Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!iuvax!cica!tut.cis.ohio-state.edu!cs.utexas.edu!sun-barr!apple!vsi1!wyse!stevew
From: stevew@wyse.wyse.com (Steve Wilson xttemp dept303)
Newsgroups: comp.arch
Subject: Re: John von Neumann, sqrt instr
Message-ID: <2376@wyse.wyse.com>
Date: 18 Aug 89 20:30:12 GMT
References: <21353@cup.portal.com> <25643@obiwan.mips.COM> <1513@l.cc.purdue.edu>
Reply-To: stevew@wyse.UUCP (Steve Wilson xttemp dept303)
Distribution: usa
Organization: Wyse Technology
Lines: 27

In article <1513@l.cc.purdue.edu> cik@l.cc.purdue.edu (Herman Rubin) writes:
>
>There is no reason why a machine with hardware division should not have
>hardware square root.  It costs almost nothing.
>-- 
>Herman Rubin, Dept. of Statistics, Purdue Univ., West Lafayette IN47907
>Phone: (317)494-6054
>hrubin@l.cc.purdue.edu (Internet, bitnet, UUCP)

Ah, but that presumes that hardware division is warranted!

The Cydra-5 included hardware divide and square-root on the some
board so your contention on this point is correct. However, I'm
not sure I'd accept the proposition that these functions were mandatory
for the majority of applications.

I know that both operations sure did a number on scheduling the inner-most
loop.  Both operations had a long latency, thus caused scheduling
headaches.  The other point is that this board was fairly expensive.
Does the occurance rate of divide/square root in scientific computing
justify the cost?

How does the scientific computing community feel about this functionality?

Steve Wilson

The above are my opinions, not those of my employer.