AMBER Archive (2004)

Subject: AMBER: Leap & x86_64 / PGI (was: xleap solvatebox/oct problem)

From: Hannes Barsch (hbarsch_at_physik.uni-bielefeld.de)
Date: Mon Nov 29 2004 - 11:02:15 CST


Dear Jiten, David and others,

I have some difficulties getting tleap to work on a dual opteron machine
with PGI fortran compiler installed. As David suggested (below) I compiled
tleap using gcc (3.2.3.20030502) what did not result in a correct working
tleap executable, either. How did you compile tleap with the gcc compiler,
Jiten? What configure options did you use?

I tried to generate a config.h by running './configure -opteron ifort' as
just changing pgcc to gcc in config.h resulted in linker problems. I used
this config file for tleap only, changing the include in line 1 in
src/leap/src/leap/Makefile to match the 'ifort' file. The compilation runs
fine but the executable (exactly as the pgcc compiled executable) shows
strange behaviour when using the solvatebox command: Compared to the result
on a P4 desktop with Intel compiler the waterbox is same size in x but much
smaller in y and z direction. Furthermore on the Opteron system, there are
almost twice as many water molecules pushed into the smaller box what leads
to the density being 4.5 as high as on the desktop system.

Thanks for any hint!

Regards,

Hannes

-----Ursprüngliche Nachricht-----
Von: owner-amber_at_scripps.edu [mailto:owner-amber_at_scripps.edu]Im Auftrag
von Jiten
Gesendet: Freitag, 16. Juli 2004 11:07
An: amber_at_scripps.edu
Betreff: Re: AMBER: xleap solvatebox/oct problem

Dear Prof. David,

It works well with tleap when I compile with gcc. But I am still unable sort
out the xleap problem.

csm43|/theo1/tmp/jiten:821> -I: Adding /theo1/tmp/amber8/dat/leap/prep to
search path.
-I: Adding /theo1/tmp/amber8/dat/leap/lib to search path.
-I: Adding /theo1/tmp/amber8/dat/leap/parm to search path.
-I: Adding /theo1/tmp/amber8/dat/leap/cmd to search path.
Segmentation fault (core dumped)

Sincerely,

Jiten

----- Original Message -----
From: "David A. Case" <case_at_scripps.edu>
To: <amber_at_scripps.edu>
Sent: Friday, July 16, 2004 2:02 AM
Subject: Re: AMBER: xleap solvatebox/oct problem

> On Fri, Jul 16, 2004, Jiten wrote:
> >
> > Here is the similar problem with lteap installed in Opteron compiled
with
> > PGI. I find that the water molecules are far away from the UNIT. Any
> > suggestions to fix this problem will be highly appreciated.
>
> There have been both public and private reports of problems with LEaP
> compiled with pgcc. Can you try compiling with gcc instead?
>
> (Work in ongoing to find the pgcc problem, but there is nothing to report
> yet. And, this might not be your problem, but the symptoms sound
similar.)
>
> ....thanks...dac

-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber_at_scripps.edu
To unsubscribe, send "unsubscribe amber" to majordomo_at_scripps.edu