[gmx-users] mpi problem]
Ruben Martinez Buey
ruben at akilonia.cib.csic.es
Fri Mar 21 23:50:57 CET 2003
Hi all,
I´ve still have a problem with mpi. I can run mdrun with only one
processor (np 1), but not with more than one (np > 1). This is the
error:
Reading file prueba.tpr, VERSION 3.1.4 (double precision)
Reading file prueba.tpr, VERSION 3.1.4 (double precision)
Steepest Descents:
Tolerance = 1.00000e-02
Number of steps = 100000
MPI: MPI_COMM_WORLD rank 2 has terminated without calling MPI_Finalize()
MPI: Received signal 10
____________________________________________________________________________
And the otuput files are:
core.13520690
ener.edr (empty)
md0.log:
Log file opened: nodeid 0, nnodes = 8, host = jen50, process = 13440144
:-) G R O M A C S (-:
Go Rough, Oppose Many Angry Chinese Serial killers
... ... ... ...
CPU= 0, lastcg= 606, targetcg= 2409, myshift= 4
CPU= 1, lastcg= 1243, targetcg= 3046, myshift= 5
CPU= 2, lastcg= 1859, targetcg= 57, myshift= 6
CPU= 3, lastcg= 2219, targetcg= 417, myshift= 5
CPU= 4, lastcg= 2551, targetcg= 749, myshift= 5
CPU= 5, lastcg= 2883, targetcg= 1081, myshift= 4
CPU= 6, lastcg= 3215, targetcg= 1413, myshift= 4
CPU= 7, lastcg= 3604, targetcg= 1802, myshift= 3
nsb->shift = 6, nsb->bshift= 0
Listing Scalars
nsb->nodeid: 0
nsb->nnodes: 8
nsb->cgtotal: 3605
nsb->natoms: 7963
nsb->shift: 6
nsb->bshift: 0
Nodeid index homenr cgload workload
0 0 996 607 607
1 996 995 1244 1244
2 1991 996 1860 1860
3 2987 995 2220 2220
4 3982 996 2552 2552
5 4978 996 2884 2884
6 5974 996 3216 3216
7 6970 993 3605 3605
parameters of the run (nodeid=0):
... ... ... ...
and md1:
Log file opened: nodeid 1, nnodes = 8, host = jen50, process = 13525852
, md2, md3,...md7.log
Thanks again for your kind attention,
with best regards,
Ruben
More information about the gromacs.org_gmx-users
mailing list