AMBER Archive (2000)

Subject: Re: your mail

From: schayg_at_puskin-a206.sote.hu
Date: Thu Oct 12 2000 - 04:41:46 CDT

It would be the best to get it compile on the Athlon system, try to tweak
around the compiler options. For eg. turn off all pentium specific
optimizations, and get as low opimization level as possible (maybe also
omit these.) I know, this slows down the code, but would enable a rough
test if the problem is really due to bad processor. I had success on AMD
K6-2 3D at 475 Mhz.

-----------------------
Gusztav SCHAY
PhD student
Semmelweis University Budapest
Dept. of Biophysics and Radiation Biology
Budapest
Puskin u. 9.
H-1088
Hungary

phone: +36-1-2662755 #4033
fax: +36-1-2666656
email: schayg_at_puskin-a206.sote.hu
-----------------------

On Tue, 10 Oct 2000, Christian Pilger wrote:

> Dear AMBER-Users,
>
> I tried AMBER 6 on several different plattforms. However under LINUX (Suse
> distribution 6.4, Kernel: 2.2.16) I experienced the following problem:
> while AMBER 6 readiliy compiles and runs on an Intel Celeron 333, it does
> not compile on a new machine with an AMD Athlon 800. Employing the
> executables generated by the Celeron 333 machine, the calculation (a simple
> minimization) proceeds, but after about 50 (not really reproducable) steps
> I get the "RESTARTED DUE TO LINMIN FAILURE". However, exactly the same
> input files give reasonable results on the other platforms, including the
> Celeron 333. Furthermore I frequently experienced crashes of the Athlon
> 800 machine while running AMBER, which never happend on the Celeron 333.
> Has anyone of you experienced similar problems ? Could it be a matter of
> defect hardware ?
>
> Any comments are highly appreciated.
>
> Best regards,
>
> Christian
>
> -----------------------------------------------------------------
>
> Christian Pilger Uni-GH Paderborn
> FB 13 - Organische Chemie
> Warburger Str. 100
>
> D-33098 Paderborn/Germany
> Tel.: 05251-60279/-602183
> Fax : 05251-603245 email: cpilger_at_oc30.uni-paderborn.de
>
> Word Wide Web: http://oc30.uni-paderborn.de/~cpilger
>
> -----------------------------------------------------------------
>
>