Re: AMBER: Different result for serial and parallel sander with IREST=1 (AMBER 8) !

From: David A. Case <case.scripps.edu>
Date: Mon, 3 Apr 2006 07:57:36 -0700

On Mon, Apr 03, 2006, Pradipta Bandyopadhyay wrote:
>
> I compiled AMBER 8 in both in serial and parallel mode.
>
> When I tested the serial and parallel for a longer job with the following
> input file:
> upto 400 steps the results are identical (i.e. if we consider 4 digits
> after decimal - as sander prints). However, after that the two simulations
> start to diverge slowly.

This is expected; please see pp. 9-10 of the Amber 8 Users' Manual.

....dac

-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber.scripps.edu
To unsubscribe, send "unsubscribe amber" to majordomo.scripps.edu
Received on Wed Apr 05 2006 - 06:07:05 PDT
Custom Search