QuantumATK Forum

QuantumATK => General Questions and Answers => Topic started by: baizq on January 20, 2013, 09:56

Title: Error: BAD TERMINATION OF ONE OF YOUR APPLICATION PROCESSES
Post by: baizq on January 20, 2013, 09:56
Dear all,

I was running a devce calculation when encountered such a problem: The program was killed right after equivalent bulk calculation and the following error message appeared:

Code
/home/z400/QuantumWise/atk-12.8.0/atkpython/bin/atkpython: line 7: 26207 Killed                  $EXEC_DIR/atkpython_exec $


+------------------------------------------------------------------------------+
|                                                                              |
| Equivalent Bulk  [Finished Sun Jan 20 13:03:08 2013]                         |
|                                                                              |
+------------------------------------------------------------------------------+

===================================================================================
=   BAD TERMINATION OF ONE OF YOUR APPLICATION PROCESSES
=   EXIT CODE: 137
=   CLEANING UP REMAINING PROCESSES
=   YOU CAN IGNORE THE BELOW CLEANUP MESSAGES
===================================================================================



I resubmitted the job and the same error was still there.

I am quite puzzled...Is it due to short of memory?

Thanks,
baizq
Title: Re: Error: BAD TERMINATION OF ONE OF YOUR APPLICATION PROCESSES
Post by: Anders Blom on January 20, 2013, 22:09
Most likely. The error message is from the queue system, not ATK.

Try to make sure you only have one MPI process per node, that minimizes the memory usage (each MPI progress essentially contains a full copy of the calculation). If that is still too much, see http://quantumwise.com/documents/tutorials/latest/MemoryUsage/index.html/ for tips on how to estimate and reduce the memory usage.
Title: Re: Error: BAD TERMINATION OF ONE OF YOUR APPLICATION PROCESSES
Post by: baizq on January 22, 2013, 16:08
Hi Anders,

I followed the tutorial to optimize the memory requirement by avoiding saving the self-energy and reducing the mixing history steps. Now the job runs pretty fine...Thank you for your help.

baizq