Path: utzoo!censor!geac!torsqnt!news-server.csri.toronto.edu!cs.utexas.edu!usc!apple!das
From: das@Apple.COM (David Shayer)
Newsgroups: comp.sys.mac.misc
Subject: Re: The New Macs: Greedy Compromises?
Message-ID: <47215@apple.Apple.COM>
Date: 8 Dec 90 23:39:11 GMT
References: <1990Dec5.072931.4079@cs.ucla.edu> <11447@goofy.Apple.COM> <1990Dec7.085447.19397@cs.ucla.edu>
Organization: The Troll Den
Lines: 21

In article <1990Dec7.085447.19397@cs.ucla.edu> lange@lanai.cs.ucla.edu (Trent Lange) writes:
>I was using that fact that it has a 25 MHz 68040, rather than the IIfx's
>40 MHz 68030.  All published reports are that the 68040 is approximately
>3x faster than the 030 at the same clock speed, and 5-10x faster on
>floating point.  A quick "average" of those (and considering the 40 MHz
>clockspeed of the IIfx's 030) makes the NeXTCube (and NeXTStation)'s
>raw processing power about three times the IIfx's.  It's also nice that
>the NeXTs have DMA that they actually *use*.
A/UX uses the DMA.  Its only the Mac OS which doesn't use the DMA.

>And, since you mentioned Mathematica, the 68040 NeXT has been benchmarked
>to run it 50% faster than a Sparcstation 1+.  So much for RISC
>architectures.  :-)
Performance on RISC machines (and CISC machines, for that matter) often
has much more to do with the quality of the compiler than the speed of
the processor.  Unless you knew that the compiler used on each machine
produced code of comparable quality, I don't you think have much of a
benchmark.

David
"There are lies, damn lies, and benchmarks."  Motorola PR sheet