Threadripper CPU 32 cores with bifrost?

[abercaine]

Hi,
I was wondering if Bifrost is able to fully leverage the power of a 32 cores threadripper cpu.
I know for instance that houdini solvers are not playing that well with the 32 cores TR, and devs over there do not really recommend it.

So anybody has an insight on this with the current Bifrost? I would also be interested to know technically why solvers have trouble like that with so many cores. Is it an architecture limitation or it's more on the software side that might evolve in the future?
(I also aware that there was this windows issue capping the performance) 

Here is a youtube video with a guy doing a test with Bifrost with a TR 2990wx 32cores, at that time a did exactly the same setup with my i7 4930k (overclocked to 4.5)
and i had exactly the same simulation time!

thanks

[jan]

Microsoft fixed the 64 core ceiling issue months ago and AMD assured us the chip was working on a windows platform.

In my experience there's always been a bit of a performance hit with windows over linux but 25% is massive!

[abercaine]

maybe you could give a try and download the houdini files to run the benchmark; the 3990x is not listed would be interesting to see the result compared to the 3970x

https://www.vfxarabia.co/post/houdini-benchmark-cores-vs-clockspeed-updated

[jan]

Ran the big flip ocean sim. CPU activity dips frequently...

25 mins on the nose. Shaved 11s off the 3960X. Looks like those extra cores really don't come into play...

[abercaine]

yeah maybe softwares need to adapt to this type of cpus
the sweet spot seems to be the 3970x for rendering, simulations and also UI interactivity (which is mostly single threaded)   

[Christoph_Schaedl]

Would it be possible to sim on 32cores with the command line Bifrost and use the other 32cores for the current session?

http://help.autodesk.com/view/BIFROST/ENU/?guid=Bifrost_Common_bifcmd_Using_bifcmd_html

Hase someone tried that?

[jan]

Ooooo. That's interesting.

I'm currently batching a sim launching instances of mayabatch for every 20 frames. I don't think I want to shift to a command line bifrost solve since I want to also kick out playblast frames for preview. But I could restrict the mayabatch process

In my case the simulation is broken into 3 discrete fluids. It would be an interesting test to restrict the mayabatch process to 24/32 cores and run 2 processes concurrently. Data read/write might then become an issue I suppose.

I'll give this a try.

[abercaine]

i guess it will be better when bifrost will support distributed simulation.
We can already do this in houdini with PDG to overcome this issue about not using all cores.