Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!uwm.edu!linac!att!ucbvax!WATSON.IBM.COM!jbs
From: jbs@WATSON.IBM.COM
Newsgroups: comp.arch
Subject: float/float = integer, remainder
Message-ID: <9105171235.AA20130@ucbvax.Berkeley.EDU>
Date: 17 May 91 05:18:29 GMT
Sender: daemon@ucbvax.BERKELEY.EDU
Lines: 17


        David Hough states:
A couple of issues have been confused in recent comp.arch discussion.
One is whether argument reduction for elementary transcendental functions
involves simultaneous computation of a quotient (preferably in integer
format) and a remainder; another is whether instructions for that purpose
should be included in computer instruction set architectures.

        Well the reason they have been "confused" is that people advo-
cating the float/float = integer, remainder instruction have stated
that such an instruction would be useful for argument reduction for
elementary transcendental functions.  David Hough appears (his post was
little unclear) to agree with me that this is not the case.
        If you are going to put an instruction in your architecture
that may run for 1000's of cycles and causes all sorts of design
headaches it would be nice if it did something useful.
                       James B. Shearer