+- tyFlow Forum (https://forum.tyflow.com)
+-- Forum: tyFlow Discussion (https://forum.tyflow.com/forum-1.html)
+--- Forum: General Discussion (https://forum.tyflow.com/forum-2.html)
+--- Thread: first observations (/thread-111.html)
first observations - super gnu - 04-04-2019
Hi Tyson, like many ive been very excited to get my hands on your plugin.
so far ive got as far as installing it, and going through the example scenes.
my first two observations are:
the cloth 001 file locks up max when i try to open it (max2018, windows 7)
second, opencl implementation seems sub-optimal, but thats probably my imbalanced system.
i have an old 6 core 3930k running at 4.6 ghz, 64 gb ram, and 2x titanX maxwell gpus.
when i playback any of the test scenes, i get 100% cpu usage, which is BRILLIANT coming from pflow... however a) only one gpu is used (any plans to implement multi gpu solving?) , and b) my MAXIMUM gpu usage is 40% of one gpu.. most scenes dont get above 20% gpu usage..
im guessing this is due to my cpu not being meaty enough to feed the gpu with sufficient data, but thought id mention it.
im sure you are gonna be swamped with feedback as people start to get their teeth into this plugin... amazing work. MAX feels like a new toy for the first time in a decade.
oh, one last thing.. my browser says this forum is not secure.. is that a problem at my end or yours?
thanks! Robin.
RE: first observations - Heerockss - 04-04-2019
(04-04-2019, 07:29 AM)super gnu Wrote: Hi Tyson, like many ive been very excited to get my hands on your plugin.I would suggest that the GPU Usage may be low because only some parts of the simulation can be calculated on the GPU. Not sure about that tough.
so far ive got as far as installing it, and going through the example scenes.
my first two observations are:
the cloth 001 file locks up max when i try to open it (max2018, windows 7)
second, opencl implementation seems sub-optimal, but thats probably my imbalanced system.
i have an old 6 core 3930k running at 4.6 ghz, 64 gb ram, and 2x titanX maxwell gpus.
when i playback any of the test scenes, i get 100% cpu usage, which is BRILLIANT coming from pflow... however a) only one gpu is used (any plans to implement multi gpu solving?) , and b) my MAXIMUM gpu usage is 40% of one gpu.. most scenes dont get above 20% gpu usage..
im guessing this is due to my cpu not being meaty enough to feed the gpu with sufficient data, but thought id mention it.
im sure you are gonna be swamped with feedback as people start to get their teeth into this plugin... amazing work. MAX feels like a new toy for the first time in a decade.
oh, one last thing.. my browser says this forum is not secure.. is that a problem at my end or yours?
thanks! Robin.
Oh and 3930k@4.6 is definitely meaty enough ^^
RE: first observations - super gnu - 04-04-2019
ok, im not sure if this is expected behaviour either: playing with the lovely "worms" example scene.. ive added an animated collision teapot, and a ton more worms.
i was getting lots of interesections and worms getting stuck to each other and the tespot.. so i just upped every steps multipler i could find. time step to 1/4, particle solver to 30, physx to 20,16,8 i also enabled "ccd" in physx, since it seemed a good thing to have (ah the joys of randomly pushing buttons)
anyway, its working better it seems accuracy wise. , but cpu usage is now fluctuating around 15-30% and gpu usage is 0-5% doesnt seem terribly efficient...