Re: amber-developers: test case failures

From: Robert Duke <>
Date: Thu, 16 Mar 2006 18:53:31 -0700

Guys -
I need to check out the output for Mengjuei's 2/23 pmemd bug posting;
are little things, but easily fixable if indeed the delta is some
descriptive output change I did not track. If anyone is aware of other
changes in the nmr code that I may not have tracked, please speak up.
code is not particularly well maintained or used, as far as I know, on the

pmemd end (I don't plow through the sander nmr source trying to figure out

what has changed all the time).
For the last two bug posting's by Ross, they look like no-op's to me; the
nfft1-3 grid size issue is known; I am more conservative in hitting a
minimum grid density in pmemd than sander is, and also if you compile
with fftw you can use a prime factor of 7. Thus you won't exactly
sander results unless you specify nfft1-3 explicitly. And as far as bad
compiler issues on the sgi altix, we have had that since day 1. We should

note specific versions that have been observed to give grief, though (I
actually not had problems to any extent in the last year myself; there
lots of problems with compiler stability in the early ifort ia64 8.0
releases though).
Regards - Bob

----- Original Message -----
From: "David A. Case" <>
To: <>
Sent: Thursday, March 16, 2006 8:18 PM
Subject: amber-developers: test case failures

>I encourage those of you that are developing/testing Amber 9 to take a
> at this wiki page:
> If there are problems you have encountered that are not there, please
> them;
> If there are questions directed at you (currently Carlos, Bob, Ross,
> M.)
> please see if you can answer them.
> ...thx...dac
> p.s. looks like Wei has made xleap work with 64-bit machines! This
> be in Amber9 CVS in a day or two, and we will post a bugfix for Amber 8
> as well.
Received on Wed Apr 05 2006 - 23:49:40 PDT
Custom Search