[ansible]

The closest real-world thing would be something like The Connection Machine, which combined 64Ki processors in a hypercube. Individually, they had just a small amount of RAM for programs.

Today's GPUs are more about executing the exact same small program in a massively parallel way, and not individually programming each of the compute elements.

You know... it would be a fun project to make a TIS-100 in actual hardware on an FPGA or something. Actually, if you aren't the FPGA type, you could just write an emulator on an 8-bit microcontroller, and connect them all together. Bonus points for having each one with a display that shows its current execution state.