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.
LU factorization Out Of Memory
Posted 2010年8月26日 GMT-4 11:36 4 Replies
Please login with a confirmed email address before reporting spam
Dear All,
I run a comsol/matlab script (let's say SourceFile) and it works.
Then I pick up one parameter and make it change in a loop (let's say LoopFile).
It runs for many steps, until LU factorization Out Of Memory occurs.
I think the following things are strange:
1) if, in SourceFile, I manually set that parameter to the value for which LoopFile breaks, and I run the file, it works without going out of memory.
2) I must close and open again the application "comsol with matlab", otherwise for every script the application goes out of memory during LU factorization.
If anybody has a suggestion about what might be wrong, I would really appreciate it.
Thanks
I run a comsol/matlab script (let's say SourceFile) and it works.
Then I pick up one parameter and make it change in a loop (let's say LoopFile).
It runs for many steps, until LU factorization Out Of Memory occurs.
I think the following things are strange:
1) if, in SourceFile, I manually set that parameter to the value for which LoopFile breaks, and I run the file, it works without going out of memory.
2) I must close and open again the application "comsol with matlab", otherwise for every script the application goes out of memory during LU factorization.
If anybody has a suggestion about what might be wrong, I would really appreciate it.
Thanks
4 Replies Last Post 2010年8月27日 GMT-4 01:44
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Well I've had similar LU Out of Memory problems, however this is simply due to the fact that I have not enough memory on my comp 4GB.
What was interesting was that for every time I ran the study, ever so often the Memory footprint of COMSOL would increase, almost like a memory link. If I ran it about 4-5 times it would get so bid that it would literally go over the tipping point for my computer resources and cause a crash. Therefore I would have to close and restart COMSOL where it would be at a lower memory footprint so to speak, annoying.
Perhaps this is what is happening, everytime time you do a loop of 10 or so analysis calls, the memory footprint incrementally increases to a point where one of the study calls fails?
Regards Mike F
What was interesting was that for every time I ran the study, ever so often the Memory footprint of COMSOL would increase, almost like a memory link. If I ran it about 4-5 times it would get so bid that it would literally go over the tipping point for my computer resources and cause a crash. Therefore I would have to close and restart COMSOL where it would be at a lower memory footprint so to speak, annoying.
Perhaps this is what is happening, everytime time you do a loop of 10 or so analysis calls, the memory footprint incrementally increases to a point where one of the study calls fails?
Regards Mike F
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Yes, Michael,
it seems like that, cause otherwise it would go out of memory also if I run the SourceFile and, instead, this one always works fine.
I would like to add that I followed some advices on the internet to get rid of some problems that matlab2007a gives with win7.
In particular I substituted the original folder jre1.5.0_07
C:\Program Files\MATLAB\R2007a\sys\java\jre\win32\jre1.5.0_07
with the folder
C:\Program Files\Java\jre6
properly renamed.
Thanks
it seems like that, cause otherwise it would go out of memory also if I run the SourceFile and, instead, this one always works fine.
I would like to add that I followed some advices on the internet to get rid of some problems that matlab2007a gives with win7.
In particular I substituted the original folder jre1.5.0_07
C:\Program Files\MATLAB\R2007a\sys\java\jre\win32\jre1.5.0_07
with the folder
C:\Program Files\Java\jre6
properly renamed.
Thanks
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Matlab though useful can be annoying when it comes to optimization calls, I have this annoying problem where over time it accumulates Memory until it eats all of it up :), apparently it is due to a conflict with Windows memory management and matlab, any file created by matlab below a certain size something like 512Kb is not removed from the memory as it should normally be, hence after about 10,000 calls the memory begins to build :D
I hope you get your problem fixed
Regards Mike
I hope you get your problem fixed
Regards Mike
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
I had a similar post regarding Comsol too: www.comsol.com/community/forums/general/thread/8542/
Also, when I had a talk with some groups, they mentioned that garbage collection of comsol 3 (when run with matlab) was not that good, and they were calling their own routines (I think simply calling java gc) at the end of their matlab files.
Also, when I had a talk with some groups, they mentioned that garbage collection of comsol 3 (when run with matlab) was not that good, and they were calling their own routines (I think simply calling java gc) at the end of their matlab files.