Re: Constructed Computer Architectures (Concomps?)

On Tue, Feb 10, 2009 at 6:48 AM, Wesley Parish
<wes.parish@...> wrote:
> It's hardly on the same level, but during 1993/94 I fooled around with
> eradicating the reliance on a CPU with its consequent reduction of everything
> else to the level of peripherals.  I was at the time much taken with the Unix
> concept of everything as a file, and the SCSI concept of everything as
> an "autonomous" node, meaning that it had its own "intelligence".
In the case of Commodore 8-bit computers, that was how all the
"peripherals" worked, a few years before SCSI.  Of course, that also
meant that peripherals were about as expensive as the computers
themselves.  :)  For instance, the floppy disk drive was essentially a
(serially) networked storage device with its own CPU and operating
system, that responded to requests from the host computer to perform
file operations on its behalf.  As long as you were using the standard
disk format/filesystem, the computer had nothing to do with the
low-level writing of disk blocks; it just said "hey, disk drive, give
me program file FRED" and the disk drive said "Sure, boss, here
y'go!".  But if you wanted finer control, you could also say "Hey,
disk drive, here's a bunch of data.  Write it to track 19, sector 0,
'kay?"  The overhead of the serial communications command/response
protocol made this latter sort of thing slower; when Commodore came
out with the 128, which supported CP/M and hence needed lots of that
sort of custom direct disk I/O, it compensated by giving it and the
associated new floppy drive model the ability to negotiate a faster
bitrate.

None of which has anything to do with conlanging, although I do have
some conlang notes that are still sitting on a Commodore floppy in my
basement...
--
Mark J. Reed <markjreed@...>