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.
Out of memory during assembly BUT the mesh number is just 3608
Posted 24 apr 2011, 11:38 GMT-4 Mesh, Studies & Solvers Version 4.1, Version 4.2a, Version 5.2a 10 Replies
Please login with a confirmed email address before reporting spam
My operation system is windows 32bit, and the internal memory is 2GB
Attachments:
Please login with a confirmed email address before reporting spam
it ran smoothly to some 576.375 seconds, using from 0.65 to 1.35 Gb (in bout the same time too almost 1:1), then it hangs, looks like your response is getting suddenly very non-linear (T slope almost vertical) so I ended stopping it, that took 4 minutes extra ;)
Have you tried stop some of the non essential running applications, to free more RAM ? As it might need a RAM boost/spike during meshing or assembly, I do not believe the indication is peak RAM use. On my PC I have plenty, but before on my laptop I used those tricks, also with 3.5 I ran it in server client mode, that gained some 30% on the number of DoF that passed. But I'm not sure how it behaves in V4, haven't tried that
--
Good luck
Ivar
Please login with a confirmed email address before reporting spam
Hi
it ran smoothly to some 576.375 seconds, using from 0.65 to 1.35 Gb (in bout the same time too almost 1:1), then it hangs, looks like your response is getting suddenly very non-linear (T slope almost vertical) so I ended stopping it, that took 4 minutes extra ;)
Have you tried stop some of the non essential running applications, to free more RAM ? As it might need a RAM boost/spike during meshing or assembly, I do not believe the indication is peak RAM use. On my PC I have plenty, but before on my laptop I used those tricks, also with 3.5 I ran it in server client mode, that gained some 30% on the number of DoF that passed. But I'm not sure how it behaves in V4, haven't tried that
--
Good luck
Ivar
Thank you, Ivar
about the "suddenly very non-linear", do you any suggestions about how to fix such kind of problem?
Please login with a confirmed email address before reporting spam
that is related to the equations, and probably your material properties, once you localise the place, you could try to ask the solver to take smaller steps you might chain several "{range(,,) range(,,)}" ... in a "{}" "list"
--
Good luck
Ivar
Please login with a confirmed email address before reporting spam
Number of vertex elements: 24
Number of edge elements: 872
Number of boundary elements: 34868
Number of elements: 949673
Minimum element quality: 0.1935
I do not think it is an issue with the shortage of phsyical memory, at least in my case. I am running this program on a computer which has 196 Gb of RAM. I was continuously checking the windows task manager to see how much memory COMSOL is utlizing while attempting to solve this model and it is around just 12 Gb. I know I can solve this problem using a coarser mesh but that still does not answer as to why COMSOL is giving me this error message when there is more than enough physical memory available in my computer. I have attached the COMSOL model. I have removed the mesh from the model as its size was more that could be attached. But I have left the mesh setting in the model.
Could some one provide me with an explanation to this issue and possible solutions?
Attachments:
Please login with a confirmed email address before reporting spam
indeed looks strange, that is not so many elements and DoF's and you seem to have really plenty of RAM.
Are you sure it has nothing to do with your Java heap size (check/search the Forum and the KB) ?
else it could be from some weird BC settings making COMSOL run in loops, or perhaps a frame reference that makes the model excessively non linear ?
I can only suggest to use the later versions, they have greatly improved the stability of COMSOL since 4.1
And I'm not by my COMSOL WS so I cannot look at your model, it's soon the middle of the night for me ;)
--
Good luck
Ivar
Please login with a confirmed email address before reporting spam
Thanks for the reply.
I am sucessful in running the model for several coarser meshes. I ran the model from 'Normal' to 'Extra Fine' mesh setting of COMSOL. I am not able to run it for 'Extremely Fine' option as it is giving the Failed to find consistent initial values' error message. The fact that I am able to run for coarser meshes gives me the impression that the BCs are correct. As you suggested I will shift to COMSOL 4.2 and hopefuly this issue would be resolved.
Once again, thanks.
Bobby
Please login with a confirmed email address before reporting spam
Bobby
Ruston, LA.
Please login with a confirmed email address before reporting spam
when nothing else works, (or even before) often support can help particularly for such specific more programming / RAM management issues. Pls report back here as it is of general interest.
I use often a few million DoF, but has so far seldom got your error, but I have managed often to saturatemy 48Gb of RAM ;)
--
Good luck
Ivar
Please login with a confirmed email address before reporting spam
thanx in advance
Please login with a confirmed email address before reporting spam
FYI : I managed to remove this error (Out of memory ...) by playing with meshes.
It is not about how many elements you use. You need to figure out where to use finer meshes.
Hope this helps you.
Note that while COMSOL employees may participate in the discussion forum, COMSOL® software users who are on-subscription should submit their questions via the Support Center for a more comprehensive response from the Technical Support team.