T.R | Title | User | Personal Name | Date | Lines |
---|
3827.1 | ...and f77 begat f90... | HDLITE::SCHNEIDER | whatever # of VPs it takes | Wed Apr 26 1995 14:39 | 20 |
| Hmm, I should probably let our fearless moderator Steve L. answer,
since he's right in the thick of F90 development. But I think the short
answers are:
F90 is used for the same kinds of applications that FORTRAN has always
been an beautifully appropriate language for; mostly, numerically
intensive programs used in engineering and science.
It's brought to you by those fine folks in whatever Technical Languages
and Environments is called these days.
I read a tone of scepticism in your note, and I think I understand
where that might come from. But, I think if you looked into it with an
open mind, you would likely find that this is a good business for us to
be in. If I misread, and you're just curious about what F90 is, see
TLE::F90_ALPHA_OSF (which seems to cover OpenVMS, too, despite the
name). KP7'll do ya.
Semper FORTRAN,
Chuck
|
3827.2 | Thanks for the cue... | QUARK::LIONEL | Free advice is worth every cent | Wed Apr 26 1995 16:48 | 37 |
| Fortran is still the world's most popular programming language for
scientific and engineering applications. It is also the language
programmers reach for when they want the highest possible performance
for their mathematical applications. Consider this - every system
vendor which promotes their system's SpecFP performance uses Fortran to
get it.
Fortran 90 is the most recent revision of the Fortran language
standard, adding many features which have been long requested by
Fortran users - many of which were implemented as extensions to the
Fortran 77 language. Digital is ranked the world's leading Fortran
vendor (in dollar sales) by Dataquest - our VAX FORTRAN (now DEC
Fortran) is undeniably the "gold standard" of the Fortran industry -
all the other vendors tout "VAX FORTRAN compatibility" (where they
can). Digital's Fortran compilers have an excellent reputation for
features, performance and quality - we are often asked by users of
non-Digital systems if we could bring out DEC Fortran on their
platform.
We introduced DEC Fortran 90 for Digital UNIX last summer, and it has
quickly gained a position of respect in the marketplace. What's new is
DEC Fortran 90 for OpenVMS Alpha, a product we have had great demand
for (it seems almost daily I get calls and mail messages from customers
and sales staff asking if/when we'll have such a product.) We
submitted the VMS version to the SSB yesterday - it will ship on the
"May" (read June) ConDist CD.
DEC Fortran and Fortran 90 is developed by the "System Compilers and
Environments" group headed by Bill Blake, part of the Core Technologies
Group.
Fortran has, as long as I've been around, been extremely profitable for
Digital. If you have any questions about Fortran or Digital's
Fortran products, feel free to send them to me or ask in any of our
assorted notes conferences.
Steve (DEC Fortran development)
|
3827.3 | High Performance Fortran also part of DEC Fortran 90 | HPCGRP::JHARRIS | | Thu Apr 27 1995 11:25 | 33 |
| Another significant difference between DEC Fortran 90 and every other Fortran
compiler in the world is that DEC Fortran 90 includes full High Performance
Fortran (HPF), a set of extensions designed for writing portable parallel
programs. The compiler can parallelize HPF programs for clusters of Alpha
workstations, for SMP Alpha systems, and for clusters of Alpha SMPs. This
parallel HPF capability is currently available only on systems running Digital
Unix.
The High Performance Fortran language was produced by the High Performance
Fortran Forum, a collection of application programmers, users, vendors, and
universities that includes all major computer vendors and good representation
from all of the other groups. Digital was instrumental in driving the HPF
language work.
DEC Fortran 90 is the first full HPF compiler in the world (*), and the first
native HPF compiler in the world (other products are source-to-source
translators that only handle a small subset of HPF). This is truly
industry-leading technology; everyone else (IBM, SGI, HP, Sun, Cray, Convex) has
to play catch-up.
The HPF work was done in the High Performance Computing group, part of Bill
Blake's System Compilers and Environments group.
Anyone with questions about HPF can contact me directly, or post a note here or
in one of the Fortran notes conferences. We are also working on WWW pages for
DEC Fortran 90 and its HPF capabilities, which (when done) will be available
from the Digital home page.
Jonathan Harris
HPF Project Leader
(*) Full HPF except for transcriptive argument passing and dynamic remapping.
|
3827.4 | | QUARK::LIONEL | Free advice is worth every cent | Thu Apr 27 1995 11:47 | 12 |
| A caveat to what Jonathan said - HPF support is available on
Digital UNIX only at this time. The parallelization support code was
written specifically for what we used to call "OSF/1 workstation
farms" connected by a Gigaswitch. OpenVMS doesn't yet have equivalent
features - or even kernel threads, though I understand that is in the
works.
I wrote a small paper on the history and future of Fortran a couple
of months ago at the request of a customer - I'll dig it out and
post it here.
Steve
|
3827.5 | | TOOK::MORRISON | Bob M. LKG1-3/A11 226-7570 | Thu Apr 27 1995 13:06 | 4 |
| Thanks for the info so far.
I am delighted to hear that Fortran is still widely used and is highly pro-
fitable for Digital. Fortran is my favorite programming language and I would
hate to see it become obsolete.
|
3827.6 | | ASABET::YANNEKIS | | Thu Apr 27 1995 15:40 | 5 |
|
Why would wecall our new Fortran ... Fortran 90 ... instead of Fortran
95 or some other name that doesn't make it sound 5 years old?
|
3827.7 | | HDLITE::SCHAFER | Mark Schafer, Alpha Developer's support | Thu Apr 27 1995 16:51 | 4 |
| Standards! Folks that care about them will know that our Fortran
meets the ANSI X3.198-1992 Fortran 90 standard.
Mark
|
3827.8 | | MU::porter | now with less than 1% vms | Fri Apr 28 1995 13:53 | 7 |
| >
> Why would wecall our new Fortran ... Fortran 90 ... instead of
Fortran
> 95 or some other name that doesn't make it sound 5 years old?
"We" didn't name the language anything. We simply
used the name that the language already has.
|
3827.9 | | ATLANT::SCHMIDT | E&RT -- Embedded and RealTime Engineering | Sat Apr 29 1995 16:53 | 10 |
| Greg:
> Why would wecall our new Fortran ... Fortran 90 ... instead of Fortran
> 95 or some other name that doesn't make it sound 5 years old?
"<anything> 95" is going to have a bad taste in the customers'
mouths when a certain major operating system doesn't ship in
August of '95 as promised.
Atlant
|