Will we see advances in microprocessor speed during the next decade similar to what we saw in the past two decades? If so, how?
Researchers at DARPA asked this question, or, rather, asked: “What sort of technologies would engineers need by 2015 to build a supercomputer capable of executing a quintillion (1018) mathematical operations per second?” (IEEE, 2011).
The short answer is, it can’t be done by 2015 and, possibly, anytime in the near future. The reason for this is something called the power wall.
Modern supercomputers are based on groups of tightly interconnected microprocessors. For decades, successive generations of those microprocessors have gotten ever faster as their individual transistors got smaller—the familiar Moore’s Law paradigm. About five years ago, however, the top speed for most microprocessors peaked when their clocks hit about 3 gigahertz. The problem is not that the individual transistors themselves can’t be pushed to run faster; they can. But doing so for the many millions of them found on a typical microprocessor would require that chip to dissipate impractical amounts of heat. Computer engineers call this the power wall. Given that obstacle, it’s clear that all kinds of computers, including supercomputers, are not going to advance at nearly the rates they have in the past (IEEE, 2011).
The other part of the power wall — my version, anyway — is the amount of electricity required to run the machine. To run an exaflops-class supercomputer would essentially require a nuclear power plant next door — which is not exactly feasible.
So much for blazing fast speed with which to game and write The Great American Novel!
Still, how fast is fast? The infographic below uses bended bars to describe advances in microprocessor speed — i.e., “number crunching” — since 1993. (You can click on the image to view and zoom on a larger version of it.)
Figure 1 — “Number Crunching”, courtesy IEEE Spectrum via the article, Next-Generation Supercomputers.
If you would like to read a summary of the expected limitations of next-generation supercomputing, read the 3-page article in IEEE Spectrum called, “Next-Generation Supercomputers“. If you would like to read the 278-page .pdf research report on the limitations of next-generation supercomuting, go to doe.gov.
Is there any data on the infographic above that surprises you? Would you have designed the data representation the same or differently?