<HTML>
<HEAD> <META http-equiv=Content-Type content="text/html; charset=us-ascii">
<STYLE title="table borders">.htmtableborders, .htmtableborders td,
.htmtableborders th {border : 1px dashed lightgrey ! important;} </STYLE>
<STYLE type=text/css>html, body { border: 0px; } span.macro, span.macro ul,
span.macro div, span.macro p {background : #CCCCCC;} </STYLE> <STYLE
type=text/css> p{margin-bottom: 0.15em;margin-top:
0.15em;}body{font-family:arial, helvetica, sans-serif;font-size:10pt;};
</STYLE> </HEAD> <BODY><FONT style="FONT-SIZE: 10pt; FONT-FAMILY: arial,
helvetica, sans-serif"> <DIV><BR> </DIV></FONT> <DIV>In God We
Trust</DIV> <DIV>Hello GMX users</DIV> <DIV>I am trying to run one gromacs
simulation in parallel. I have successfully compiled gromacs 4
with mpi support on a cluster with x86_64 architecture with 4 cpus
(intel core 2 Quad 6600).<BR>when I run mdrun with
single-processor ,it works fine and everything is ok, but when I
run the same simulation with 4 cpu, <BR>it finishs again normally and
I don't get any error messages , but my protein breaks and box of sol
becomes deformed.</DIV> <DIV>These are first few lines of my output (on
screen)</DIV> <DIV>NNODES=4, MYRANK=3, HOSTNAME=localhost.localdomain<BR>
NODEID=3 argc=16</DIV> <DIV>NNODES=4, MYRANK=0,
HOSTNAME=localhost.localdomain<BR>NNODES=4, MYRANK=2,
HOSTNAME=localhost.localdomain<BR>NNODES=4, MYRANK=3,
HOSTNAME=localhost.localdomain<BR>NODEID=2 argc=15<BR>NNODES=4, MYRANK=1,
HOSTNAME=localhost.localdomain<BR>NODEID=1 argc=15<BR>NODEID=3 argc=15<BR>
NODEID=0 argc=15</DIV> <DIV> </DIV> <DIV>here is my commands and mdp
file :<BR>commands:</DIV> <DIV>grompp -f sp.mdp -c fprmd.gro -r
fprmd.gro -p n.top -o sp.tpr -n n.ndx -maxwarn 1000</DIV>
<DIV>mpirun -np 4 mdrun -s sp.tpr -o sp.trr -c fsp.gro -g sp.log -e sp.edr
-n n.ndx -N 4</DIV> <DIV> </DIV> <DIV>mdf file:</DIV> <DIV> </DIV>
<DIV>
title
= n.pdb <BR>restraining<BR>
warnings
= 10</DIV> <DIV>
cpp
= /lib/cpp</DIV> <DIV>
define
= -DFLEXIBLE</DIV> <DIV>
constraints =
none</DIV> <DIV>
integrator =
steep </DIV> <DIV>
nsteps
= 10000</DIV> <DIV>
nstcomm
= 1</DIV> <DIV>
comm_mode
= Linear</DIV> <DIV>
comm_grps
= protein </DIV> <DIV>
nstxout
= 250</DIV> <DIV>
nstvout
= 1000</DIV> <DIV>
nstfout
= 0</DIV> <DIV>
nstlog
= 10</DIV> <DIV>
nstenergy
= 10</DIV> <DIV>
nstlist
= 10</DIV> <DIV>
ns_type
= grid</DIV> <DIV>
rlist
= 1.2</DIV> <DIV>
coulombtype = PME</DIV>
<DIV>
rcoulomb
= 1.2</DIV> <DIV>
rvdw
= 1.4</DIV> <DIV>fourierspacing =
0.12</DIV> <DIV>
fourier_nx =
0</DIV> <DIV>
fourier_ny =
0</DIV> <DIV>
fourier_nz =
0</DIV> <DIV>
pme_order
= 4</DIV> <DIV>
ewald_rtol =
1e-5</DIV> <DIV>optimize_fft
= yes<BR>
emtol
= 10.0</DIV> <DIV>
emstep
= 0.01</DIV> <DIV><BR>whould you please help me to overcome this
problem. <BR>Many thanks in advance for your help and your reply.</DIV>
<DIV><BR>Yours truly <BR>Karim Mahnam<BR>Institute
of Biochemistry and Biophysics (IBB)<BR>
Tehran University <BR>P.O.box 13145-1384<BR>Tehran <BR>Iran <BR><A
href="http://www.ibb.ut.ac.ir/">http://www.ibb.ut.ac.ir/</A></DIV></BODY>
</HTML>