6x6, i have to figure this out. maybe im using to much of cells.

oh, i won a bit better performance by using copyTo()

i reached a bit a better result by change the size of the raster and i did a bit a other structure, but im not that much happy with… now with 220 particle its running with 40 frames, but i wanna reach 50 frames with 300 particles, before at 200 particles with < 20 frames.

:smiley: 500 pieces with 30 fps and i still see some parts to speed this thing up :slight_smile: cool

Be sure to use PStats to find bottlenecks. That can save you a lot of time. :wink: