Re: amber-developers: pgf90 and netcdf

From: Scott Brozell <>
Date: Sat, 1 Apr 2006 19:48:28 -0700


On Sat, 1 Apr 2006, David A. Case wrote:

> On Sat, Apr 01, 2006, Scott Brozell wrote:
> > I then did this
> > pgf90 -c -o dlamch.o _dlamch.f
> > According to the manual the default is -O1 -tp p7 in this situation.
> > All 9 previous hangers passed.
> To me it sounds as if putting -O1 in FFLAGS would do the trick for
> your platform. Is that what you think?


Here's the test report.
22 possibly failing serial tests.
They all are insignificant.
21 are qmmm2 and 1 is jar.
It looks like the truncation could be increased on the qmmm2 and
would decrease the false positives.

I suppose I should move on to parallel ?

> > I know your position on distributing all dependencies, but
> > for COLUMBUS, I decided that all platforms will have a native
> > blas and lapack. This hasnt failed yet.
> What do you do for these platfoms?
> 1. PIV with G95 compiler
> 2. PIV with ifort compiler, etc.
> Do you require MKL? What happens on AMD chips? This all seems like a
> steep price to pay for Amber, especially since the performance
> are modest.

We support non-mkl, non-acml linux for g77(not compatible with all
programs), g95, ifort, and pgf90 using
LIBS=-llapack -lblas
So we assume they exist and can be found.
Admittedly, I have been surprised that blas and lapack seem to be
(on a small lab, recently aquired machine)
uname -a
Linux zirconium 2.6.9-11.EL #1 Fri May 20 18:17:57 EDT 2005 i686 athlon
locate liblapack.a

(on ~medium iron)
uname -a
Linux piv-login1 2.4.24-pre1 #6 SMP Wed May 19 09:47:10 EDT 2004 i686 i686
locate liblapack.a

> ...thanks for all the testing...dave

You're welcome; I enjoy working in the Amber community.

Received on Wed Apr 05 2006 - 23:49:31 PDT
Custom Search