Path: utzoo!attcan!utgpu!jarvis.csri.toronto.edu!mailrus!bbn!bbn.com!clements
From: clements@bbn.com (Bob Clements)
Newsgroups: comp.arch
Subject: Re: Registers in memory
Message-ID: <43492@bbn.COM>
Date: 29 Jul 89 15:56:39 GMT
References: <8907281547.AA28878@maxwell.ece.cmu.edu> <3602@csd4.milw.wisc.edu>
Sender: news@bbn.COM
Reply-To: clements@BBN.COM (Bob Clements)
Organization: Once DEC, but not recently!
Lines: 18

From article <8907281547.AA28878@maxwell.ece.cmu.edu>, by mbjr@maxwell.ece.cmu.edu (Mauricio Breternitz):
>      The (now ancient) IBM 1130 also had the three index registers
> at memory locations 1, 2, 3

The not-quite-so-ancient-as-an-1130 DEC PDP-5 had the Program Counter
in core memory, at location zero.

The bootstrap loader (which read programs at 110 BPS from the reader
of the Model 33 TTY) used this feature to transfer to the newly
loaded program.  Just read a datum into location zero, and suddenly you
were no longer running in the bootstrap.

The PDP-8, largely compatible with the PDP-5, had its PC in flipflops.
So the bootstrap had to recognize stores to zero and treat them as
jumps.

/Rcc
Bob Clements, K1BC, clements@bbn.com