home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.sys.apple2
- Path: sparky!uunet!snorkelwacker.mit.edu!ira.uka.de!chx400!josef!avalon.physik.unizh.ch!gudat
- From: gudat@avalon.physik.unizh.ch (gudat henrik)
- Subject: Re: AE Fastmath
- Message-ID: <1992Nov9.165018.29134@ifi.unizh.ch>
- Sender: news@ifi.unizh.ch (USENET News Admin)
- Nntp-Posting-Host: avalon
- Organization: Informatik Club Uni Zurich
- References: <9211072123.AA08918@gc3.gac.peachnet.edu>
- Date: Mon, 9 Nov 1992 16:50:18 GMT
- Lines: 29
-
- In article <9211072123.AA08918@gc3.gac.peachnet.edu> flowney@GC3.GAC.PEACHNET.EDU (Frank Lowney) writes:
- >I have an AE fastmath card that served quite well to speed up a statistics
- >program written in Applesoft BASIC (AppStats). This was accomplished by a
- >set of Ampersand commands that AE developed and shipped with the product.
- >I'd like to extend the functionality of this math coprocessor to other,
- >newer areas of my computing activities like HyperCard GS, AppleWorks 3.0
- >and AppleWorks GS (the spreadsheets particularly).
- >
-
- Just forget it. The AE FastMath is the worst coproc card I've ever seen.
-
- The "math" chip is so stupid and slow that it would slow down most math
- operations significantly.
-
- The problem is that the FM chip has only 48 bit resultion, and the SANE
- software always calculates with 80 bits. You'd have to convert values
- from one format to another and spend more time on converting and loading
- the chip's registers than doing it without FM.
-
- For AppleSoft it's fine because these rouztines are 8 bit (and thus a
- bit slower than 16 bit code) and have exactly the same precision.
-
- henrik
-
- --
- Bright Software CH gudat@avalon.physik.unizh.ch Bright Software USA
- P.O. Box 18 any demo: US$3.00 P.O. Box 120
- 4153 Reinach 2 info on GSymbolix, ShadowDial, Exeter, ME 04435-0120
- Switzerland Gate, SpaceFox, Mandelbrot U.S.A.
-