[inforoots] Operating Systems from CDC

[Peter G. Capek]

I wrote:

>  In true MIT fashion, the hackers at the AI lab, primarily
>  Greenblatt and Gosper, I believe, created ITS, the Incompatible 
Timesharing
>  System.   How incompatible?  Well, it ran on a PDP-6 and then a PDP-10, so
>  not even the word length was the same.

And Mike Albaugh responded:

 >  Huh? The PDP-6 and PDP-10 had/have 36-bit words, so _that_ much would
 >  be compatible with the 704/709/7090/7094 (et al.) (and 3600, but not 
6600 :-)

And Tom Van Vleck said (privately): 

 >  Don't forget Tom Knight, who is still at AI.  Wonderful folks all.
 >  But the PDP-6 and PDP-10 had the same word length as the 7094,
 >  36 bits.

And you're absolutely right.  Brainfart on my part.  Somehow I was 
thinking momentarily
that the PDP-6/10 used 32 bit words.

And I fully agree about tk, too.   Shouldn't have forgotten him..

Here, by the way, is a site about ITS:   http://www.its.os.org/

I also wrote:

>    At some point, based on additions to the CPU architecture,

and Mike Albaugh responded:

 >  Which is to say, the ability for the CP to initiate an Exchange Jump.
 >   (CEJ/MEJ?)

Yes, although it's slightly more than that: the introduction of a 
monitor vs. problem
state for the CPU.   With this change, the CPU could always perform an 
Exchange
Jump, but when in monitor mode, theCPU program could specify the address 
of the
exchange package;  when in problem mode, the address used for the 
exchange package
was one which had been specified by the operating system.

But the 3600 was a 48-bitter.

      Peter Capek