Discussion Forum

Hi,

I do have some issues withe flow analysis in general but I did not get a freeze yet, can you upload the model or a some version of the geometry you are using?

Cheers

Good Day,
Yes, I too have noticed more of this behavior with 4.2. With task manager one will see 'not responding'. COMSOL will report the simulation is using 3-5GB but this is well below the system capacity at 32GB. Reset History does not seem to have any effect, nor does deleting the mesh and solutions. It seems more common if there is more than one session of COMSOL open.

Not all sessions can be recovered so a 'save' before solution or a mesh is an absolute MUST.

For very long time dependent simulation I will usually split the study in parts and use the restart features.. so that I have a chance to loo at the result before deciding to cotinue it or not...

it is very helpful in many situations and is in fact time saving.

I think It could be used in your context to ID the cause of the freeze...

Good Luck

Hi

My 4.2 has been running nicely on Win 7-64b, until last MS OS security update (but not fully sure its linked), now sometimes it jams/crashes, and creates many hs-err files. It seems to be linked with a file saving / write-back issue. It has started with other MS programmes as well as IE and html files too ;)

--
Good luck
Ivar

Hi,

I do have Win7-64bit too, and I have a couple of important security updates pending with MS OS, could you please tell me the update number so I may avoid it?

Cheers

Hi

first of all I cannot pinpoint yet exactly what is happening, and even if it is linked to a security patch.
I have got 7 patches to my OS in July and about 3 times that many in June, so which one ? if any of them ?
My problem is unpredictable and sneaky, not anything clear and easy to find, but its rather new (I expect at most a month or so)

--
Good luck
Ivar

Hi,

I would have had a look at the list of updates. Anyway at the beginning of the month I started having some issues with Java, here is an example of the header of the error file:

#
# A fatal error has been detected by the Java Runtime Environment:
#
# EXCEPTION_ACCESS_VIOLATION (0xc0000005) at pc=0x000007feef8e19e9, pid=2988, tid=4988
#
# JRE version: 6.0_20-b02
# Java VM: Java HotSpot(TM) 64-Bit Server VM (16.3-b01 mixed mode windows-amd64 )
# Problematic frame:
# C [flgeom.dll+0x3919e9]

Anything similar to yours? In my case I had just to 'get rid of' MS Defender and install Java from Sun.

Cheers

Hi guys!

I am not sure if it's a problem due to the MS OS update! It happens even under Linux OS, especially for long time dependent studies! Actually there are lots of mini-problems.. !!!
However, there are many hints to encounter such problems as JF has proposed! but it's specific to the nature and the type of the study!

This is the cost of making COMSOL more "powerful" by offering various options as recovery files, restart, swap and tmp issues...etc!!

Lets wait for other improvements.... :-)

Cheers!

I repaired my COMSOL installation from DVD and after that I run the Update 2.

Now it seems that COMSOL works ok on VISTA64! For 24 hours I haven't got any crashes!

Michael

Hello,

My COMSOL 4.2 also gets frozen on a LINUX workstation during transient simulations. I have the feeling that it comes from the convergence plots displayed during the simulation. I hope the next patch will solve this issue...

Regards,

Alain Glière

I use Win7 - 64Bit - 8 Core - 64G RAM with 128G (256G capacity) Solid State Disk configured as virtual Memory.
(The project planned to update to 128G Ram soon)

I simulated the case (acoustic time transicent, with it meshed only 6 points per wavelength and total about 4.1M calculation points in 3D (for Case Cement Bond LWD). It crashed the PC to blue screen several times already. The networking was not connected during the simulation in order to avoid any update from company's servers.

The last crash blue screen shows "BAD_POOL_HEADER" after 5 days of running (about 40% done showed on the progress indicator on COMSOL, and it seemed COMSOL worked properly).

The first crash occured just a few hours of running the same case.

The case is identical among those crashes. The difference is that I disabled Win-7 Power save configuration to high performance (no power sleep mode -- that is default set by Win 7?). Otherwise, the PC tuned into sleep mode soon during the simulation and somehow crash happened thereafter (may not be related). At least, the case ran faster. Explicitly reconfigure the virtual memory to the solid state DISK (128G). I did configured it to full before.

It definitely improved the durations from starting simulation to crash. Still I could not finish my case simulation. I guess it would take 10 days for one complete run.

What I observed and adjusted along each crash was:

1. The case uses more than 64G so virtual memory was very active during the simulation. (shown on DISK activities on Win Task Manager). COMSOL memory indication was 63G/107G. I am not sure 107G indicated some peak memory usage. I did not see that kind of memory usage from Task Manager (I monitored it once every 3 hours during the office time)
2. COMSOL also generates some huge temp file on C drive under my log profile folder in AppData folder (the last one was about 21G and updated during the last 5 days of running until the crash).
3. COMSOL can run high speed when everything is in RAM (close to 100% of CPU). But only 30% on average when Virtual memory kicked in. (I tried using 3 points per wavelength or so, the case fiished in 7 hours without crash). I did see the negative corelation in between the CPU usage and Virtual Memory disk activities.
4. I did have COMSOL frozen cases when I use mesh upto 14 points per wavelength or so and it generated about 28M calculation points. It never reached 1% in progress after 24 hours of running (COMSOL core usage is less than 12%). And then PC crashed to BLUE SCREEN. (could be that the virtual memory was full. but i could not check it due to the blue screen)

By the way, I ran similar scale of cases before on XP with 3.5. I did not see such PC crash.

Any Solutions?

I tried one more time. In this time I reduced the number of degrees of freedom to about 2.8M by reduce the mesh density to 5 points per wavelegth. The simulation is finished in 20 hours without crash. The memory usage indicator in COMSOL showed about up to 42G/54G during the simulation. I did not see activity on virtual memory. And CPU uasage is around 100%. My question is what that 54G means? I noticed that Task Manager showed about 41 or alike for COMSOL. So i guessed it was related to 42G.

It seams to be a feature that Windows Task Manager tells that Comsol Multiphysics is "Not Responding", when you make any operations with it. I have Windows 7 64bit OS and 8Gb memory. My COMSOL model file is 19Mb. After 30sec - 2 min Comsol is recovered from the "Not Responding" state and next step can be done. Is this "normal" behaviour?