On Thu, Mar 26, 2015, Glenn Carrington [bs10g3c] wrote:
> I initially used the Langevin thermostat as per the AMBER tutorial, but
> it took ages to get 100 steps on the GPU. Switched to Berendsen and
> the GPU simulation ran noticeably faster. I have not got the timings
> to hand, but I'm guessing a x2 speed increase at least. However,
> as mentioned I don't think I can use this thermostat with implicit
> solvent. As the structure has 47,000 atoms, do you think the difference
> between Anderson and Langevin will be more pronounced due to the
> system size? I'm hoping so, as the simulation will be painfully slow
> otherwise.
Be sure to look at the benchmarks shown on the Amber web page. The nucleosome
GPU values (closest to yours) are for 25,000 atoms. Doubling the number of
atoms is likely to increase execution time by a factor of 4, and you should
see if your timings are in rough agreement with that estimate. It is
certainly true that Amber's GPU implementation of GB is not optimal for large
numbers of atoms.
I think the Anderson thermostat should be fine for GB, but I have not ever
used it that way. But my GB runs are usually on much smaller systems.
....dac
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Thu Mar 26 2015 - 10:00:07 PDT
