VBRANDT@DBNUAMA1.BITNET (09/06/88)
This is directed primarily at the people at Atari, but since it contains
both praise for Atari (!) and information that might interest others, I post it
here.
I have just bought an "Atari SFP004(tm)" -- that's a 16 MHz 68881 FPU for
the expansion slot of the Mega series STs. It comes in a grey cardboard box
that's so new it still has spelling errors on it ('coprocesser' :-). I have
had my dealer install it, and now it sits in my ST waiting for work.
All the people I asked, including my dealer and representatives of Atari
Germany, used the same word for the software supplied with the FPU: "none".
Now I reread the one (1) sheet of paper that came with the 68881, and I noticed
that Atari *DOES* offer software support for the FPU, and I quote (from Atari
document C300443-001 Rev. A):
"The Atari SFP004 Developer's Kit contains more information on the co-
processor and the software you can use with it. The kit also contains a
disk of sample source and object files for a set of high precision
mathematical routines. (These routines are compatible with the Alcyon C
compiler.)"
This sounds interesting ! Since I don't have the dev package, I don't have
the Alcyon C compiler either, but 'sample source' would definitely be helpful.
Unfortunately, Atari Germany knows nothing about this (as usual). The sheet
also gives three addresses to contact for additional info: one in the US, one
in Canada and one in the UK. Now for my questions:
- What exactly is in the 'SFP004 Developer's Kit' ?
- Where is it available ? Will it *ever* be available in Germany ?
- Where should I ask for more info?
- What is the 'suggested price' for it in the US ?
Also, the sheet lists two other sources of information, the Motorola 68881
FP Coprocessor User's Manual, which I have (thanks to Motorola UK, Ltd., for
sending me a free copy when I just asked for general info !!!), and something
called "Software links math chip to 68000/family ,uPs AR233 (Reprinted from
EDN; Jan. 23, 1986)". What is this ? Could someone please tell me if this is
a journal, magazine etc. ?
Finally, the praise to Atari: Thanks for making this product available, one
that deserves the 'power without a price' label. (For those who wonder, the
'suggested price' in Germany is DM 398). One of the two other suppliers of
FPU cards for the ST in Germany (Weide of Hilden, W.Germany) has already
lowered the price for the Mega version from DM 898 to DM 498. Admittedly,
they supply FP libs for 6 different compilers ... still, they are expensive.
Things aren't so bright on the software side as yet:
- Heimsoeth is working on 68881 support for Turbo C (no details yet). I
don't know of other C compilers yet.
- "gdat" of Bielefeld have announced an APL.68000 version that incorporates
68881 support, but MicroAPL cannot specify any delivery date yet (I suppose
MicroAPL spend more time working on their new Mac APL these days.. ).
- I'm trying to find out more on the OSS/CCD Pascal, but no answer yet.
If anyone out there has some PD source and/or binary for FP routines that
make use of the 68881 FPU, I'd be *VERY* interested in a copy (if it's big,
send email first).
Any and all info is most welcome indeed.
PS: This has grown rather long, but I thought it might be of interest.
Bitnet: VBRANDT@DBNUAMA1 Volker A. Brandt
UUCP: ...!unido!DBNUAMA1.bitnet!vbrandt Angewandte Mathematik
ARPAnet: VBRANDT%DBNUAMA1.BITNET@CUNYVM.CUNY.EDU (Bonn, West Germany)good@atari.UUCP (Roy Good) (09/08/88)
All subsidiaries have now received a copy of the SFP004 Developer's Kit, which includes demos, sample routines and documentation. There should be no reason why they cannot provide information. Available applications and compilers is a separate issue, over which we have little control. All major compiler developers have been notified of the hardware addressing used (remember, this board operates in peripheral mode rather than coprocessor mode, in Motorola terminology). The sample code would allow you to "hand-carve" code to access the chip in 68K assembler, but it is up to the compiler people to generate object code directly from C, or have "intelligent routines" which can use the 68881 or emulate it. And once the compilers are there, applications will follow. Most of this is documented in the Dev.Kit manual, or in the sheet included with the SFP004. The 'three addresses' (US, Canada and UK) are cited for one reason only - this is the English/American [:-)] manual, and not the final native language one (ie the German one will indicate Raunheim, the French one Atari France etc) > - What is the 'suggested price' for it in the US ? > We have not set a price on it - it is part of the Developers' Kit update ($20) which also includes the TOS 1.4 Developer Release. > "Software links math chip to 68000/family ,uPs AR233 (Reprinted from > EDN; Jan. 23, 1986)". What is this ? Could someone please tell me if this is > a journal, magazine etc. ? This is Electronic Design News, a US periodical. I am sure European computer mags have run similar articles about 68000 (rather than 68020/030) interfaces to 68881. ----------------------------------------------------------------------------- Roy J. Good Product Development, Atari Corporation Views expressed are my own. Atari may agree or disagree; they have the right. -----------------------------------------------------------------------------
ljdickey@water.waterloo.edu (Lee Dickey) (09/08/88)
In article <8809061532.AA26180@ucbvax.Berkeley.EDU> VBRANDT@DBNUAMA1.BITNET writes: > > This is directed primarily at the people at Atari, but since it contains >both praise for Atari (!) and information that might interest others, I post >it here. ... [ lots of good stuff about 68881 ] ... > - "gdat" of Bielefeld have announced an APL.68000 version that incorporates > 68881 support, but MicroAPL cannot specify any delivery date yet (I suppose > MicroAPL spend more time working on their new Mac APL these days.. ). > - I'm trying to find out more on the OSS/CCD Pascal, but no answer yet. This interests me. Thank you for the info. Until recently, APL.68000 running on an Atari, Mac, or Amiga claimed to be the fastest APL on a micro-computer. This news makes it all the more interesting! -- L. J. Dickey, Faculty of Mathematics, University of Waterloo. ljdickey@WATDCS.UWaterloo.ca ljdickey@water.BITNET ljdickey@water.UUCP ..!uunet!watmath!water!ljdickey ljdickey@water.waterloo.edu