[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Supercard



Keith wrote:
Michael J. Mahon wrote:

Keith wrote:

It's so easy to do, you might as well, assuming you've got room for it
in the FGPA (two processor cores is likely to eat a lot of room)

Matt


Based on my experience with the Xilinx Spartan XC3S1000 (Digilent
Spartan-3 Board) I would guess that you could just about squeeze an
entire AppleCrate onto one. And the communications channel could be
much faster than that used on an AppleCrate. Don't know if anyone would
be interested in that or not. You could probably clock them faster than
an AppleCrate too, but maybe not much faster since you would probably
be time domain multiplexing the SRAM to provide enough memory to the
CPU cores.

Cool!  ;-)

Of course, with both on-chip and shared memory communication possible,
the entire networking premise of AppleCrate would be unneeded.  A much
closer coupling should be supported--closer to SMP.

It's fun to think about hi-tech approaches to lo-tech problems!  ;-)

-michael

Parallel computing for 8-bit Apple II's!
Home page:  http://members.aol.com/MJMahon/

"The wastebasket is our most important design
tool--and it is seriously underused."


How much of an Apple II would one need to emulate in order to get the
client-side of AppleCrate working? Assume that it would use the same
networking as your box is using, and not some new invention. It would
seem that you could leave out a lot of the logic that makes an Apple II
and Apple II. Is this correct? If it's not too bad I might try to do it
for fun.

The only thing required to participate in a NadaNet network (which is
what AppleCrate machines do) is the ability to bit-bang the serial
network line at 94-95 Apple cycles per byte (8 cycles per bit) and
sense them if receiving (while maintaining phase lock within 3 cycles).

The rigidity of the timing requirements makes it hard to do in any
environment in which interrupts can cause jitter of more than a
microsecond per byte, like PCs and Macs, unless interrupts can be
disabled for the duration of a packet, which could be nearly 25
milliseconds (which is also problematic for PCs and Macs).

There should be no problem at all meeting these requirements using
a microcontroller or hardware.  Of course, the protocols for actually
using the network for something would also need to be implemented,
and they currently exist only as 6502 code, optionally invoked by
ampersand routines in Applesoft.

-michael

Parallel computing for 8-bit Apple II's!
Home page:  http://members.aol.com/MJMahon/

"The wastebasket is our most important design
tool--and it is seriously underused."