yade-users team mailing list archive

[Anton Gladky <question228175@xxxxxxxxxxxxxxxxxxxxx>]

Question #228175 on Yade changed:
https://answers.launchpad.net/yade/+question/228175

Anton Gladky proposed the following answer:
2013/5/6 Bruno Chareyre <question228175@xxxxxxxxxxxxxxxxxxxxx>:
> Also, is it really 11362 or is there a typo? With this number of particles you should get the result in minutes.

Maybe there are too much iterations?

Actually, Ocean, why did you decide, that your simulation lasts too long?
Did you compare the same simulation with other programs?

>From my point of view, Yade code is relatively good parallelized
already. We can, of course, try to implement some more techniques,
but it will unlikely influence the common performance sufficiently.

If you have a lot of particles (say, >5*10^5), I would recommend you
to try some MPI-based codes (liggghts or esys-particle), but you
should know, that communication between MPI-nodes/processes
is an also very time-consuming operation.

Anyway, if you are going to contribute cuda/opencl module, we are
always glad to review and apply your code.

Cheers,

Anton

-- 
You received this question notification because you are a member of
yade-users, which is an answer contact for Yade.