Path: utzoo!attcan!uunet!dg!rec
From: rec@dg.dg.com (Robert Cousins)
Newsgroups: comp.sys.next
Subject: Re: NeXT with an '040
Message-ID: <125@dg.dg.com>
Date: 5 Apr 89 17:21:00 GMT
References: <9876@polyslo.CalPoly.EDU> <6492@cbmvax.UUCP>
Reply-To: rec@dg.UUCP (Robert Cousins)
Organization: Data General, Westboro, MA.
Lines: 42

In article <6492@cbmvax.UUCP> daveh@cbmvax.UUCP (Dave Haynie) writes:
>in article <9876@polyslo.CalPoly.EDU>, mdeale@algol.acs.calpoly.edu (Myron Deale) says:
>
>>   Heck, I for one would like to see a migration to RISC. An 88K has
>> acceptable performance :-)
>
>Moto's been claiming that the '040 will edge out the current 88100+2*88200
>in all-around performance.  Of course, the current 88ks only run at 20MHz
>or so....
>
>> Myron
>-- 
>Dave Haynie  "The 32 Bit Guy"     Commodore-Amiga  "The Crew That Never Rests"
>   {uunet|pyramid|rutgers}!cbmvax!daveh      PLINK: D-DAVE H     BIX: hazy
>              Amiga -- It's not just a job, it's an obsession
My experience with the 88K is that it is quite fast and capable of surprising
the user by its power.  On the DG AViiON series, we found that if one 
computed the MIPS rating off of simple Dhrystones and normalized to a
VAX 11/780 (in other words 1 Dhrystone MIPS = 1757 Dhrystones), the 88K
gives you more MIPS than the CPU clock rate!  Not bad for a RISC where
it was originally surmized that the ratio of CISC instructions to RISC
instructions would end up somewhat lopsided.

One of the major reasons for the 88K's improved performance is the 
enlarged number of registers (32 x 32 bits) along with the seperate
data and instruction caches (4 way set associative).  Another reason
that the 88K will do well is that it is easier to build a multiprocessor
88K machine in my experience than it is a 68K one.  Our experience with
multiple 88Ks shows that they perform wonderfully.  

Lastly, one reason why I'd buy an 88K over a 68K any day is that there is
an ABI (applications binary interface definition) which allows truly 
portable applications between vendors.  How many 68K vendors can point
to the fact that they can run the binaries from other vendors?  In fact
I do believe that some vendors are actually REMOVING binary compatibility
from their OSes to guarantee that this won't take place.  

Anyway, the 88K products are available now.

Robert Cousins

Speaking for myself alone.