Years ago I bought an 8k card of core memory. I recently reattached it to the back of a shadow box and thought some might enjoy seeing single bits of memory. If I remember correctly standard core memory had 3 wires going thru the “donut”: 1 to read, 1 to write and 1 to clear. Anyway, here’s some pictures of ancient computer memory.
9 Likes
Do you think the void sticker still applies?
2 Likes
Interesting that this one is HP branded. I’ve seen the similar ones that DEC used which are also represented in several computer museums.
Do you know what piece of HP equipment it was used in?
2 Likes
Unfortunately I don’t know anything about it.
Can you imagine having the job of threading those wires? That’s the thing that always comes to my mind when I look at it. Well, that and how far technology has come.
4 Likes
Handweavers were in charge of teaching those skills in the beginning. It’s amazing how much influence the making of textiles and textile good have on other industries and trade.
6 Likes
Is that 8 kilobytes or?
Yeah, I believe 8 kilobytes. I notice that it actually says “8k x 17” but 136k seems like it would be a weird configuration so I’m not sure what to think.
2 Likes
Using one of the numbers visible on the PCB, I believe that this board was used in the systems described here: HP 21xx - 2116 Processors
The capacity would have been 8 kword and the X17 would refer to 16 bit word length plus 1 bit parity.
Similar memory boards used by DEC were around $20k.
In the HP 2116 series the B model contains up to two 8 KWord modules. The C model contains up to four 8 KWord modules,
2 Likes
Thanks, Mike! I was hoping someone would fill in the blanks and you did!
1 Like
The three sets of wires in core memory were alpha, beta and read. alpha and beta were orthogonal to each other. In simple 2D assemblies like depicted here, these were equivalent to rows vs columns. There are as many alpha wires as one axis, the same number of betas as the other axis, and one or more read wires (depending on the size of the array and speed characteristics)
Each core required a certain amount of magnetic flux to set its magnetic polarity. Neither the alpha and beta wires carried enough current to pass this threshold, instead a single alpha and beta combination had to be energized at once to set the polarity for the core shared by both wires. Whenever the polarity of a core flipped, a small current would be induced in the read wire passing through that core. “Reading” was done by actually setting the polarity at a position to a specified orientation; If the read line detected a changed polarity then the core had been in the opposite polarity prior to the read; otherwise, the polarity was already in the specified orientation. Any flipped bits were then flipped back to preserve the memory contents.
Later on, a few bespoke implementations were explored for larger mainframes, the most interesting of which (my personal opinion) had a seperate core array for each bit in the mainframes word (I think it was seven bits in this case). This allowed the machine to read / write an entire word at once rather than a bit at a time.
an interesting fact was that women were often employed in the manufacturing process of core memory because they often had seamstress experience which made them far more efficient at threading the miniscule cores with thin wires.
1 Like