Hello,
I have a protein system in explicit water of around 56000 total atoms.
Using the latest Amber 12 updates, I found that the Andersen thermostat
(ntt=2) is slower by about 1 ns/day compared to Langevin (ntt=3). The runs
are on a single GPU using SPFP precision. But this page (
http://ambermd.org/gpus/) says that Langevin should be slightly slower, so
I'm wondering about this difference. For Andersen, I'm using vrand=1000,
which should be reasonable (it themostats to the target temperature fine)?
Thanks,
Tim
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Wed Sep 05 2012 - 10:30:02 PDT
