3 ms·
His analysis was on a machine with a very unique architecture. I've generally assumed that the problem he was analyzing was somewhat unique to that design. Th
by jandrese 8y ago
His analysis was on a machine with a very unique architecture. I've generally assumed that the problem he was analyzing was somewhat unique to that design.
The Thinking Machines computers he was analyzing were massively parallelized, they had thousand of processors, but the processors were unbelievably primitive. They operated on a single bit and had a tiny number of instructions. The part Fenyman analyzed was the routing between the processors.
You also missed the followup. When the design was nearing completion they realized that they wouldn't have the silicon for 7 buffers per chip so the hardware was finalized with 5 buffers per chip. It was enough.