🚀 go-pugleaf

bernecky@gack.eecg (Robert Bernecky) writes:

>I laid hands [and installed SHARP APL] on one at University of Waterloo
>back in the early 70's. It was a strange box, indeed. 

I used that machine.  And another one at the University of Toronto
in 1969.

The /44 was sold as a real time / process control machine, and there
were several odd operating systems to support this (DAMP, for
example).

One of the interesting things about the /44 is that it was the
cheapest hard-wired 360.  All others below the /75 were microcoded.
Bell and Newell's book (I think) pointed out how it was way off the
apparently-planned 360 family price/performance curve.

It was missing lots of 360 instructions, but it did have support for
an emulator for those instructions.  The emulator was unobtrusive
enough that OS/360 could be run without it knowing it was on a
substandard machine.  Even running normal 360 code, with the
emulator for the missing instructions, the average performance was
very good.  The missing instructions, in retrospect, are the
non-RISC 360 instructions (the term RISC wasn't used, but some of
the ideas were around).

BTW, like the KA10 (the original PDP-10), the base model had its
general purpose registers in a private core memory.  The emulator
could address this core, which made me daydream about implementing
some inner loops in that memory so that an Insert Character
instruction could modify another instruction in the loop (for
example, MVI).  I never learned enough about the machine to try
this.  Besides, I bet everyone bought the semiconductor register
option.

Hugh Redelmeier  hugh@mimosa.com  +1 416 482-8253