Workstation AMD Dual Opteron processor SUSE Linux 9.3 crash

[gurmeet2003]

I have been running ABAQUS on a workstation. Operating system is SUSE Linux 9.3. The workstation is a dual processor AMD Opteron machine with 8GB memeory. I have been using ABAQUS on this machine for a couple of years without any problem. However recently my ABAQUS jobs have been ending due to a computer crash. The screen freezes and no files are saved.

The ABAQUS version is 6.6-1. Most of times the crash occurs when an ABAQUS job is running. But a couple of times it has occured when ABAQUS was not running. This is a Linux machine not supported by the IT group. Hard ware testing by software provided by IBM has not indicated any hardware error. Any suggestions will be appreciated.

Thanks,

Gurmeet

 

[SWComposites]

Could be an overheating problem. Open up the box and thoroughly clean it. Then start it up with the cover off and verify that all of the fans are operating. The next suspect is probably the power supply - get a power supply tester and check it out.

 

[dik]

I'd also try to get a utility that stress tests the processor and memory. What type of motherboard does the system use. Is it a built up system or a commercial prefab box?

Dik

 

[Jordonlaw]

Are you now running longer jobs than you were before ? There are some known errors that generally occur after a few days of running analyses in my experience. The classic one is due to the 'gam_server' script which forced me to move to XP64 with no problems so far.

 

[Stringmaker]

All good suggestions so far! Check that that version of Abaqus supports your OS version. I think it should I'm not certain though.

Also, this may sound stupid but it's the cause of so many problems: Graphics Drivers. Be sure you have the recommended drivers for you GFX card using that version of Abaqus.

Let us know what you come up with.

-Brian

 

[gurmeet2003]

Thanks for all the suggestions. It took a long time to resolve the problem. Memory tests by the IBM software passed the memory. So they changed a number of parts but not the memory. The probelm was not solved. Subsequently our IT guys changed the memory for test (ignoring the advise of the software). Computer began to run ok in the test room which is mantained at a cool temperature (60 def F). However when we moved the computer into my office, which is warmer it began to crash again.

There are two fans in the computer. One of the fans was set up by IBM as a standby. It used to kick on and off. Our IT guys changed the setup to run the standby fan continuously. This solved the problem and computer is now running satisfactorily.

There were two problems: 1. Bad memory 2. thermal problem.

I would suggest that one should not pay any attention to the memory troubleshooting software. It is worthless. Also it is not clear to me why the computer developed a thermal problem after running ok for 1-1/2 years.

Gurmeet