@MoolahNasreddin Thanx a lot ! Problems with simulation it's hard to multithread it , because each calculation is dependant to the next one. And python in Blender don't support multithreading module I think too. Because of that , the number of core change anythings :-( I hope to increase performance with Cython.
@tubento I already try to instance metaball instead of sphere and it's too long or just not render at all when you have too much particles. I know some developper work on a "polygonizer" , a density space triangulation. Probably better performance in my case. If not , I must too think about my own system but for now I have to complete my engine :o)
@tubento Thank's a lot ! I have a lot of other idea to increase performance. Like dynamic substep. But I have to continue more basic things in my engine before thinking about that. Cython (convert Python code to C compiled code) is another thing I must to take a look when my engine is done.
@d3sertst0rm haha , yes it's look too long again but if I can get a minimum of 10 to 20 times faster with Cython (convert Python to C) , it's starting to be interesting
Man, it's already looks outstanding! Hope you'll make it simulating in 15-30 minutes on double core Athlon (or Intel - nevermind).
MoolahNasreddin 1 month ago in playlist More videos from PyroEvil
@MoolahNasreddin Thanx a lot ! Problems with simulation it's hard to multithread it , because each calculation is dependant to the next one. And python in Blender don't support multithreading module I think too. Because of that , the number of core change anythings :-( I hope to increase performance with Cython.
PyroEvil 1 month ago
@PyroEvil
Ah, I forgot about this but already heard!
MB After Depsgraph Blender Devs' team will do something about this old problem...
MoolahNasreddin 3 weeks ago
One more thing are you thinking of using metaballs in future?
tubento 1 month ago
@tubento I already try to instance metaball instead of sphere and it's too long or just not render at all when you have too much particles. I know some developper work on a "polygonizer" , a density space triangulation. Probably better performance in my case. If not , I must too think about my own system but for now I have to complete my engine :o)
PyroEvil 1 month ago
Great going pyro hurray for faster calculations.
tubento 1 month ago
@tubento Thank's a lot ! I have a lot of other idea to increase performance. Like dynamic substep. But I have to continue more basic things in my engine before thinking about that. Cython (convert Python code to C compiled code) is another thing I must to take a look when my engine is done.
PyroEvil 1 month ago
@d3sertst0rm haha , yes it's look too long again but if I can get a minimum of 10 to 20 times faster with Cython (convert Python to C) , it's starting to be interesting
PyroEvil 1 month ago
yay now it only takes a day to sim xD
d3sertst0rm 1 month ago