Hi Brian,
Thanks for your suggestions.
On 5/1/2012, at 4:00 p.m., Brian Radak wrote:
> This won't necessarily help, but what ntc and ntf settings are you using?
ntc and ntf are both set to 2, as befits a run with SHAKE.
> Another often overlooked aspect of SHAKE in AMBER (and minor pet peeve
> since it's usually miscited) is that 3 point waters (maybe all rigid
> waters?) don't actually use the true SHAKE algorithm as other bonds to
> hydrogen, they use the SETTLE algorithm. Now I couldn't tell you in
> detail what that means but I would guess that a rigid water with additional
> bonds/constraints could muck up the math there.
Owing to the importance of getting Amber's TIP3P waters to work, I would hazard a guess that Scott and the others who work on the GPU code dealt with this situation. In his messages to me (on the Amber bugzilla), Scott suggested that the error I was receiving was the result of either a hydrogen bonded to two or more oxygens, or an oxygen bonded to three or more hydrogens. (Of course, what that does to a simulation involving genuine hydronium ions is anyone's guess.) His very specific choice of words further suggests that the "hydrogen bonded to an oxygen and another hydrogen" scenario is OK. I will, of course, accept correction on this point if the GPU code really does choke on rigid TIP3P waters.
Scott also told me, though, that the check for these excessive bonds has only been implemented in his GPU code, to avoid some weird and wonderful errors. This, I think, is why my error is peculiar to the GPU code. I suspect that if I were to run it via CPU using ordinary PMEMD, it would work OK, but I had hoped for the performance boost that GPUs give.
Cheers,
Ben
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Thu Jan 05 2012 - 13:30:04 PST
