<html>
<head>
<meta content="text/html; charset=ISO-8859-1"
http-equiv="Content-Type">
</head>
<body bgcolor="#FFFFFF" text="#000000">
On 30/04/2012 3:38 PM, Anirban Ghosh wrote:
<blockquote
cite="mid:CAJqxE7cE8AsRPZ1thfEExMt2d3RgNYLpQ3VNLmVD4J2DTetBMA@mail.gmail.com"
type="cite"><br>
<br>
<div class="gmail_quote">On Sat, Apr 28, 2012 at 10:22 PM, Andrew
DeYoung <span dir="ltr"><<a moz-do-not-send="true"
href="mailto:adeyoung@andrew.cmu.edu" target="_blank">adeyoung@andrew.cmu.edu</a>></span>
wrote:<br>
<blockquote class="gmail_quote" style="margin:0 0 0
.8ex;border-left:1px #ccc solid;padding-left:1ex">
Hi,<br>
<br>
Typically, I use Gromacs 4.5.5 compiled with automatic
threading. As you<br>
know, automatic threading is awesome because it allows me to
start parallel<br>
runs without calling mpirun. So on version 4.5.5, I can start
a job on eight<br>
CPUs using simply the command:<br>
<br>
mdrun -s topol.tpr -nt 8<br>
<br>
However, now I am using a different node on my department's
cluster, and<br>
this node instead has Gromacs 4.5.4 (compiled without
automatic threading).<br>
So, I must use mpirun to start parallel runs. I have tried
this command:<br>
<br>
mpirun -machinefile mymachines -np 8 mdrun -s topol.tpr<br>
<br>
</blockquote>
<div><br>
</div>
<div>I suppose this mdrun executable is not mpi-enabled. Have
you compiled mdrun with "--enable-mpi" option?</div>
<div><br>
</div>
<div>-Anirban </div>
<div> </div>
<blockquote class="gmail_quote" style="margin:0 0 0
.8ex;border-left:1px #ccc solid;padding-left:1ex">
where mymachines is an (extensionless) file containing only
the text "c60<br>
slots=8". (c60 is the name of the node that I am using.)<br>
<br>
I get this error message:<br>
<br>
"Missing: program name. Program mdrun either does not exist,
is not<br>
executable, or is an erroneous argument to mpirun."<br>
<br>
This is strange, because mdrun is, I think, in my path. For
example, if I<br>
type "mdrun -h", I get the manual page for mdrun (version
4.5.4).<br>
<br>
Then I tried the command "which mdrun", and it gave me this
output:<br>
<br>
/usr/local/gromacs/bin/mdrun<br>
<br>
So, next I tried to call mdrun via mpirun using the specific
path for mdrun:<br>
<br>
mpirun -machinefile mymachines -np 8
/usr/local/gromacs/bin/mdrun -s<br>
topol.tpr<br>
<br>
This starts running my simulation, but when I look in "top",
the simulation<br>
is only running on a single CPU; there is only one entry for
mdrun in "top",<br>
and it has only %CPU=100 (not eight different entries for
mdrun, nor one<br>
entry with %CPU=800). Also, the simulation is going at the
speed I would<br>
expect for running on a single CPU -- it is very slow, so I am
convinced<br>
that, as "top" suggests, mdrun is running on only one CPU.<br>
<br>
Strangely, my colleagues are able to run jobs in parallel
using the exact<br>
commands that I described above. So apparently something is
wrong with my<br>
user ID, although there are no error messages (except the
error message<br>
about "Missing: program name" that I described).<br>
<br>
If you have time, do you have any suggestions for other things
that I can<br>
try? Do you think that something could be wrong with my bashrc
file?<br>
</blockquote>
</div>
</blockquote>
<br>
Get a simple MPI test program and prove how you can run it in
parallel. Then worry about GROMACS.<br>
<br>
Mark<br>
</body>
</html>