Re: [AMBER] gpu and cpu runs

From: Jason Swails <>
Date: Tue, 15 Jul 2014 08:43:55 -0700

On Tue, Jul 15, 2014 at 12:23 AM, Mary Varughese <>

> Sir,
> This is a previos mail,
> Topic: different RMSF between CPU and GPU (Ross Walker)
> if you are using NTT=3 the random number generator on the CPU is
> different from that used on the 3 GPU runs which would account for the
> differences in such short runs.​

> I am getting the same issue (15ns) runs are quite different. i thought
> its due to ig=-1.
> i have used ntt=3;
> So am i doing anything wrong.
> ​​
> could do suggest any other option.

​I think I recall the email you're talking about, but I certainly don't
remember the minute details​ (specifically how different the RMSF was
between the CPU and GPU). These kinds of references to past emails are of
limited use when you don't provide a link to the archived copy.


1. It sounds like you're simply repeating this question, as though you
don't trust the answer from that post. If you have evidence to suggest why
the explanation in that email is not correct, give it and formulate a more
complete question.

2. 15 ns is a very short simulation, and it's highly unlikely that any
measurable properties have converged for any except the smallest of
systems. You have no reason to expect that unconverged simulations will
resemble each other.


Jason M. Swails
Rutgers University
Postdoctoral Researcher
AMBER mailing list
Received on Tue Jul 15 2014 - 09:00:03 PDT
Custom Search