AMBER Archive (2006)

Subject: Re: AMBER: Different results between serial and parallel runs in AMBER 8

From: David A. Case (
Date: Tue Mar 07 2006 - 10:23:21 CST

On Tue, Mar 07, 2006, Ilyas Yildirim wrote:

> But it
> is mind bugging to see different results when inpcrd does not have any
> velocity info in it.

Sigh....I had hoped to have explained the cause of that result in my earlier
e-mail. A program that relies on a pseudo-random number generator needs to
take special steps to make sure that the exact sequence of pseudo-randoms
is maintained, regardless of the number of cpus. (That is, this needs to be
done *if* one wants the "comfort" of having identical results.)


The AMBER Mail Reflector
To post, send mail to
To unsubscribe, send "unsubscribe amber" to