Hi,
On Mon, Jun 5, 2017 at 7:39 AM, Vidhya Srivatsan
<vidhyasri1609.gmail.com> wrote:
> 3. MPI / CUDA is soime sort of heterogeneous computing where some processes
> are done in CPU and some in GPU
I'd just like to clarify this point a bit. In the MPI/CUDA version the
majority of the calculation is still done on the GPU, as with the
serial CUDA version. The difference is that you can spread the
calculation across GPUs (peer-to-peer) and/or run multiple instances
on different GPUs for e.g. ensemble methods like REMD, where each
replica can have it's own GPU/GPUs.
Hope this helps,
-Dan
>
> Again, this is a general overview of CPU, GPU jobs and my reply is not
> specific to AMBER
>
>
>
> On Mon, Jun 5, 2017 at 12:47 AM, Robert Wohlhueter <
> bobwohlhueter.earthlink.net> wrote:
>
>> Am I right in concluding that the CUDA version (of pmemd) uses a
>> cuda-device, but not multiple cores? That the MPI version used multiple
>> cores, but not a cuda-device? And that the MPI-CUDA version will work
>> only if there are multiple cuda-devices avaiable?
>>
>> Bob Wohhueter
>>
>> _______________________________________________
>> 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
--
-------------------------
Daniel R. Roe
Laboratory of Computational Biology
National Institutes of Health, NHLBI
5635 Fishers Ln, Rm T900
Rockville MD, 20852
https://www.lobos.nih.gov/lcb
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Mon Jun 05 2017 - 05:30:02 PDT
