Have you tried actually running 32 mpi threads on your 16 cores? My bet is it probably works pretty well costing you only a few percent in performance. Changing the code to allow you to run say 32 replicas with 16 mpi threads instead of requiring the current minimum of 32 is really not going to help you very much. You cannot magically make the calculations faster by doing this. I would just run with 32 threads on your 16 cores and accept that in this case you are only getting say 95% of the performance you could get if you spent ages unnecessarily overcomplicating the code.
All the best
Ross
On Aug 24, 2011, at 4:26, Dmitry Mukha <dvmukha.gmail.com> wrote:
> Yes it is what I mean. For me, I'd like to run REMD on 1 machine, but if it
> was implemented for serial execution, it would be easy to add scalability
> allowing to run, let say, 16 thread handling 32 replicas (some integer
> multiplier of replicas, 2 or more). It would be certainly faster than
> distributed computing may do this task.
>
> 2011/8/23 Carlos Simmerling <carlos.simmerling.gmail.com>
>
>> it's possible he means normal REMD, but run serially. no reason they need
>> to
>> run simultaneously- I think Pande may have done something similar. it's
>> just
>> going to take a very long time in wallclock. also the code is not set up
>> for
>> this at present.
>>
>>
> --
> Sincerely,
> Dmitry Mukha
> Institute of Bioorganic Chemistry, NAS, Minsk, Belarus
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Tue Aug 23 2011 - 14:30:02 PDT
