MPI problem: Assertion failed in dapl_module_util.c

MPI problem: Assertion failed in dapl_module_util.c

Hello,

This question has been asked before, but I don't think the poster got an answer. I am trying to compile a user's code. It runs for a while, but then stops with the error message shown in the subject. A bit more detail:

[71:wardlaw187] unexpected disconnect completion event from [0:wardlaw182]
Assertion failed in file ../../dapl_module_util.c at line 2682: 0
internal ABORT - process 71

This is impi 4.0.1, with ifort 12.0.0 20101006, on RHEL/CentOS (the login/compile node is RHEL 5.3, the compute nodes use a slightly modified (by Bull, the vendor) CentOS 5.3). The hardware is a Westmere cluster with Infiniband interconnect.

Any ideas?

Thanks in advance,

Herbert

4 post / 0 nuovi
Ultimo contenuto
Per informazioni complete sulle ottimizzazioni del compilatore, consultare l'Avviso sull'ottimizzazione

Hi Herbert,

It looks like this is rather application issue.
Probably one process dies and and other processes cannot get completion event. Could you please set I_MPI_DEBUG_COREDUMP=1 and start you application (much better if it was compiled with '-g'). It should generate a core file. Take a look at the backtrace and let me know if you think that this is Intel MPI issue.

Regards!
Dmitry

Quoting Dmitry Kuzmin (Intel)
Hi Herbert,

It looks like this is rather application issue.
Probably one process dies and and other processes cannot get completion event. Could you please set I_MPI_DEBUG_COREDUMP=1 and start you application (much better if it was compiled with '-g'). It should generate a core file. Take a look at the backtrace and let me know if you think that this is Intel MPI issue.

Regards!
Dmitry

Dear Dmitry,

I came across the same problem. I tried to set the I_MPI_DEBUG_COREDUMP=1 but no core file dumped so that I could not know the detail of the abortion. The application I run was mpiBLAST, a very famous HPC application of biology.

Can you show more hints? Thank you!

Could you please submit a ticket on the premier.intel.com?
I'm not sure that we should investigate the issue on this forum.

Regards!
Dmitry

Lascia un commento

Eseguire l'accesso per aggiungere un commento. Non siete membri? Iscriviti oggi