-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Hello Matt--

> I have installed the most recent version of xplor (2.27) on a 64bit
> Ubuntu 10.10 system  running on an Intel i5core 460M processor.  I
> have 2cores and 4 threads to work with for running my jobs in parallel
> (each thread 2.53GHz)l.  When I use the command "xplor -smp #
> xplorscript.inp -o xplorscript.out" (no quotes) I don't see an
> improvement in my run time.  The command gets all the threads to fire
> with no error messages but I am not seeing an improvement over running
> just 1 thread at a time.  I have been testing my installation with the
> the anneal.inp example found in /eginput/protG. For 1 and 2 threads
> (-smp 1 or 2) the run time is about 2.5min for all 4 threads (-smp 4)
> the runtime is about 6min.  It appears that each processor is working
> on each structure rather than each processor working on a different
> calculation. When the job is complete I have redundant .pdb files for
> each calculation.   
> 

For the old XPLOR interface, you must explicitly add parallelization
logic to each script. An example is given in eginput/protG/anneal.inp in
the Xplor-NIH distribution. For the Python interface, this is handled
transparently by the StructureLoop class.

best regards--
Charles
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.10 (GNU/Linux)
Comment: Processed by Mailcrypt 3.5.9 <http://mailcrypt.sourceforge.net/>

iEYEARECAAYFAk1AV4MACgkQPK2zrJwS/laS5ACghfr7jo22CMn9Axrlxv4OS814
LxAAnisXa9tD5sux1PUY0I0IGSfXX75p
=d4ps
-----END PGP SIGNATURE-----

Reply via email to