billsey@agora.UUCP (Bill Seymour) (04/24/89)
From article <4607@freja.diku.dk:, by dizzy@freja.diku.dk (John Bach Jensen):
: I recently wrote to ask how to connect a 68881 to the 68000 in my AMIGA 1000,
: without having to buy a 68020 board. I have received quite a few letters
: (thanks) who recommend me to buy the CMI-board. I seriously considered buying
: this board, but then I saw Peter Mitchell's article, which described the pin-
: connection between the 68000 and the 68881, and it looks like I might save a
: little dinero (means a lot to me), if I do it that way. BUT (and here comes my
: questions):
Now, of course, I would rather you by the CMI Processor Accelerator,
since I work for CMI... But failing that... :-)
: 1. How must I do the chip-select decoding if I want to make it
: compatible with the 1.3 math.library ? Can it be done at all ?
There is a nice app note from Motorola on just this. Get in touch
with your Mpot rep and they should be able to fix you up with the complete
specs. To keep it compatible with the 1.3 IEEE (And 1.4, of course) you have
to set up a MATH_RESOURCE with a pointer to the base address for your 68881.
This is documented in the 1.3 Native Developers Update, and I believe also
in last years DevCon disks....
: 2. Can I clock the 68881 at, say 14 MHz, without additional hardware
: (The CMI-board raises the clock-frequency, but I have no idea how)
We clock the processor faster, that's the hard part. The 68881 runs
async, so you can pretty much run it at any speed you want. I've found that
different speeds of math chip don't make a whole lot of difference in your
end speed. I clock mine at 20MHz on the 2000 and 16MHz on the 1000... But
that's just because those were the chips I had.
: 3. Can I (this might be a really stupid question) connect then 68881
: to the expansion-slot on the side.
I don't know how easy that would be to do... Maybe Dave can give more
help here... It's easier to do it internally, that's for sure!
: 4. Last but not least, is this so complicated that i would be better off,
: buing the CMI-board. I am *not* (as you might have guessed) an expert
: in this area.
It's certainly easier to buy our board... You'll probably have less
conflicts with other expansion devices also. Noise is a real big factor to
take into consideration on something like this, and doing it on a piece of
perf board would make it *real* hard to keep clean! Our newest accelerator
for the 1000 (Introduced at AmiEXPO NY) gives you a socket for Commodore's
ROM chip (which you can then run at 14MHz) and an optional battery backed
clock. The clock is the same setup as Commodore uses on the 500/2000 with
just a different address, so you can easily patch 'SetClock' to be used with
ours. We provide a little utility that does the patch for you...
: Thanks a lot in advance, and once again thanks to the people who responded to
: my original letter.
:
: John Bach Jensen.
: Student at the Institute of Datalogy, University of Copenhagen.
: E-mail to dizzy@freja.diku.dk
--
-Bill Seymour ...tektronix!reed!percival!agora!billsey
...tektronix!sequent.UUCP!blowpig!billsey
Creative Microsystems Northwest Amiga Group At Home Sometimes
(503) 691-2552 (503) 656-7393 BBS (503) 640-0842