>Thanks for the info Ross! :)
>
>Just to clarify. When you say single GPU runs, do you mean that there is
>only one card in the machine, or does this also include two GPUs not
>working together, but performing a solo run each. i.e. not communicating
>with each other.
solo run each.
>
>Also what number exactly counts as a low value of ntpr, ntwr, ntwx, ntwe
>or
>ntwv? Is 250 ok?
This is highly simulation specific - generally the more atoms in your
simulation the smaller these values can be before they effect performance.
My suggestion would be to try it and see. Typically values of 1000 or more
or good or you will create massive files. NTWR should always be set large,
typically so it is triggered only every hour or so.
>
>Additionally, could you let me know what bandwidth value I need to have a
>minimum of to run two GPUs in a machine that are not working together?
You need a minimum of PCI-E x16 for reasonable multi-GPU performance.
Lookup whatever that bandwidth should come out as.
All the best
Ross
/\
\/
|\oss Walker
---------------------------------------------------------
| Associate Research Professor |
| San Diego Supercomputer Center |
| Adjunct Associate Professor |
| Dept. of Chemistry and Biochemistry |
| University of California San Diego |
| NVIDIA Fellow |
|
http://www.rosswalker.co.uk |
http://www.wmd-lab.org |
| Tel: +1 858 822 0854 | EMail:- ross.rosswalker.co.uk |
---------------------------------------------------------
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.
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Mon Jun 10 2013 - 07:30:02 PDT
