Re: Software for uSPEED math card

[mjmahon@aol.com (Michael J. Mahon)]

brad wrote:

>"Michael J. Mahon" <mjmahon@aol.com> wrote in message
>20040131135737.13722.00000755@mb-m27.aol.com">news:20040131135737.13722.00000755@mb-m27.aol.com...
>> I recently acquired a uSPEED math card for the Apple II series,
>> made by Applied Analytics, Inc.
>>
>> According to published reviews, it came with a version of Applesoft
>> that was patched to use if for floating-point arithmetic and functions.
>>
>> Does anyone have this software, or know where it can be found?
>
>i recently got a saturn systems math co-processor card for the apple ii with
>the amd 9511 coded math chip
>
>did anyone ever find the programs needed from this post to get one of these
>beasts to work...i think a prev post said these cards used the same math
>chip with the 9511 in the descrip ..again mine was AMD
>
>and yes i do have 2 sets of 8" drives and apple lobo controller if they are
>on 8" disks...not exactly sure what would be needed to xfer them to a 800k
>disk or some such.. but...doable i guess
>
>anyway...let me know if someone tripped over a link or has the stuff...maybe
>there is something i got doc wise or otherwise you may need
>
>anyway thanks for your time..and see my other post on saturn systems apple
>ii math co-processing card for what i managed to snag

Hi, Brad!

First, a question:  is this card made by Saturn Systems or Computer Station?
The auction suggests the latter, and the photo leaves it pretty ambiguous.  ;-)

You are correct, this is one of a small set of Apple II math accelerators
based on the AMD9511.  They were produced in various clock speeds,
from 2MHz to 4MHz, and offered a big speedup in floating-point arithmetic
and trig functions compared to a 1MHz 6502 running the routines in the
Applesoft ROM.

No one has come forward with the software designed to make the card
work within Applesoft--yet.

I have the chip datasheet (from Paul R. Santa-Maria), and it is instructive.
The actual AMD datasheet only contains electrical, timing, and mechanical
data, and was apparently supplemented by a programming manual, which
I have not found.  But the Intel 8321 datasheet describes a fully compatible
chip, complete with programming information.

Using the Intel datasheet, I wrote some Applesoft routines to "exercise"
the chip and test it.  It works pretty well, but it only supports 32-bit FP
numbers, with a signed 24-bit mantissa--a little short compared to the
32-bit (with separate sign) of Applesoft 40-bit FP numbers.

I have the MicroSpeed ("uSpeed") card, which was designed to rely on
using the 6502 RDY line to pause the processor (if necessary) until a result
was ready.  Since my primary machine contains a Zip Chip, which does not
support the RDY line, it is a bit tricky to use the card at full speed on my
primary machine.

Speed tests show that the chip accelerates a 1MHz Applesoft program
up to 5x, or so, depending on the amount and type of floating-point
computation.  My 8MHz Zip Chip speeds up just about every Applesoft
program by about the same amount, so, ironically, the FP accelerator
is not a big win for many programs (very trig-heavy programs might be
an exception).

In short, the idea of a math accelerator offered a very short-lived advantage
compared to the much more generally useful cached fast processor.

-michael

Check out amazing quality sound for 8-bit Apples on my
Home page:  http://members.aol.com/MJMahon/