[gmx-users] K20 test

Szilárd Páll szilard.pall at cbr.su.se
Thu Apr 11 15:58:55 CEST 2013


Hi,

No, it just means that *your simulation* does not scale. The question
is very vague, hence impossible to answer without more details

However, assuming that you are not running a, say, 5000 atom system
over 6 nodes, the most probable reason is that you have 6 Sandy Bridge
nodes with 12-16 cores each and 24-32 threads supported. If you just
start mdrun with one MPI rank/node it will use 24-32 threads per
process which is way too much for parallel runs. Try running with less
threads per rank and on less nodes and see if that helps.

Cheers,
--
Szilárd


On Thu, Apr 11, 2013 at 3:23 PM, 陈照云 <chenzhaoyun06 at gmail.com> wrote:
> Hi!
>    When I run gromacs-4.6.1 with k20. I meet a question.
>    I have 6 nodes.And each node has one K20.And I use one process on one
> node with one gpu. But the test result shows that the runtime of one node
> is less than that of six nodes.Is the scalability of GPU not good?
> Thanks!
> --
> gmx-users mailing list    gmx-users at gromacs.org
> http://lists.gromacs.org/mailman/listinfo/gmx-users
> * Please search the archive at http://www.gromacs.org/Support/Mailing_Lists/Search before posting!
> * Please don't post (un)subscribe requests to the list. Use the
> www interface or send it to gmx-users-request at gromacs.org.
> * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists



More information about the gromacs.org_gmx-users mailing list