Topic Options
#27896 - 07/11/11 04:40 PM Parallel CHARMM for large systems?
Wes Offline
Forum Member

Registered: 06/27/11
Posts: 3
I am trying to set up a replica exchange run with a solvated protein system containing about 130,000 atoms. CHARMM is terminating upon reading the PSF with this error:

*** CHARMM terminated abnormally
*** Last command: read psf card unit 10
*** CHARMM output follows:
CHARMM> read psf card unit 10
MAINIO> Protein structure file being read from unit 10.
PSFRD2> Reading PSF in the expanded format.
TITLE> * GENERATED BY CHARMM-GUI (www.CHARMM-GUI.ORG) ON JAN, 08. 2009.
TITLE> * SETUP PERIODIC BOUNDARY CONDITION
TITLE> * DATE: 7/10/11 1: 9:37 CREATED BY USER: john
TITLE> *

***** LEVEL -1 WARNING FROM <PSFRD2> *****
***** EOF during read
******************************************
BOMLEV ( -1) IS REACHED - TERMINATING. WRNLEV IS 5


This causes MPI processes to die:


---------------------------------------------------------------
MPI_ABORT was invoked on rank 0 in communicator MPI_COMM_WORLD
with errorcode 13.

NOTE: invoking MPI_ABORT causes Open MPI to kill all MPI processes.
You may or may not see output from other processes, depending on
exactly when Open MPI kills them.
---------------------------------------------------------------
---------------------------------------------------------------
mpirun has exited due to process rank 0 with PID 2093 on
node c002.local exiting without calling "finalize". This may
have caused other processes in the application to be
terminated by signals sent by mpirun (as reported here).
---------------------------------------------------------------


However, if I use the same scripts to generate a solvated protein-fragment system with only 30,000 total atoms, my replica exchange setup works without issue. Any ideas why the same setup would work with 30k atoms and not 130k?

Thanks.

Top
#27899 - 07/11/11 05:04 PM Re: Parallel CHARMM for large systems? [Re: Wes]
rmv Online   content

Forum Member

Registered: 09/17/03
Posts: 8314
Loc: 39 03 48 N, 77 06 54 W
I'm not sure this qualifies as a bug report yet, at least for CHARMM itself.

The error messages indicates an error in reading the PSF, which suggests that the number of atoms may not be the only difference between the systems. Have you tested the initial model (PSF+COOR files) independently?

What happens if you try a non-parallel version of CHARMM? What happens with just regular MD instead of replica exchange?
_________________________
Rick Venable
computational chemist


Top
#27902 - 07/11/11 05:17 PM Re: Parallel CHARMM for large systems? [Re: rmv]
Wes Offline
Forum Member

Registered: 06/27/11
Posts: 3
Yes, I have tested the PSF+COOR files. There doesn't seem to be a problem there.

Trying MD with non-parallel CHARMM also works just fine with my larger system.

Top

Moderator:  BRBrooks, bucknerj, chmgr, John Legato