Hi,
I wonder how to speed up CFDEM.
I am running a particle-laden jets. To reach the steady state, I have to run for a long time.
Could anyone tell me where I can modify to speed up, especially in DEM/in.xxxx
Thanks.
rq
A project by DCS Computing and CFDEMresearch
This website uses cookies to ensure you get the best experience on our website. By continuing to browse the site you are agreeing to our use of cookies. This website uses cookies. By continuing to browse the site you are agreeing to our use of cookies. More info
ckloss | Fri, 08/26/2011 - 14:54
there are multiple ways to
there are multiple ways to speed up your DEM simulation. First of all, you should look at the timing statistics what comsumes the most of the time. If e.g. the neigh time is more that say 10% you can tune the skin distance and binsize, if the other time is very high, you may suffer from load imbalance - but it all depends on your problem and also the machine you are running on. If you have further questions, feel free to ask (please use the DEM forum).
Christoph
venes520 | Wed, 03/28/2012 - 11:30
Hi Christoph, Can you give
Hi Christoph,
Can you give more specific instructions on how to speed up the simulation?
You mention the timing statistics, how do I access this timing statistics?
And how doesn't the skin distance and timestep affect the simulation speed (not the total simulation time).
By the way, is there a way to determine the optimal tiemstep and skin distance?
Thanks in advance.
dnojiri | Thu, 05/31/2012 - 13:58
Hellou Does "load imbalance"
Hellou
Does "load imbalance" means that the number of atoms alocated to each processor differ greatly?
if the answer is yes. How can I make the load balanced? Does liggghts have commands to achieve this?
Im currently pouring atoms in a cylinder, and then stirring them with a stl paddle.
The simulation runs faster with one processor than with four processors. Which is contrary to what I was expecting.
Any pointer that you could give me?
cgoniva | Wed, 06/06/2012 - 15:27
Hello, there is a load
Hello,
there is a load balancing method, but not in the public release.
Concerning the moving mesh performance, there will be a better parallel perfromance of moving meshes in future versions.
Cheers,
Chris