Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!iuvax!rutgers!cbmvax!daveh
From: daveh@cbmvax.UUCP (Dave Haynie)
Newsgroups: comp.sys.next
Subject: Re: A kind word...
Message-ID: <8726@cbmvax.UUCP>
Date: 27 Nov 89 23:21:08 GMT
References: <21597@brunix.UUCP>
Distribution: usa
Organization: Commodore Technology, West Chester, PA
Lines: 49

in article <21597@brunix.UUCP>, rca@brunix (Ronald C.F. Antony) says:

> there are a few things to say:
> a) "It's too slow"

> it is not that much slower than a sparcstation with X, although
> many say NeXT's window system is slow. It may have less MIPS, 
> but when a lot of programs run, a RISC machine comes down in
> speed pretty soon (context-switching). 

Depends on what you consider slow.  The SparcStation certainly
has a larger CPU context than anything 68030 based; SPARC is a
register-window RISC.  There are plenty of RISC machines that
aren't much different in context-swap time than a 68030, and a
couple that are better.

Which really doesn't amount to a hill of beans if you're talking
about single-user performance under something UNIX-ish.  The 
SPARC CPU in the SparcStation has about twice the integer
performance of the NeXT 68030 implementation, and about 10x
the floating point performance.  That's one of the main reasons
we have SparcStations here at work; they're used for IC design
work.  

Sun's been traditionally bad at making fast displays.  Our 7.16 MHz
68000 machines look good compared to Suns as far as display speed
goes.  I don't think I'd pick a SparcStation for desktop publishing
work.  I'm pretty happy with that on my Amiga 2500/30.  A NeXT might 
be a good choice too, though it's about twice the price.  Lots of
folks use Macs for this job too.  Obviously for DTP, speed is not
the most critical of factors, once you've hit a point of being
acceptably fast.

> Read just what (was it
> UNIXWORLD ?) wrote about the DECstation 3200 which is faster
> than a SPARCstation: if it comes down to 'real'-applications
> and not just benchmarks, it is about the speed of a good 386.

Then again, a good '386 system these days runs at 33MHz, has 
32-64k of external 0 wait cache, and beyond that some interleaved
SCRAM.  That's going to do quite a few things things noticably 
faster than a NeXT, even given the inherent advantages of an '030
over a '386 at the UNIX OS.  

> Ron
-- 
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