Path: utzoo!utgpu!news-server.csri.toronto.edu!rpi!zaphod.mps.ohio-state.edu!swrinde!elroy.jpl.nasa.gov!aero-c!gumby.dsd.trw.com!deneva!news
From: thomsen@spf.trw.com (Mark R. Thomsen)
Newsgroups: comp.sys.next
Subject: Re: New HP machines No News for NeXT??
Message-ID: <28485109.67E3@deneva.sdd.trw.com>
Date: 2 Jun 91 01:59:36 GMT
References: <1991Jun1.003344.2109@potomac.ads.com>
Sender: news@deneva.sdd.trw.com
Organization: TRW Inc., Redondo Beach, CA
Lines: 41

John T. Nelson writes
  Shame it can't outperform a SPARC though :-)

But, can you, using a 68040 NeXT, get more work done or write
a comparable program faster than the guy using the SPARCstation?
Do you feel better about the quality of your work?

One more thought on this thread. My group is starting to evolve
a strategy towards computers that assumes there will be multiple
processors in a computer system. The NeXTcube w/ NeXTdimension
is an example - you have a 68040, a 56001, and an i860 in one
box. The CPU chips are swallowing up the computer board. Given
this we catagorize processors into

   the flexible master - runs the control program, coordinates,
    has the general event/tasking loops, and is mandatory

   the powerful but intelligent slaves - runs more dedicated
    programs, has more arithmetic performance and probably more
    bandwidth than the master, and is closer to a peripheral
    including the screen

   the idiot savants - more algorithmically speciallized and
    tends to run small programs or operates based on parameter
    settings

In a NeXT you have the 68040 master, the i860 'intellislave',
and the 56001 savant. The C-Cube would have been (will be?) a
savant. Because of this increase in the number of programmable
components in the computer system, and that i860s and DSP
chips will evolve in parallel to more standard CPU
architectures, and that the distributed approach allows
superior system performance, our excitement over a new hot
chip is a bit cooler than years past. The CPU might do
an FFT faster than another CPU, but wouldn't you rather have
that done on a DSP that will still outperform the fastest
CPUs for those tasks? In a stereo, unless your preamp is
bad, buying a better preamp might not make the result much
better - a system assessment.

Mark R. Thomsen