The problems you are seeing could be due to a number of reasons, all related
to the mpi installation and compilation.
First of all does your MPI installation work? Can you run some of the simple
ping-pong tests that come with mpi?
Secondly you should verify that the mpi installation was built with the same
version of the compiler as you are using to compile amber. Also check that
the libraries you linked are for the appropriate mpirun that you are using.
I.e. a common error on Linux machines is to link against an custom
installation of mpich but then not modify the users path correctly so that
the mpirun that is used is actually the one in /usr/bin that corresponds to
the default installation of lam. Such issues can lead to some very weird and
often undecipherable errors.
Also are you submitting this through a queuing system? I would start by
getting an interactive login on one of the nodes and just run the tests
there. Make sure you set the environment variable DO_PARALLEL to the correct
value for your system. Although you may run into problems if you have a
version of mpirun that requires several interconnect options etc to be after
the program name. Either way you are probably best speaking to the
administrator of your cluster here and get them to take a look at things.
You should also try creating a simple run yourself for which you have the
single processor answers for (say printed every step for 100 steps) and then
run this and see what happens. Sometime, especially with queuing systems,
there are problems running the scripts that fire up the test cases in
parallel. In this case you should have your queing system give you a set of
nodes for interactive use, most batch queueing systems support such modes,
and then you can set the DO_PARALLEL options and run the makefile
interactively on that node.
All the best
Ross
/\
\/
|\oss Walker
| HPC Consultant and Staff Scientist |
| San Diego Supercomputer Center |
| Tel: +1 858 822 0854 | EMail:- ross.rosswalker.co.uk |
|
http://www.rosswalker.co.uk <
http://www.rosswalker.co.uk/> | PGP Key
available on request |
Note: Electronic Mail is not secure, has no guarantee of delivery, may not
be read every day, and should not be used for urgent or sensitive issues.
_____
From: owner-amber.scripps.edu [mailto:owner-amber.scripps.edu] On Behalf Of
snowyowls
Sent: Thursday, April 13, 2006 01:15
To: amber.scripps.edu
Subject: AMBER:a problem in parallel compiler
dear all
i have compiler amber8 on a linux based claster
and the compiler finished without any error message
but when i run make test.sander
it failed with output here:
cd dmp; ./Run.dmp
This test not set up for parallel
cannot run in parallel with #residues < #pes
cd adenine; ./Run.adenine
This test not set up for parallel
cannot run in parallel with #residues < #pes
==============================
================================
cd cytosine; ./Run.cytosine
0 - MPI_COMM_RANK : Null communicator
[0] Aborting program !
[0] Aborting program!
p0_2419: p4_error: : 197
/export/mpiintel/bin/mpirun: line 1: 2419 Broken pipe
/home/faculty/jmj/Amber8/amber8/test/cytosine/../../e
xe/sander "-O" "-i" "in.md <
http://in.md/> " "-c" "crd.md.23" "-o"
"cytosine.out " -p4pg /home/faculty/jmj/Amber8/amber8/test/cytosine/PI2
331 -p4wd /home/faculty/jmj/Amber8/amber8/test/cytosine
1 - MPI_COMM_RANK : Null communicator
[1] Aborting program !
[1] Aborting program!
p1_31866: p4_error: : 197
./Run.cytosine: Program error
make: *** [test.sander.no_lmod] Error 1
the system administrater told me there is something wrong with my opration
what's wrong?
-----------------------------------------------------------------------
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 14 2006 - 03:25:46 PDT
