[Simh] Non-numeric character in IBM 1620 P & Q addresses

[Paul Koning]

> On Dec 25, 2015, at 2:21 AM, Bob Supnik <bob at supnik.org> wrote:
> 
> ...
> So either no core is selected or multiple cores are selected. Now what happens? I suspect it's something bad.

FWIW: given how core works, if more than one core select line is active, the read cycle will see the OR of the two selected cores (since either of them magnetized will produce the pulse due to the magnetization flip).  On the restore (write) cycle, assuming a single global inhibit line it will write the same to both cores.  So you're going to see the OR of the two words, if the core uses 1 for magnetized cores (AND otherwise).

If you have segmented inhibit, as in the CDC 6000 core modules, things get more interesting.  Suppose the inhibit decode is working right but the X or Y decode has a fault, and the two cores selected are in different inhibit regions.  If so, the read would produce the OR of the two bits, and the write would restore the core whose inhibit is driven, but you'd get a 1 in the other core since its inhibit is not active.

I suppose this sort of thing could happen in the wild with a fault in the select logic, which explains where there are "no dual address" memory diagnostics.

	paul