Re: [AMBER-Developers] Fortran experts, a word of advice please

From: <steinbrt.rci.rutgers.edu>
Date: Wed, 25 Jul 2012 09:38:22 -0400 (EDT)

Hi Jason,

> Try just removing the "<compile=optimized>" from the top of that
> particular
> source file.
>
> What is the performance hit when you do that? (And more importantly, does
> it solve the problem?)

well, performance remains unchanged, sadly the problem does not go away
:-/ I deleted the line and ran make clean && make sander.MPI, that should
have changed the optimization level, right? Strangely, even without the
<compile=optimized> directive, I see the compiler options '-ip -O3 -xHost'
for softcore.F90. How exactly is it decided what source file gets which
option flags? My make-fu seems too weak to understand this.

If I use '-fp-model precise' by hand *only* for the softcore.F90 compile
line, the desynchronisation problem disappears as well.

We could make -fp-precise become a requirement for TI calculations, but it
is bad that all other sander runs would suffer (slightly) as well. I was
hoping that there is an in-file compiler directive like 'use this
additional switch just to compile this source file'

Does something like this exist?

Thomas

Dr. Thomas Steinbrecher
formerly at the
BioMaps Institute
Rutgers University
610 Taylor Rd.
Piscataway, NJ 08854

_______________________________________________
AMBER-Developers mailing list
AMBER-Developers.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber-developers
Received on Wed Jul 25 2012 - 07:00:02 PDT
Custom Search