Skip site navigation (1)Skip section navigation (2)
Date:      Wed, 16 Dec 1998 19:19:31 -0500
From:      "Allen Smith" <easmith@beatrice.rutgers.edu>
To:        Steve Kargl <sgk@troutmask.apl.washington.edu>, chuckr@mat.net (Chuck Robey)
Cc:        nate@mt.sri.com, sthaug@nethelp.no, bright@hotjobs.com, bs_13943_34262@adimus.de, hackers@FreeBSD.ORG
Subject:   Re: Fortran in the base system (was Re: sysinstall)
Message-ID:  <9812161919.ZM6960@beatrice.rutgers.edu>
In-Reply-To: Steve Kargl <sgk@troutmask.apl.washington.edu>  "Re: Fortran in the base system (was Re: sysinstall)" (Dec 16,  5:45pm)
References:  <199812162203.OAA75899@troutmask.apl.washington.edu>

next in thread | previous in thread | raw e-mail | index | archive | help
On Dec 16,  5:45pm, Steve Kargl (possibly) wrote:
> [Attributes might be screwed here]
> 
> According to Chuck Robey:
> > On Wed, 16 Dec 1998, Nate Williams wrote:
> > > 
> > > Read what I said.  It is only used by engineers that have already
> > > existing Fortran code.  It doesn't mean new code isn't written, but the
> > > new code that is written tends to be written by folks who already have
> > > written lots of Fortran code.
> > 
> > Actually, besides the mountain of legacy code, it vectorizes (where ANSI
> > C doesn't) onto supercomputers, so academics are often into Fortran.
> > These guys (from my own experience) want big workstations, and aren't
> > really terribly interested in PC-based OSs.  A smallish program to them
> > is 200 megs in size.
> 
> Dual PII 450 MHz with 1 GB memory.  You're hits some serious computing
> power.   The Portland Group sells HPF (high performance Fortran) for
> SMP systems and clusters for linux.  I haven't tried HPF yet, but
> PGI's F90 compiler works under our linux emulation.

As well as engineering, a lot of biochemistry (molecular modelling,
energy minimization, structure prediction, etcetera) code is written
in Fortran. Given the cost of swap space and memory on non-PC
machines, and that energy minimization takes a _lot_ of swap space
(and, preferably, memory) even for small molecules (we've got about a
gig of combined memory and swap space on one of our machines that's
used a lot for this, and it keeps running out...), we've been looking
at the possibility of getting in a FreeBSD-based machine for just this
purpose. A good Fortran compiler would be a necessity for this,
although it might be worth buying something with automatic
parallelization. (With such an automatic parallelizer, an equivalent
of Beowulf for FreeBSD would be quite nice also - it would enable
using lots of much cheaper non-Intel microprocessors, one per
machine.)

	-Allen

-- 
Allen Smith				easmith@beatrice.rutgers.edu
	

To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-hackers" in the body of the message



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?9812161919.ZM6960>