Dear Dr. Case,
Thank you very much for your patch. I succeeded in compiling parallel
version of NAB on two platforms after applying this patch. However, I am
not able to run a program yet. The test program is
molecule m;
float x[2000], f[2000], v[2000];
float dgrad, fret;
float t1,t2;
int ier, mytaskid, rank, size;
file fp;
mytaskid = mpiinit( argc, argv, rank, size );
fp = fopen("gbrna_long.traj","w");
m = getpdb("gbrna.pdb");
readparm(m, "gbrna.prmtop");
mm_options( "ntpr=100, ntwx=100, gb=1, kappa=0.10395, cut=99.0, diel=C,
tempi=300., rattle=0");
mme_init( m, NULL, "::Z", x, fp);
setxyz_from_mol( m, NULL, x );
t1 = second();
ier = md(3*m.natoms, 1000, x, f, v, mme );
t2 = second();
printf( "md returns %d; elapsed time was %8.3f\n", ier, t2-t1 );
putxv( "gbrna_long.x", "rattle md", m.natoms, 0.0, x, v );
fclose(fp);
mpifinalize();
I compiled it with nab the same way as serial code.
On the lammpi machine, the output message is
MPI_Init: unclassified: MPI already initialized (rank 3, MPI_COMM_WORLD)
Rank (1, MPI_COMM_WORLD): Call stack within LAM:
Rank (1, MPI_COMM_WORLD): - MPI_Init()
Rank (1, MPI_COMM_WORLD): - main()
Rank (0, MPI_COMM_WORLD): Call stack within LAM:
Rank (0, MPI_COMM_WORLD): - MPI_Init()
Rank (0, MPI_COMM_WORLD): - main()
Rank (2, MPI_COMM_WORLD): Call stack within LAM:
Rank (2, MPI_COMM_WORLD): - MPI_Init()
Rank (2, MPI_COMM_WORLD): - main()
Rank (3, MPI_COMM_WORLD): Call stack within LAM:
Rank (3, MPI_COMM_WORLD): - MPI_Init()
Rank (3, MPI_COMM_WORLD): - main()
While on the openmpi machine,
warning:regcache incompatible with malloc
warning:regcache incompatible with malloc
[node060:20875] *** An error occurred in MPI_Init
[node060:20875] *** on communicator MPI_COMM_WORLD
[node060:20875] *** MPI_ERR_OTHER: known error not in list
[node060:20875] *** MPI_ERRORS_ARE_FATAL (goodbye)
[node060:20876] *** An error occurred in MPI_Init
[node060:20876] *** on communicator MPI_COMM_WORLD
[node060:20876] *** MPI_ERR_OTHER: known error not in list
[node060:20876] *** MPI_ERRORS_ARE_FATAL (goodbye)
Please help me out.
Thanks a lot,
Shaowen
-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber.scripps.edu
To unsubscribe, send "unsubscribe amber" to majordomo.scripps.edu
Received on Fri Apr 18 2008 - 21:17:59 PDT
