Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!tut.cis.ohio-state.edu!unmvax!aplcen!haven!adm!smoke!gwyn
From: gwyn@smoke.BRL.MIL (Doug Gwyn)
Newsgroups: comp.sys.apple
Subject: Re: Floating Point Engine
Message-ID: <10841@smoke.BRL.MIL>
Date: 28 Aug 89 06:59:54 GMT
References: <8908250820.AA01027@trout.nosc.mil>
Reply-To: gwyn@brl.arpa (Doug Gwyn)
Organization: Ballistic Research Lab (BRL), APG, MD.
Lines: 22

In article <8908250820.AA01027@trout.nosc.mil> lbotez@pro-sol.cts.com (Lynda Botez) writes:
>Anyone on the net have the math coprocessor called the "FloaPoint Engine"?

Yes, I recently obtained one via ByteWorks.

>I'm curious to hear about it; also is it compatible with a TWGS?

It's 68881-based.  It works.  For best results, you need a compiler that
can generate in-line code for it.  I don't know whether or not it would
work with a TWGS, but since it's a slot peripheral (not using the COP hook
in the 65816), it ought to work with the TWGS.  You can put the FPE in any
peripheral slot; a corresponding INIT for each one is provided.

>Though it is another major "big bucks" peripheral ($200+)... it sure sounded
>interesting to me.

Well, as usual it depends on your needs.  It can be used to speed up any
software that uses the SANE toolbox routines.  However, SANE is so slow
that your applications may be providing their own floating-point emulation,
which the FPE will NOT help with.  Orca/C supports the FPE quite nicely.
Benchmark programs that make heavy use of floating point typically speed
up by several orders of magnitude.