The Power of the NORC 3753

The quality of Wikipedia articles reflects the inputs and the people who have taken the time to edit. This is the first page which made my blood boil and I did my first edit.

FP performance, as far as I can tell can be traced back by an argument for the LLL purchase of the 1st CDC 6600 before Congress. About 1964. It was regarded at the time as a concrete measure. I would argue that over time as the code went from 1-D to 2-D to 3-D FP performance took less significance. It is well known by the people who use and buy machines that it's the von Neumann memory-CPU bottleneck which throttles performance.

A whole some what long esoteric history is involved.

The guys who sign the checks would take issue. This is what computational complexity is about. The Kings of Science, the physicists, control all this. Computists are just pawns in their book.

The Brits would take issue with the latter.

Specialized string search machines exist. They mapped the genome and more.

The problem is that there is no consistent definition for raw power. Even a clock cycle is not measured properly.

They do more index calculation and other aspects to get to the FP numbers than they give credit. They make it look worse. An FP to solution "counts;" an FP op to graphics does not.

Depend what the purpose of recognition is. I can think of machines that certain people don't want recognized.

The real question is one of software. Because of an ad(vert) it's otherwise just a paperweight. In Mary Shafer's words they are computation hanger queens if you can't use them, and a lot of machines in the past weren't usable. Power is just not a good word.

--