From: Nicolas Welte (welte_at_chemie.uni-konstanz.de)
Date: 2002-07-29 11:25:04
Hi Nick,
ncoplin@orbeng.com wrote:
> Could you identify which chips/pins these other wires go to? Do you think
> that this would be a easy way to also engineer a faster 65802 CPU into the
> C128? In the past I tried doing something on the by multiplying the clock
> phase on the 8502 side, but as the RAS and CAS signals are produced by the
> VIC it would crash. How does the Z80 get away with it?
Reengineering this device is one thing on my to-do list, of course. First
comes building that GAL-Blast device to read out the GAL, if it isn't
protected.
As I understand the circuit, it runs the Z80 at 8MHz until it requests a
memory access. Then it is halted by the GAL until it is time for Z80 memory
accesses. The original Commodore way is different: The Z80 is run at 4MHz
half of a 1MHz Phi2 cycle, and during the other half it is halted completely
to allow for a memory access, not matter if it is needed or not.
This only works for the Z80 with its extra status lines that signal memory
and i/o accesses, with a 6502 family member you're out of luck with such a
simple approach. IMO the speedup for the Z80 will only be effective for
opcodes that eat up a lot of cycles. Short opcodes with lots of memory
accesses will be as slow as ever, I guess.
A few days ago I also found an advertisment for this device in an old 64'er
issue from '89: The device was sold by Rossmöller for DM 99, they also had an
8MHz CP/M cartridge for the C64, which also sold for DM 99! An 8MHz CP/M
cartridge for the C64 was also published by c't magazine:
http://www.ix.de/ct/inhverz/search.shtml?T=z80+c64&Suchen=suchen
For a faster 6502 you have to do it just the other way around: Run the 6502
on its own fast memory subsystem and give the VIC/DMA system some memory
access slots to this fast memory. You have to buffer the data for the slow
VIC with latches, so the data that is fetched at 8MHz is still accessible to
the 2MHz device.
Nicolas
Message was sent through the cbm-hackers mailing list
Archive generated by hypermail 2.1.4.