On 11/3/2005 at 12:26 PM Paul Koning wrote:
Ok, let's test that. 1LT?
UGH! That was the long "stranger" tape driver overlay to 1MT. Ugly beast,
too--and an exercise in futility. 30 years ago, I spent entirely too much
time trying to make it run for some stupid benchmark. This was back on the
607's. You had to read the tape and write to CM at the same time. Lose
your place in the pyramid and you had to backspace to the beginning of the
record and try all over...and over...and over...again.
Before my time... I worked on PLATO when it ran on
6000 series NOS.
Since the PPU's on the 7000 had hard-wired buffer areas in CM, they
couldn't access all of CM. So the burden of the OS fell on the CPU. 7000
SCOPE used a very clever system of staggered RA's and overlapping FL's for
different levels of privilege. So, the job supervisor task had an FL that
included itself and the FL of the job it was supervising, but its code and
RA was placed immediately below the job, so it had unrestricted access to
the job's FL. Same sort of idea for the buffer and I/O managers. Very
clever implementation before protection "rings" got the credit for the
idea.
Good luck with 1LT--there's been much hair left on the floor from folks who
worked on that thing.
Cheers,
Chuck