Re: hello, whereFore Apple.COM

Lee Larson (lmlars01@homer.louisville.edu)
Tue, 3 Jun 97 21:01:07 -0400


>At 2:38 PM -0400 6/3/97, Garance A Drosehn wrote:
>>seth <sjohnson@smart.net> wrote:
>>>
>>> I would also like to remind Mac contributors that Motorola has
>>> an extension that promises to speed math processes on PPC hardware.
>>
>>I must admit I'm a bit leary of trying this extension while working
>>on the deschall client... :-)

I've used the Motorola routines for quite a while, and they do make a
significant difference in speed for some things. But I would not expect
them to speed up the DES client because they are mostly for
transcendental functions and some string handling.

By the way, I suspect one of the reasons Apple's routines are slower is
because they're trying to stick by the letter of the law as written in
the Standard Apple Numerics Environment (SANE). SANE is not fast, but it
is a fairly accurate implementation of the IEEE floating point standards.
Motorola was not held to such constraints. There are many people who
prefer predictability to raw speed.

Lee Larson http://www.louisville.edu/~lmlars01/ (502)852-6826
Mathematics Department, University of Louisville, Louisville, KY 40292 USA