Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!rice!titan.rice.edu!preston
From: preston@titan.rice.edu (Preston Briggs)
Newsgroups: comp.arch
Subject: Re: int x int -> long for * (or is it 32x32->64)
Keywords: arithmetic,arbitrary precision,benchmark,modular arithmetic
Message-ID: <1990Sep13.161546.5794@rice.edu>
Date: 13 Sep 90 16:15:46 GMT
References: <41425@mips.mips.COM> <353@kaos.MATH.UCLA.EDU> <119977@linus.mitre.org>
Sender: news@rice.edu (News)
Organization: Rice University, Houston
Lines: 28

In article <119977@linus.mitre.org> bs@linus.mitre.org (Robert D. Silverman) writes:

>One thing is clear throughout ALL of this. Machines like the SPARC
>which provide NO integer multiply/divide are not worth much for
>this type of work. This is why I question ALUs like the i860
>which also has no integer multiply/divide, yet does have a fancy
>graphics unit.

I expect you're right.  The i860 is supposed to crunch floating-point.
The integer half is supposed to do the necessary integer work to
support addressing.  I guess they intend the graphics portion to
display the results...

I argued about this with a guy who wanted to put a chess program on
an i860.  It seems like a big mismatch of problem and resources.

Personally, I'd rather see the 860+ drop the graphics instructions
and give me a "4-address" multiply-accumulate instruction.
Then we could do some crunching.

For your stuff, a good integer processor with no FP or graphics
or whatever sounds like the ticket.  Unfortunately, most machines
use integers to do characters, addresses, and sometimes bits.  
Are there exceptions?  Digital signal processor chips?

-- 
Preston Briggs				looking for the great leap forward
preston@titan.rice.edu