← Thu, Sep 24 digest

RAM: the forgotten history (2024)

read article ↗ hn thread ↗

A technical history of computer working memory before integrated circuits, examining relays, vacuum tubes, acoustic delay lines, and magnetic core storage.

read it It offers a deeply researched and engaging technical tour of early computer memory architectures that shaped modern computing constraints.

summary

The article explores the forgotten history of electronic working memory, arguing that RAM constraints were the primary bottleneck in early computing rather than basic programmability. It traces the evolution of memory from electromechanical relays and vacuum tube bistable circuits to quirky dynamic solutions like mercury delay lines and torsion memory. Finally, it covers the era of magnetic drum and solid-state magnetic core memories before the advent of large-scale integration chips.

key points

  1. Konrad Zuse’s Z3 computer used roughly 2,000 electromechanical relays for its memory unit compared to 600 for the rest of the machine.
  2. Early dynamic memory solutions included mercury delay lines that used toxic liquid to propagate sound waves as stored bits.
  3. Magnetic core memory used woven arrays of microscopic ferrite beads where sub-threshold magnetic fields flipped individual cores at wire intersections.

who should read it

Software and hardware engineers interested in computer history and the physical foundations of computing.

discussion warm

consensus

Commenters discuss the non-volatility of magnetic memories and whether ancient core stacks or magnetic drums can still retain readable data decades after being powered off.

pushback

One commenter notes that magnetic coatings on drums and platters are prone to flaking and mechanical degradation over time compared to core memories.

best comment by jandrese ↗

Usagi Electric managed to recover the code from a 1960s era LGP-21.

https://www.youtube.com/watch?v=IBjh0SaA5dc

https://www.youtube.com/watch?v=39SwIl3wQDw

The coating wasn't a problem, it was a couple of the heads that sort of failed.

It links to a practical demonstration showing the successful recovery of code from a 1960s magnetic drum computer.

Generated by an LLM from the article and thread. Check the sources before relying on it.