Re: Ada on supercomputers (Was: Re: federal requirements to use ada)

[network.ucsd.edu!sdd.hp.com!news.cs.indiana.edu!lynx!spectre.unm.edu!john @nosc.mil (John Prentice)]

In article <1992Jan15.154220.17182@freedom.msfc.nasa.gov> robichau@lambda.msfc.
nasa.gov writes:
>In <g_hg5xf@lynx.unm.edu> john@spectre.unm.edu (John Prentice) writes:
>
> >Speaking for the numerically intensive government world however, I would 
> >have to say that Ada is a non-starter.  I know of no major numerical codes
> >being written in Ada and there is a virtual dirth of compilers for any 
> >of the supercomputers (are there ANY?).  Some lip service is paid to these
> >"mandates", but the only people I know doing so are officials who are
> >at best distantly connected to the real world of computing.  None of this
> >is a statement about the pros or cons of Ada, it is just an observation
> >of the current state of numerical computing.  I don't see this changing
> >in the future.
>
> Cray has a rather nice compiler for their supermachines; release 2.0 does
> some automatic vectorization, and my understanding is that its runtime
> performance is quite close to their C and FORTRAN compilers.
>
> I believe that Concurrent and Convex also have available Ada compilers for
> their machines, and I vaguely recall that Amoco and Exxon were investing
> heavily in doing some numerics work on Convex machines in Ada.
>
> Of course, there aren't many tools of any type available for machines with
> such a small installed base.
>

I don't wish to suggest (and didn't in my original post) that Ada is not
being used at *all* in the numerically intensive world, simply that it is
not *significantly* used and I don't think ever will be.  I was not aware
of the supercomputer compilers, though this doesn't change my basic point.
Ultimately time will tell whether I am right or not, all I am reporting
is my perspective from deep within the trenches of that community. 

John
-- 
John Prentice                   "I would rather be climbing"
john@spectre.unm.edu