Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!tut.cis.ohio-state.edu!ucbvax!mtxinu!shore
From: shore@mtxinu.COM (Melinda Shore)
Newsgroups: comp.os.misc
Subject: Re: Re^2: Unix bigotry
Message-ID: <760@mtxinu.UUCP>
Date: 23 Feb 89 00:25:18 GMT
References: <4434@freja.diku.dk> <5900004@hpfcdc.HP.COM>
Reply-To: shore@mtxinu.UUCP (Melinda Shore)
Organization: mt Xinu, Berkeley
Lines: 23

In article <5900004@hpfcdc.HP.COM> marc@hpfcdc.HP.COM (Marc 'Sphere' Sabatella) writes:
>If you are going to shell out the major $$$ for a Cyber 200, ETA-20, or Cray 2,
>you don't want to saddle it with an operating system that is not going to take
>full advantage of the hardware.

There are several errors in your assumptions.  First, "Unix" does not
necessarily mean "interactive."  NQS, the batching system provided with
Unicos and other large Unix systems, is surprisingly effective.  There
are sites that run their Crays under Unicos but batch-only.  Secondly,
most supercomputer users run enormous Fortran applications that do
very few system calls, so that compiler quality is more of an issue than
operating system performance.  (In the case of Cray machines, the compiler
used under Unicos is the same as that used under COS.)  Thirdly, again
in the case of Cray machines, Unicos performs very much like COS but
actually provides somewhat *faster* i/o.

Because the way in which large-scale scientific computers are used is
so different from the way in which smaller more conventional machines
are used that it is certainly worth rethinking the criteria used to talk
about appropriateness of operating systems.
-- 
Melinda Shore                                     shore@mtxinu.com
Mt Xinu                                      ..!uunet!mtxinu.shore