[Sidefx-houdini-list] Distributed Fluid Sim question

Beatriz Lorenzo bealobo at gmail.com
Tue Aug 11 09:04:57 EDT 2009


I think it depends on the sim. As you need an overlapping region for
distributing the voxel regions, if the sim is very complex and fluid is
moving in many directions (=needs a lot of feedback from the adjacent
voxels) it might end up in lots of artifacts if you distribute in too many
jobs.

But, I've only done a few tests, so it'd be great to hear what you end up
with.

About multithreading, you can ask sidefx if they have any benchmark to see
up to how many procs the multithreading is reliably working. If it scale ok
up to 8 cores, and the sim is really complex, then I'd go for the first
option.

Cheers


Beatriz


2009/8/7 <craigleehoffman at aol.com>

>
> If I have an 8 core machine with a lot of RAM, what is best when trying to
> get the fastest performance for a large complex Fluid Sim?
>
>
> Computing the Fluid Sim in one job on the machine and letting it
> multithread and calculate as best it can?
>
> Splitting the Fluid Sim up into 8 smaller jobs and distributing one job per
> core on the one machine?
>
> Splitting the Fluid Sim up into 4 smaller jobs and giving each job two
> cores?
>
> Has anyone tried such tests?
>
> Thanks,
> Craig
>
>
>
> _______________________________________________
> Sidefx-houdini-list mailing list
> Sidefx-houdini-list at sidefx.com
> https://lists.sidefx.com:443/mailman/listinfo/sidefx-houdini-list
>



More information about the Sidefx-houdini-list mailing list