1 Aug
2021
1 Aug
'21
12:10 p.m.
On 8/1/21 11:45 AM, ben via cctalk wrote:
I suspect if they had the proper virtual memory, it
would have been picked
up as a Unix cpu, instead. It is the only common non segemented 16 bit
cpu I can think in that time frame.
There were ways around that. You could run two, with one slightly
lagging the other, such that one would hit a trap condition before the
"real one" (was that Daisy?) Or you could limit the instruction set to
the restartable ones (Apple).
I recall attending a trade show (NCC, WESCON...?) where the Moto reps
were handing out literature (but not chips). After a brief reading of
the 68K manual, I approached the rep with a question about implementing
virtual memory. He looked wearily at me (he'd probably been asked the
same question a hundred times before) and quietly said, no, there wasn't
enough information saved in a trap to ensure restartability of all
instructions.
You also have to recall that at the time, virtual memory Unix was very
new--I think it was introduced with BSD in 1979 or so. I still have a
BSD tape from Ernie (Co VAX) from that time.
--Chuck