Discussion Closed This discussion was created more than 6 months ago and has been closed. To start a new discussion with a link back to this one, click here.
Parallel computing MatLab Comsol [warning -Identifier: root is not serializable]
Posted 2015年11月5日 GMT-5 04:37 Interfacing Version 5.0 2 Replies
Please login with a confirmed email address before reporting spam
Dear All,
I am running COMSOL servers in parallel using parallel computation toolbox (Matlab) to solve optimization problem.
For each time when the model gets loaded using parfor i get the below warning:
Warning: COMSOL Model Object
Name: LIB_9AH_1D.mph
Tag: Model
Identifier: root is not serializable
I am not sure if COMSOL server runs parallely.
Any reasons for this warning and can i ignore it? Further i am not sure if this warning is given by COMSOL or matlab.
Code Snippet:
parfor i = 1:N
comsolPort=[2036 2037];
t = getCurrentTask();
labit=t.ID;
mphstart(comsolPort(labit))
import('com.comsol.model.*');
import('com.comsol.model.util.*');
end
....
....
Thanks in advance for your help.
yours sincerely,
Chris
I am running COMSOL servers in parallel using parallel computation toolbox (Matlab) to solve optimization problem.
For each time when the model gets loaded using parfor i get the below warning:
Warning: COMSOL Model Object
Name: LIB_9AH_1D.mph
Tag: Model
Identifier: root is not serializable
I am not sure if COMSOL server runs parallely.
Any reasons for this warning and can i ignore it? Further i am not sure if this warning is given by COMSOL or matlab.
Code Snippet:
parfor i = 1:N
comsolPort=[2036 2037];
t = getCurrentTask();
labit=t.ID;
mphstart(comsolPort(labit))
import('com.comsol.model.*');
import('com.comsol.model.util.*');
end
....
....
Thanks in advance for your help.
yours sincerely,
Chris
2 Replies Last Post 2015年11月7日 GMT-5 01:45
Please login with a confirmed email address before reporting spam
Posted:
9 years ago
I think the problem is the other way around. COMSOL runs shared-memory and distributed parallel without MATLAB out of the box. This can be tested fairly easy with the FNL and many application library models.
Now add the MATLAB livelink. Each MATLAB license is good for one node only on the distributed parallel cluster job. Within each node, MATLAB will run parallel in shared memory with the parallel version of the MATLAB license. However, once you jump to distributed parallel, then you need one MATLAB license for each node that gets used. There is no equivalent of a FNL for MATLAB ! You can test this by running MATLAB standalone without COMSOL on your distributed parallel cluster.
Now add the MATLAB livelink. Each MATLAB license is good for one node only on the distributed parallel cluster job. Within each node, MATLAB will run parallel in shared memory with the parallel version of the MATLAB license. However, once you jump to distributed parallel, then you need one MATLAB license for each node that gets used. There is no equivalent of a FNL for MATLAB ! You can test this by running MATLAB standalone without COMSOL on your distributed parallel cluster.
Please login with a confirmed email address before reporting spam
Posted:
9 years ago
Dear Mr.James,
Thank you for the answer. In my case, i am not using distributed parallel. It is just a single node / Single computer with multiple cores. Hence once i start the matlab using 'matlabpool open 3', 3 matlab workers are connected.
May be that is the reason i could not think its a matlab license issue.
Thank you,
yours sincerley,
Christober
Thank you for the answer. In my case, i am not using distributed parallel. It is just a single node / Single computer with multiple cores. Hence once i start the matlab using 'matlabpool open 3', 3 matlab workers are connected.
May be that is the reason i could not think its a matlab license issue.
Thank you,
yours sincerley,
Christober