Hi, All:
I am currently running LIGGGHTS simulation test on a large cluster, if there is no limitation on number of nodes and processors per node, I have two companion questions:
1. Does anyone have an idea how many particles that LIGGGHTS/CFDEM is currently able to simulate?
2. When we do multi-processor simulation, how many number of particles is a reasonable number for one processor?
Thanks for any suggestions in advance!
Roy Li
cstoltz | Sat, 05/04/2013 - 02:09
No real single answer to
No real single answer to either question. Depends on your simulation setup, hardware, length of simulation, and patience. I've run a hopper flow simulation in which the hopper w/ 1e7 particles on 60 cores and made it through about a million steps in a week.
Regards,
Chris
dbxmcf | Sat, 05/04/2013 - 17:14
Thanks, Chris, I just want to
Thanks, Chris, I just want to know the current limit of LIGGGHTS that anyone has tried.
Roy
ckloss | Mon, 05/06/2013 - 08:24
>>Thanks, Chris, I just want
>>Thanks, Chris, I just want to know the current limit of LIGGGHTS that anyone has tried.
The maximum I've heard of (with LIGGGHTS) was about 1e8 particles. With LAMMPS, see eg http://lammps.sandia.gov/bench.html
Cheers
Christoph