[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: CCS 7811 Math Coprocessor
mdj wrote:
Greetings,
A kindly gentleman enriched my Apple II collection recently with (yet
another) Apple II+ and a collection of expansion cards including a CCS
A/D Converter and a 7811 Math Copro, complete with original
documentation :-D
The CCS 7811 has essentially the same coprocessor schematic as the
uSpeed FP Coprocessor card, with the exceptions that the uSpeed card
contains an alternate 4MHz clock generator for the fast version of
the 9811, and the CCS card has sockets for 256 bytes of slot ROM.
Unfortunately I don't have the software for the card, and from the
docs it seems it would only work in software mode on a Apple II, not a
II+, due to the requirement of a patched Applesoft.
Actually, you can load Applesoft into the Language Card and patch away.
I'd love to find the software/firmware for this card!
A cursory glance of the schematic in the manual also indicates that
the board asserts /RDY, presumably to halt the 6502 during a
calculation so that the next instruction can fetch the result. This
would render the device useless on an accelerated Apple II :-(
Unfortunately, yes--but an 8MHz Zip chip actually produces a
comparable or better speedup.
However, the manual also indicates that the board has a pair of
jumpers for rigging up /IRQ, which leads me to believe the board can
be operated in 'asynchronous' mode, which might make it compatible.
It would also be necessary to disconnect the RDY drive, by lifting
pin 3 of the 'LS03, for example.
The benchmarks listed in the manual indicate the performance gain on
DIV and MUL ops are pretty modest, and probably worse than an
accelerated Apple II doing soft DIV/MUL. However, the trig/exp/log
operations show an impressive 5 (and sometimes 6) orders of magnitude
improvement, which would probably smoke my accelerated IIe :-)
That seems a little high... An unaccelerated Apple II computes a
trig function in a couple hundred milliseconds, and the 9511 coprocessor
requires on the order of 5000 cycles (at 2MHz in the 7811 card). So I
put the speedup at about 200/2.5, or a factor of 80--a little less than
two orders of magnitude.
Anyway, just thought I ask here if anyone else has experience with
such a board, before I run off on some wild goose chase. Modifying the
Pascal p-machine to use this board would provide a spritely boost in
performance, and make for a fun weekend project.
I've played with one, and encountered the same disincentives as you:
incompatibility with accelerators and missing Applesoft patches.
It does do IEEE (SANE) 32-bit floating point considerably faster than
the software routines, but it would take some serious patching to
integrate it usefully.
-michael
NadaNet networking for Apple II computers!
Home page: http://members.aol.com/MJMahon/
"The wastebasket is our most important design
tool--and it's seriously underused."