[molpro-user] Paralell jobs
wangm9 at cardiff.ac.uk
Tue May 12 15:31:47 BST 2009
Possible checklist might be:
(1)The way you installed Molpro (including the way you built Global
Arrays). Make sure GA works properly firstly.
(2)I/O: make sure the scratch files are written into fast file system
(e.g., local disk). Some low-performance file system (e.g., NFS) could
slow down the calculations significantly. Check settings for TMPDIR,
(3)The way you run molpro. How many processors/cores are there on your
machines, and how many processes do you run molpro? E.g., running molpro
with 4 processes on a machine with 2-cores may take much longer than normal.
(4)The ab initio method you chose. Some methods show good scaling in
parallel, but not for all the methods. Try another method, and see what
Cristina Sanz Sanz wrote:
> Dear all,
> It seems I'm running a mpp version in the 64 bit machine, so the problem
> is not the version of MOLPRO. Any other suggestion?
> Linux-2.6.9-42.0.2.ELsmp/hydra-0-7.local(x86_64) 64 bit mpp
> version DATE: 6-May-09 TIME:
> Thank you,
> On Mon, 2009-05-11 at 15:44 +0100, Manhui Wang wrote:
>> Dear Cristina,
>> In the beginning of output (just after "Checking input..."), it
>> will tell you the information about version and patch level,etc.
>> Best wishes,
>> Cristina Sanz Sanz wrote:
>>> Dear Grant,
>>> I thought I was running the mpp version, is there any way I can check which one I'm running?
>>> Thank you,
>>> -----Original Message-----
>>> From: Grant Hill [mailto:jghill at wsu.edu]
>>> Sent: Fri 08/05/2009 19:36
>>> To: c.sanzsanz at bham.ac.uk
>>> Cc: molpro-user at molpro.net
>>> Subject: Re: [molpro-user] Paralell jobs
>>> A simple suggestion to get things started, but are you sure you're
>>> running the mpp version and not mppx?
>>> Molpro-user mailing list
>>> Molpro-user at molpro.net
School of Chemistry, Cardiff University,
Main Building, Park Place,
Cardiff CF10 3AT, UK
Telephone: +44 (0)29208 76637
More information about the Molpro-user