Omnibus core memory compatibility?

[Brent Hilpert]

Philipp Hachtmann wrote:
> 
> I have the following stuff:
> 
> G104 X/Y drivers
> G227 Sense/Inhibit
> G619A 4K core stack
> 
> The 4K stack is broken and lost forever.
> 
> But I have a spare G646C "4K or 8" Stack sitting around.
> I *think* it has 8K.
> 
> Does anyone know if I could use that stack with the rest of the package,
> using only 4K of the memory?
> 
> Does anyone have schematics for the 8K core system? In my documentation
> I have only information about the 4K system.
> 
> On Doug Jones' module list, there are mentioned 4K and 8K sense/inhibit
> boards. And there are mentioned X/Y driver boards, without 4K or 8K
> written on them.
> I would like to know how DEC expanded the core. Doubling X or Y lines
> would be the easiest. But why should that result in different
> sense/inhibit circuitry? The sense/inhibit connections look quite
> similar on both of my stacks...
> Or does everything change?
> 
> Thanks for clarification :-)
> 
> And if someone has a spare 4K stack, please contact me!
> 
> The same for 8K sense/inhibit and X/Y drivers.

There was a technique called "anti-coincident addressing" that permitted the
address space to be doubled without additional X/Y/inhibit drivers or sense
amplifiers. All that was required was a little additional logic in the address
decoding/selection and the additional cores. See
http://www.cs.ubc.ca/~hilpert/e/coremem/index.html#anticoincidence .

I'm not familiar with the DEC boardsets you're referring to but it's
conceivable this technique was used to produce a 4K/8K combination boardset in
some form.