29 Oct
2010
29 Oct
'10
11:39 p.m.
On 29 Oct 2010 at 22:58, Ian King wrote:
And we're discussing this because...? :-)
I think it stems from the discussion of PDP-11 "firsts". Memory-
mapped I/O was brought up and I mentioned that the CDC 7600 CPU
enjoyed memory-mapped I/O, but Erik disagreed. Since there's not a
lot of information wandering around about 7600 I/O system structure,
it could be that he's not familiar with it.
A shame that much of the 7600 information may be lost. Bitsavers only
has a couple of documents related to 7000 SCOPE, none of which
discusses the OS organization, which was unusual for CDC and probably
much of the industry at the time.
My take on the PDP-11 trailblazing was UNIBUS, which essentially let
I/O and memory cards be plugged into the same bus without (much)
regard to placement. Memory-mapped I/O appears to be an outgrowth of
that philosophy. I don't think anyone even mentioned UNIBUS in the
PDP-11 discussion...
--Chuck