*Dear AMBER family,*
* *
I’m trying to install Amber10 on a cluster composed of eight machine. I have
installed Sander and Sander.MPI, and pmemd successfully on each node. I can
employ sander.MPI with no problem. But when I try to use pmemd, for example
in the form of “mpiexec -n 2 pmemd -O -i min-igb.in -o min-igb.out -p
dna-igb.prmtop -c dna-igb.inpcrd -r min-igb.rst -x min-igb.mdcrd” to run
parallel, the optimization does not start and I get the following errors:(
Fatal error in MPI_Bcast: Internal MPI error!, error stack:
MPI_Bcast(786)........: MPI_Bcast(buf=0xd6e008, count=457040, MPI_BYTE,
root=0, MPI_COMM_WORLD) failed
MPIR_Bcast(230).......:
MPIC_Send(36).........:
MPIDI_CH3_RndvSend(70): failure occurred while attempting to send RTS packet
(unknown)(): Internal MPI error![cli_0]: aborting job:
Fatal error in MPI_Bcast: Internal MPI error!, error stack:
MPI_Bcast(786)........: MPI_Bcast(buf=0xd6e008, count=457040, MPI_BYTE,
root=0, MPI_COMM_WORLD) failed
MPIR_Bcast(230).......:
MPIC_Send(36).........:
MPIDI_CH3_RndvSend(70): failure occurred while attempting to send RTS packet
(unknown)(): Internal MPI error!
It is to be noted that pmemd works fine when I run it on two cores of one
machine. However when I try to run it on two or more cores of different
machines, it crashes.
System/software configuration of each machine is:
Intel 1.8 GHz Intel Core 2 Duo
10/100 Ethernet switch
Fedora 12 Linux
Python 2.5
MPICH2-1.0.8
Intel FORTRAN Compiler 11.1
Intel C++ Compiler 11.0
I patched Amber with the last bugfixes in the website and I configured
sander and pmemd with the support of MKL, respectively, as follows:
./configure -mpich2 ifort
./configure linux_p4 ifort mpich2
I searched very much, but couldn't find any solution for this problem.
Any suggestions would be greatly appreciated. Thanks in advance.
Ali
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Mon Aug 02 2010 - 01:00:03 PDT
