Path: utzoo!utgpu!jarvis.csri.toronto.edu!rutgers!cbmvax!daveh
From: daveh@cbmvax.commodore.com (Dave Haynie)
Newsgroups: comp.sys.amiga
Subject: Re: Recommendations needed about A2500/30 as research workstation
Message-ID: <9986@cbmvax.commodore.com>
Date: 5 Mar 90 06:35:20 GMT
References: <6623@cps3xx.UUCP> <2017@sauron.Columbia.NCR.COM> <22494@pasteur.Berkeley.EDU> <50518@lll-winken.LLNL.GOV>
Reply-To: daveh@cbmvax (Dave Haynie)
Distribution: usa
Organization: Commodore, West Chester, PA
Lines: 50

In article <50518@lll-winken.LLNL.GOV> jac@muslix.llnl.gov.UUCP (James Crotinger) writes:
>In article <22494@pasteur.Berkeley.EDU> navas@cory.Berkeley.EDU.UUCP (David C. Navas) writes:

>>The '030' is, of course, a faster chip than an equivalently clocked '386.
>>The '486' is, of course, faster yet.
>>David Navas                                   navas@cory.berkeley.edu

>   This seems to be a widespread myth among Motorola product users.
>MIPS magazine (now Personal Workstations) has tested a wide variety of
>'386 machines and a few '030 machines, and the '386 machines always
>come out slightly on top (and in computing power/$ they come out way
>on top (of course MIPS has never tested an A2500/30)). 

Hmm, I read the same magazine, and seem to hear something different.  You sure
you aren't reading through blue-tinted glasses?  I'm not going to suggest one
compare a Mac against the latest 33MHz '386 machine with scads of external 
cache.  But comparing some of the better '030 machines (HPs, Apollos, and the
occasional Sony), the '030 machines have always come out on top running similar
software.  It's a close test, and if you run one of these '386 program launchers
on a '386 with Dhrystone or some other cache-sized benchmark, you get a bit
better performance than an '030 machine under UNIX.  Or, for that matter, a
'386 machine under UNIX.  But for the AIM benchmarks, UNIX vs. UNIX, etc. the
'040 machines come out ahead.  In fact, a recent issue has a 50MHz HP '030
machines outperforming all tested '486 machines on equivalent integer
benchmarks.  The '486 did win on the floating point, but I wouldn't have
expected any '030 to have much of a chance.

>   However, if Motorola's info is accurate, the '040 is faster than an
>equivalently clocked '486.

It's always dependent on the system design.  Only, with these newer chips, the
system design matters less, since you've got more % of performance based only
on chip design, thanks to on-chip floating point and large on-chip caches.

One thing in the favor of '386 machines is that competition has forced 
manufacturers to build better '386 machines at lower prices than similar 
'030 machines.  The HP and Apollo machines carry Workstation prices.  Even
the NeXT and Mac IIci aren't cheap, and of those two, they're both quite
moderate designs -- you can get '386 machines at 33MHz with external cache
for the $8-$10k or so you'll pay for either of these '030 boxes.  You will
generally have to spend a few hundered extra to put a math chip in your
'386 (nearly every '020 or '030 system comes with one).

>   Jim


-- 
Dave Haynie Commodore-Amiga (Systems Engineering) "The Crew That Never Rests"
   {uunet|pyramid|rutgers}!cbmvax!daveh      PLINK: hazy     BIX: hazy
                    Too much of everything is just enough