Hand-wired core for ROM?

[Brent Hilpert]

On 2015-Apr-10, at 9:32 AM, geneb wrote:
> On Fri, 10 Apr 2015, William Donzelli wrote:
> 
>> You should probably cut him a little slack, since core and rope are
>> easily confused, as they are so similar. Rope could be rewired in the
>> field, but I certainly would not want to be the FE with that job.
>> 
> Yeah.  I think I'm just going to walk away from it. :)

Core RAM and "core rope" ROM sure are often confused, but technically I would disagree that they are similar.
They have about as much in common with each other as they do with the power transformer in the other corner of the chassis of whatever equipment they are in, that is to say what they have in common is using electro-magnetics and having a magnetic core.

The operating principles and design and implementation of core RAM and core rope ROM are very different. Notably:

	- Core rope ROM does not magnetise the core for it's memory function.
	Core rope ROM is really just pulse transformers with multiple primaries on each transformer.
	The presence or absence of a primary for a particular address (whether or which way the address wire goes through the core) is the memory function.
	This is in contrast to core RAM of course, which relies on changing the remanent magnetic state of the core for the memory function.

	- Core rope ROM has one magnetic core per the word-width of the memory. That is, a memory of (say) 1024 16-bit words would have 16 cores.
	Core RAM has one magnetic core per memory bit, a memory of 1024 16-bit words has 16,348 cores.

	- Core rope ROM typically or often did not use doughnut-shaped cores, they were often some sort of splittable / two-part core so they could be
	 accessed for the memory-state wiring.


Strikes me as conceivable that DEC used a field-configurable core rope design for the PDP-14 program memory - something like the PDP-9 module that Mattis linked to could be field alterable, but it sounds like "joe" in the OP thread may be confusing things.