Hi,
I am currently working in the RF module on a cluster. Right now, I'm working with a very large geometry (large in that it contains many domains). It is so large that it takes a very long time for the "form union" operation to be performed. I've tried simplifying the final union operation by first doing several ordinary union operations with "keep interior boundaries" checked, the idea being to split up the process into several steps. That helped to an extent. Each one of these operations would run in a day or so, but now, it seems to have stopped working.
My question is, is there any way to use multiple processors for things like "form union" or other geometrical operations, or can that only be done for meshing and solving? I've tried it in batch mode and it only seemed to use one processor, even though I assigned it many. I'm currently using one node of the cluster with 48 processors. In earlier models, multiple processors were being used, but that was apparently for solving and meshing.
Now, I'm using the Gui mode in a torque shell on the cluster, but it's the same story. I assigned all of the 48 processors to the shell and then opened Comsol in it with the switches " -nn 1 -np 48". When I look at the usage stats, its using one processor to 100% capacity and not using the others at all. Its also using only about 11 gigs of RAM while 256 gigs are available.
Is there anything I can do to take advantage of the resources, or is that not possible for geometrical or other pre-processing operations?
Thank you in advance for any input,
A. J. Thomas
I am currently working in the RF module on a cluster. Right now, I'm working with a very large geometry (large in that it contains many domains). It is so large that it takes a very long time for the "form union" operation to be performed. I've tried simplifying the final union operation by first doing several ordinary union operations with "keep interior boundaries" checked, the idea being to split up the process into several steps. That helped to an extent. Each one of these operations would run in a day or so, but now, it seems to have stopped working.
My question is, is there any way to use multiple processors for things like "form union" or other geometrical operations, or can that only be done for meshing and solving? I've tried it in batch mode and it only seemed to use one processor, even though I assigned it many. I'm currently using one node of the cluster with 48 processors. In earlier models, multiple processors were being used, but that was apparently for solving and meshing.
Now, I'm using the Gui mode in a torque shell on the cluster, but it's the same story. I assigned all of the 48 processors to the shell and then opened Comsol in it with the switches " -nn 1 -np 48". When I look at the usage stats, its using one processor to 100% capacity and not using the others at all. Its also using only about 11 gigs of RAM while 256 gigs are available.
Is there anything I can do to take advantage of the resources, or is that not possible for geometrical or other pre-processing operations?
Thank you in advance for any input,
A. J. Thomas