It has been mentioned in the CPU page that a good way of improving performance is to run more than one processor at the same time.

This is called 'Parallel Processing'.

Parallel processing involves taking a large task, dividing it into several smaller tasks, and then working on each of those smaller tasks simultaneously. The goal of this divide-and-conquer approach is to complete the larger task in less time than it would have taken to do it in one large chunk.

Back in the 1980's people realised that the CPUs of that time were not really suitable for parallel processing.

So a small British company called Inmos designed a processor that was designed specifically for parallel processing. It was called the Transputer.

You could easily join two or more Transputers together. They also developed a programming language especially designed for parallel processing.

In a way it was ahead of its time - it was too soon to switch over to parallel processing and so the world kept on using ever faster single chips.

In 2004, the practical limits of what can be done with a single CPU has arrived - the faster a chip runs, the hotter it gets. The hotter is gets - the shorter its lifetime.

(Just as an aside - put the search term "Arrhenius reliability" into Google and see what you come up with. but not right now!)

Intel, one of the largest CPU developers in the world, have recently decided to stop focussing on turning up the speed of their CPUs. Instead they are going to develop chips that hold more than one CPU cores. These CPUs will work together in parallel.