Man Builds Giant Homemade Computer To Play Tetris (bbc.com)
An anonymous reader quotes a report from BBC: A man has finished building an enormous computer in the sitting room of his bungalow in Cambridge. James Newman started work on the "Megaprocessor," which is 33ft (10m) wide and 6ft (2m) high, in 2012. It does the job of a chip-sized microprocessor and Mr Newman has spent $53,000 creating it. It contains 40,000 transistors, 10,000 LED lights and it weighs around half a ton (500kg). So far, he has used it to play the classic video game Tetris. Mr Newman, a digital electronics engineer, started the project because he was learning about transistors and wanted to visualize how a microprocessor worked. The components all light up as the huge device carries out a task. Mr Newman hopes the Megaprocessor will be used as an educational tool and is planning a series of open days at his home over the summer. You can watch a video demonstration of the monstrosity here.
When I was somewhat younger, I was a so-called field engineer responsible for keeping some discrete element computers running.
Here's a picture of a module. This would be a single logic element such as a flip-flop, NAND gate, OR, etc.
https://www.etsy.com/listing/2...
The CPU cabinet was a huge box full of these things. The I/O controllers were in another cabinet, and the memory was in another cabinet.
The other boxes (storage, printers, card readers) had these same modules in them.
I never was main support for a CPU using those modules, but had some peripherals that had those things inside.
In more modern computers, these modules were replaced by logic cards. A PCB would have the transistors/diodes, etc to make a single element such as NAND gates, flip-flops or whatever, and these cards might have as many as 4 or even 6 logic elements on a single card. woo-eee!
I was lucky to be supporting such modern machines.
These old machines required hand-tuning such as manually synchronizing the clock signals between the near and far part of the cabinets.
The oldest machine I had to maintain was an 80 column card reader that used mechanical relays for all the logic elements. That was so long ago that the nightmares have stopped.
I used to visually monitor small computers running by using a pair of 8 bit DACs connected to the address bus with the analog outputs connected to the X and Y of an oscilloscope in XY vector mode. Where the scope trace moved around on the screen showed the branching locations of the CPU. Even without really understanding which exact locations the processor was running through you could get a heuristic view of the program in action.