Тик-так, тик-так или какво е тиктакало в Союз:

https://youtu.be/JBIhzEZkWEA

Актуална тема.

> NASA needs 14kb to control a spacecraft
> Linus (Sebastian) needs more than 1 petabyte to run some YouTube channels

At 2:00, you are describing "core" memory, which was the type of memory used in most computers up to around 1975-1980, when semiconductor memory first entered the scene. I worked on an IBM 1620, a 1401 and others, including a minicomputer made by a company called Microdata, which all used core memory. (I'm 72.)

I still have three 8K memory boards from that Microdata computer. (Failure of this type of memory was common, so I decided to save boards when they were replaced.) I remember they were priced at $3,500 each back then, or around 5 cents/bit. At that rate, 16GB of RAM (128 billion bits), now commonplace in PC's, would cost 6.4 billion dollars (without accounting for inflation)! That's over double the entire Apollo program budget. (Good thing they didn't need 16GB.)

Just as shown in this video, I can see the little cores arranged on the grid of wires. I was amazed back then, and I still find it impressive today. Here's an interesting fact about core memory: In order to read a bit, the computer would actually write a zero at that location. A sense wire running through all the bits on that plane would detect a pulse if that location originally contained a 1 because reversing the magnetic direction would induce that electric pulse. A second cycle was then required to restore the bit back to a 1 (or leave it as 0 if no pulse was detected). This was known as "destructive" read, requiring two machine cycles. One cycle on the Microdata was one microsecond (1 MHz). A modern CPU running at 4GHz is 4,000 times faster. But one cool thing: if the computer lost power, core memory was preserved! It wasn't 100% reliable, but often, when the computer was powered back on after a power failure, it could continue running where it left off!