Re: 3860 bytes per second on a 38400 bps line

From: Marko Mäkelä (marko.makela_at_hut.fi)
Date: 2003-06-05 19:02:39

On Thu, Jun 05, 2003 at 10:51:44AM -0400, Ethan Dicks wrote:
> I know I owe you a PET test, but my 2001N (U.S. version of the 3032,
> not the 2001-8) is still flakey when it comes to disk access (I'm
> still trying to test my Zmachine port), and to make matters worse,
> my laptop disappeared from a convention over the Memorial Day
> holiday.  :-(

I think someone (Nicolas Welte?) already tested c2nload on PET, and the
new release didn't change any of the three PET loaders (a universal one
that must be RUN, and two auto-starting ones, for 8032 ports #1 and #2).

On Thu, Jun 05, 2003 at 05:04:25PM +0200, MagerValp wrote:
> 0.005% is well within the error margin.

My computer thinks that 20/3840 equals 0.5%.  You have a point, especially
the real-time clocks on PCs are often very inaccurate.  But 0.5% would
correspond to 18 seconds in an hour.

My next challenge with the C2N232 will be to come up with finite state
machines for the serial bus protocol (with transitions corresponding to
timeouts and signal changes on ATN, DATA and CLK), to implement entirely
interrupt-driven routing between RS-232 and the serial bus.

Or does someone already have such state machines?  Polling code will not
be of much use to me.  So far, I've found the following documents:

http://www.funet.fi/pub/cbm/schematics/misc/iec-protocol.gif
(hopelessly inaccurate and apparently incorrect)

http://www.funet.fi/pub/cbm/crossplatform/emulators/VICE/vice-1.11.tar.gz
(file vice-1.11/doc/html/plain/serial.txt contains very convincing-looking
timing diagrams)

	Marko

       Message was sent through the cbm-hackers mailing list

Archive generated by hypermail pre-2.1.8.