How many kinds of mass-producible ROM is avaliable before 1970? [duplicate]

Question

Back in the days before mask ROM (when invented?) we got core rope memories, or Little Old Ladies memories, which required lots of human labor to build.

Were there any kinds of ROMs which were suitable for mass production before Semiconductor ROM dominated (when?) the market?

PS: I don't think the 1959 PROM satisfies the requirement. It needed lots of standalone diodes and auto assembly was not a thing yet(though auto burning might be possible?).

There are some canidates of non-volatile memories listed below beyond TROS or CCROS, andplease don't hesitate to teach me their PROs and CONs or if there was any other means.

• ENIAC's Function Tables were resistor matrix ROM
• Plated wire memory
• Thin-film memory
• Twistor memory
• Bubble memory
• Magnetic stick memory
• Two channel punched tape

LINK: Historical price of ROM

Cheapest type of Read-Only Memory allowing Random access before Year 1970

Answer 1

IBM's TROS (Transformer Read-Only Storage) and CCROS (Capacity Coupled Read-Only Storage) both were ROM-alike, entirely free of semiconductor material, and basically one-time programmed like punch tape (it could be re-programmed by exchanging the tapes with re-punched ones, and CCROS even shared the same format with IBM punch cards for its plastic sheet, so you could program your ROM with a standard punch). It was also largely mass-production ready.

I think these should satisfy your requirements.

Answer 2

I suppose it depends what counts as "mass produced". One can implement the same kind of circuits used in IC mask ROMs on a PCB with diodes. This is a PDP-11 boot ROM from the early 1970s. (Source of image and short article about the board here.)

You place a diode where you want your bits to go and solder it in. DEC also sold blanks, where you modified it to hold the data you want. (What I don't know is whether the blanks came pre-soldered with diodes in every position or not. Presumably just snipping out the 0s is much easier for the customer than soldering in the 1s, but also more expensive.)

This technology was used by MIT's Whirlwind to implement the memory for its microcode program, operational by 1951. But that machine was not mass-produced, and neither were its components.

Discrete diode ROMs would definitely be mass produced by the mid-1960s: the first electronic desktop calculator was probably the ANITA, which used a diode matrix ROM like that to encode its control program. It was available in 1961, and is the first mass-produced product I know of like that.

But I know IBM was using it in some of their designs around then, and it's possible they were turning out boards like that in significant production quantities before 1960. Perhaps someone else will help be able to narrow it down more between '51 and '61.