Path: utzoo!mnetor!uunet!husc6!bu-cs!bzs
From: bzs@bu-cs.BU.EDU (Barry Shein)
Newsgroups: comp.unix.wizards
Subject: Re: Help us defend against VMS!
Message-ID: <20393@bu-cs.BU.EDU>
Date: 4 Mar 88 17:15:20 GMT
References: <1636@tulum.UUCP> <20268@bu-cs.BU.EDU> <2219@ttrdc.UUCP>
Organization: Boston U. Comp. Sci.
Lines: 40
In-reply-to: levy@ttrdc.UUCP's message of 4 Mar 88 05:38:38 GMT



>As a Death-starian, I heartily second the motion that the UNIX system is
>great :-) -- but NOT with FORTRAN, at least not on plain-vanilla UNIX systems.
>As you said yourself, the key phrase here is "compute intensive."  I.e. you
>need all the crunch you can muster.  Code compiled by /usr/bin/f77 is not all
>that marvelously efficient, to put it mildly.  Of course the Cray people
>don't use /usr/bin/f77 :-) but if you're on a lesser machine it matters.
>(DEC now does sell a FORTRAN compiler for ULTRIX which is as good as its VMS
>version, and co$t$ about as much.)
>-- 
>|------------Dan Levy------------|  Path: ..!{akgua,homxb,ihnp4,ltuxa,mvuxa,

This (I believe, forgive me if I'm wrong) reveals a common fallacy.

Yes, VMS/FORTRAN produces incredible Vax code, I have read it myself
when running benchmarks for people, it is truly an impressive code
generator. And yes, most vanilla F77's produce much less impressive
code (although most Unix C compilers are about as good as VMS fortran,
C presents some difficulties to code generators that Fortran doesn't.)

The point is, however (eg), given $30K for a machine which would you
rather run your code on? A Sun4 at about 10MIPs or a uVax at around
1MIP?  Do you think any code generator will make up for that kind of
difference? You can find similar arguments in every price range (and
price ranges that don't exist in the Vax world, like Crays.)

No code generator can make up for the real problem, locking yourself
into the rather narrow, poor cost/performance of the Vax product.

Give me a mediocre code-generator and real iron any day.

I will say that Unix on a Vax makes the machine plausible again, or at
least leads one into other issues. It's not completely vaxes I am
complaining about (although I haven't bought one in many years, there
was always something better to be had for my money, that could change,
the 3000 series looks interesting), it's locking yourself into Vaxes
for all computing (by choosing VMS) which seems like the losing
strategy.

	-Barry Shein, Boston University