[gmx-users] GPU-Note

[Szilárd Páll]

First off, don't change the cutoff without being sure of what you're doing!

There is one thing you can try: lowering nstlist, the neighbor search
frequency; this is a free parameter and it is picked heuristically. A
larger value will reduce the cost of search but as it will increase
the Verlet buffer and with it increase the non-bonded force compute
cost. Therefore, a smaller nstlist value (e.g. 20 instead of 40) will
about double the search cost, but you should see a reduction in force
compute time. However, expect modest performance benefits.


--
Szilárd


On Mon, Sep 18, 2017 at 9:11 PM, RAHUL SURESH <drrahulsuresh at gmail.com> wrote:
> Is there something that I can do without reducing the cut off..?
>
> On Mon, 18 Sep 2017 at 10:27 PM, Szilárd Páll <pall.szilard at gmail.com>
> wrote:
>
>> It means that there is a certain amount of time the CPU and GPU can
>> work concurrently to compute forces after which the CPU waits for the
>> results from the GPU to do the integration. If the CPU finishes a lot
>> sooner than the GPU, the run will be GPU performance-bound (and
>> vice-versa) -- which is what happens here: the force compute on the
>> GPU takes longer than on CPU and as this exceeds >20%, mdrun notes is
>> so you can consider doing something about it if you can/want to.
>>
>> --
>> Szilárd
>>
>>
>> On Mon, Sep 18, 2017 at 6:39 PM, RAHUL SURESH <drrahulsuresh at gmail.com>
>> wrote:
>> > I receive the following note while doing NVT & NPT.
>> >
>> >
>> > What does it exactly mean?
>> >
>> > I am using charmm36 ff and as per documentation it is necessary to ahve
>> cut
>> > off as 1.2. How can I overcome this note.?
>> >
>> > NOTE: The GPU has >20% more load than the CPU. This imbalance causes
>> >       performance loss, consider using a shorter cut-off and a finer PME
>> > grid.
>> >
>> > --
>> > *Regards,*
>> > *Rahul Suresh*
>> > *Research Scholar*
>> > *Bharathiar University*
>> > *Coimbatore*
>> > --
>> > Gromacs Users mailing list
>> >
>> > * Please search the archive at
>> http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_List before
>> posting!
>> >
>> > * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
>> >
>> > * For (un)subscribe requests visit
>> > https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or
>> send a mail to gmx-users-request at gromacs.org.
>> --
>> Gromacs Users mailing list
>>
>> * Please search the archive at
>> http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_List before
>> posting!
>>
>> * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
>>
>> * For (un)subscribe requests visit
>> https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or
>> send a mail to gmx-users-request at gromacs.org.
>
> --
> *Regards,*
> *Rahul Suresh*
> *Research Scholar*
> *Bharathiar University*
> *Coimbatore*
> --
> Gromacs Users mailing list
>
> * Please search the archive at http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_List before posting!
>
> * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
>
> * For (un)subscribe requests visit
> https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or send a mail to gmx-users-request at gromacs.org.