On Sat, May 29, 2010 at 1:28 PM, Jim Brain <brain at jbrain.com> wrote:
The story goes that Tramiel was unhappy with the price
and availability of
the IEEE-488 cables on the PET line, and told the VIC designers to "get off
that bus!". ?They looked at the VIA 6522 and designed a serial protocol
around the use of the shift register. ?Late in the game, software developers
became aware of the 6522 Shift Register "bug", and so had to retreat back to
bit banging the bus.
Were there any versions of the 6522 (65C22?) that had the shift
register bug fixed? Does the 6526 preserve the bug? With all the
Speed DOS products that were out there, I can't imagine it would be
worth trying to (re)create the originally envisioned
hardware-implemented bit-serial interface, but once I heard the story
about the shift register bug, I'd always wondered what it would have
been like to have used a drive that wasn't so glacially slow.
So, for that bug, the serial speed went from CLK/32
to
transfer a byte down to CLK/320 or so.
I hadn't realized it was that much slower. Wow.
My first C-64 (S/N P00002007, a development machine sent from
Commodore to my first employer) came with a 1540 and some extra
operational instructions - I think it was to disable VIC-II DMA, like
the KERNEL already does during a tape load.
-ethan