bill (05/04/83)
We recently visited the makers of the Saybrook coprocessor card for the
Apple ][. It has an 8 or 12.5 MHz 68000 with 128KB of ram (expandable
to 512KB if you are rich and buy the new 256K ram chips; if you are poor
you can wait for the price to come down. The on-board address decoder
will handle either type of memory chip). The card fits into a standard
Apple slot with a little room to spare (i.e., it has a "nose job" in
front). The card has DMA to the Apple memory, which greatly speeds up
processor-to-processor communication. A separate power supply is
provided.
The operating system is the Softech P-system with Pascal and Fortran
(communicates with Apple Pascal and runs 68K Pascal). It supports
64-bit floating point, unlike Apple Pascal. Being interpreted, it is
not as fast as the 68K is capable of, but it still goes like a banshee.
They are now getting the last bugs out of the Apple to Saybrook software
interface (the person we talked to was estimating 2 weeks, but we know
how these things have a way of extending themselves...) Even so, it was
impressive. The 8 MHz board should run about as fast as the Sage (my
memory may fail me, it was my impression that their processor was 6 MHz,
but that may have changed). At 12.5 MHz there should be a substantial
speed advantage.
They plan to have a UN*X subset available later this year, which would
make it even more interesting. The C will be compiled into native code,
not interpreted using P-code like Pascal. A hard disk would be
desirable for this option. The person we talked to said that UN*X would
probably be an inexpensive (~$100?) option, but that this could depend
on how much interest there was in it.
Price for the 8 MHz board is $1550 (includes power supply and the
P-system software). The 8 MHz board is upgradable to 12.5 for several
$100's.
The manufacturer is
Analytical Engines, Inc.
3415 Greystone, Suite 305
Austin TX 78731
(512)346-8430
Fritz Benedict
Bill Jefferys
Pete Shelus
Astronomy Dept
University of Texas
(...ucbvax!nbires!ut-ngp!utastro!bill)
(...decvax!eagle!ut-ngp!utastro!bill)
( utastro!bill@utexas-11)tjt (05/07/83)
Anticipating flames in response to my previous followup ... The coprocessor board has its own memory which apparently is not accessible over the Apple II bus (the article said the 68000 could access the Apple II memory -- not vice versa). This should make it substantially easier to build sufficiently fast memory for a 12.5MHz processor. In particular, you can attach the memory more or less directly to the 68000 without bothering with bus drivers, arbiters, and the like. You can even punt a memory controller and refresh the RAM from software (a la SUN). In summary, I COULD be wrong about the 12.5MHz processor not being substantially faster than their 8MHz processor, but I would still be surprised. Tom Teixeira Massachusetts Computer Corporation (MASSCOMP) ...!decvax!genrad!masscom!tjt