PhoenixFD 3.0 with CUDA base simulation !! is this a dream ?! ![]()
you mean gpu simulation? not supported yet
BTW - I tried RF2015 GPU simulation - 35% of utilization on my gtx980… It is slower, than plain CPU simulation.
But i saw some opensource wavelet turbulence simulators that run under CUDA was 50% faster than CPU processing !!
with export of the result into a file and complex geometry interaction?
Just simple google this: ‘opensource wavelet turbulence’ . the result is list of the phd student’s projects & some of GDC & Siggraph & Pixar !! :))
The wavelet turbulence may have some benefit from GPU since it’s simpler and has more calculations per cell. But 50% increase from “CPU” is nothing… If this CPU was a standard quad core, you just get a 6-core and you are there ![]()
Would Glu3D technology be helpful?
it seems that i have to going to the Houdini … But via the VRayVolumeGrid in Max ![]()
well, this is an option too. if you have time, share the sim times to have a reference ![]()
I tested a simple fire that simulated with houdini pyrofx with openVDB export … it’s realtime simulation with write vdb cache !! opposite of FumeFX or phoenix that hours to process !!
the vdb cache size is very smaller than .aur or .f3d (at least 60% smaller) !! wooooooooow …
Since the default settings and resolutions are different, the times are sizes should be different, of course. If you equalize the settings, I’m sure that the difference will definitely not be so “wow” ![]()
our tests with equal settings do not show significant speed difference
OK … But why Dreamworks use the Houdini for productions ?? Stable + Smooth & flexible interface + Simple to control … & so … ![]()
Only v-ray volume grid !! + houdini !!