Re: [AMBER-Developers] Request for comments on Fortran vs. C versions of blas, lapack, etc.

From: Scott Brozell <sbrozell.rci.rutgers.edu>
Date: Thu, 17 Sep 2009 12:54:00 -0400

Hi,

On Thu, Sep 17, 2009 at 08:05:06AM -0400, case wrote:
> On Thu, Sep 17, 2009, Scott Brozell wrote:
> >
> > But I've heard that not all such libraries support CBLAS, e.g.,
> > ACML. I havent tested this.
>
> I'm a little lost here. I thought that ACML includes its own BLAS, so one
> would not be mixing ACML with CBLAS anyway(?) Does anyone have experience with
> this? Should we be supporting it?


ACML has BLAS, but not CBLAS according to what I heard.
gooogggle found
http://forums.amd.com/devforum/messageview.cfm?catid=217&threadid=89362
but its from 2007.
Anyway, I'm adding pgi and acml to configure.

> > icc -c -Dflex -O2 -DBINTRAJ -o nmode.o nmode.c
> > ../f2c/f2c.h(21): error: _Complex can only be used with float, double, or long double types typedef struct { real r, i; } complex;
> >
> > But the fix for this looks like the f2c.h include was superfluous.
> > One would hope that a C interface would not have such a basic flaw.
>
> I should have made more clear the fact that this whole conversation has come
> up because of portability problems with f2c.h. It happened to be the case
> that nmode.c could be hand-edited to remove the f2c.h dependency, but that
> will not be the case for other codes.
>
> > > > 4. If we stick with C, then some things like nab could still work even for
> > > > machines that don't have a Fortran compiler.
> >
> > This might be nice for DOCK since the dock6 exe does not now need fortran.
> > You say nab but probably mean sff.
>
> Well, nab is a superset of sff. And, although the dock6 exe itself may not
> need fortran, the suite does require a fortran compiler
> to compile mopac, sphgen, etc. The transition of mopac -> sqm will only
> increase the Fortran needs (going from f77 to f90 requirements).

The idea behind sff was the nab-language-less stuff of nab that
might be useful to other programs. So there's some probably small
gain by keeping it to one programming language.

There is a useable subset of DOCK that doesnt need a fortran compiler
although we don't directly support that and it would be an unusual path.
Given gfortran there's no longer much motivation to eliminate fortran
altogether. There is a popular sphgen rewrite in C++.
With Chimera and the Amber score prep changes in 6.3 one can avoid mopac
in DOCK per se.

Scott


_______________________________________________
AMBER-Developers mailing list
AMBER-Developers.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber-developers
Received on Thu Sep 17 2009 - 10:00:02 PDT
Custom Search