Becoming brainier
Electrochemical powering could help to reduce processors' heat dissipation, but there is a way to make a much bigger difference. Most of the heat from a chip is generated not by the switching of transistors, but by resistance in the wires that carry signals between them. The problem is not the logic, then, but the legwork. During the late 1990s, when transistors were about 250 nanometers across, 'logic' and 'legwork' accounted for roughly equal amounts of dissipation. But today, says Michel, “wire energy losses are now more than ten times larger than the transistor-switching energy losses”. In fact, he says, “because all components have to stay active while waiting for information to arrive, transport-induced power loss accounts for as much as 99% of the total”.
This is why “the industry is moving away from traditional chip architectures, where communication losses drastically hinder performance and efficiency”, says Garimella. The solution seems obvious: reduce the distance over which information-carrying pulses of electricity must travel between logic operations. Transistors are already packed onto 2D chips about as densely as they can be. If they were stacked in 3D arrays instead, the energy lost in data transport could be cut drastically. The transport would also be faster. “If you reduce the linear dimension by a factor of ten, you save that much in wire-related energy, and your information arrives almost ten times faster,” says Michel. He foresees 3D supercomputers as small as sugar lumps.
What might 3D packaging look like? “We have to look for examples with better communication architecture,” Michel says. “The human brain is such an example.” The brain's task is demanding: on average, neural tissue consumes roughly ten times more power per unit volume than other human tissues — an energy appetite unmatched even in an Olympic runner's quadriceps. The brain accounts for just 2% of the body's volume, but 20% of its total energy demand.
Yet the brain is fantastically efficient compared to electronic computers. It can achieve five or six orders of magnitude more computation for each joule of energy consumed. Michel is convinced that the brain's efficiency is partly due to its architecture: it is a 3D, hierarchical network of interconnections, not a grid-like arrangement of circuits.
Smart build
This helps the brain to make much more efficient use of space. In a computer, as much as 96% of the machine's volume is used to transport heat, 1% is used for communication (transporting information) and just one-millionth of one per cent is used for transistors and other logic devices. By contrast, the brain uses only 10% of its volume for energy supply and thermal transport, 70% for communication and 20% for computation. Moreover, the brain's memory and computational modules are positioned close together, so that data stored long ago can be recalled in an instant. In computers, by contrast, the two elements are usually separate. “Computers will continue to be poor at fast recall unless architectures become more memory-centric”, says Michel. Three-dimensional packaging would bring the respective elements into much closer proximity.
All of this suggests to Michel that, if computers are going to be packaged three-dimensionally, it would be worthwhile to try to emulate the brain's hierarchical architecture. Such a hierarchy is already implicit in some proposed 3D designs: stacks of individual microprocessor chips (on which the transistors themselves could be wired in a branching network) are stacked into towers and interconnected on circuit boards, and these, in turn, are stacked together, enabling vertical communication between them. The result is a kind of 'orderly fractal' structure, a regular subdivision of space that looks the same at every scale.
See what we're tweeting about
5 Comments
Add CommentIf a chip is 2-D, but with devices on both sides, then devices on both sides can be cooled, for example, by emmersion in a flowing liquid. If most of the heat is in the wires, then put the wires on the outside where they can be cooled. While a 3-D architecture lets devices be closer to each other, you have the issue of how to cool stuff that isn't on the surface. One possible way of fixing that is to introduce holes for coolant to pass through the body of the device. How small can they be made? Can all devices be close enough to coolant? Does the liquid have to be forced through small tubes under high pressure? Perhaps a study of capilaries can help.
Reply | Report Abuse | Link to thisI dont believe in engineers !
Reply | Report Abuse | Link to thisEniac was NOT the first "modern electronic computer" and that was "built at the end of WW2".
Reply | Report Abuse | Link to thisThe first was built at Bletchley Park in Britain DURING WW2, and was instrumental in the code-breaking work that was carried out there on coded German radio transmissions.
A solution for cooling massive array's of computers circuits looms as the single largest obstacle to computing's future may have been solved using multiple low orbit processing centers cooled by space and powered by solar radiation? Computing using Low Orbit remote data processing centers that utilize encrypted microwave transmission to user centers for distributions to subscribers is the future?
Reply | Report Abuse | Link to thisGoogle could save $100's of Billions over Ten years of operations in energy costs it will spend to run current Data center and cooling costs. Space data centers would allow for the immediate development of future computing, exaflop machine.
Ron Nussbeck
Great article - thanks. I was wondering if to the best of your knowledge anyone is actively working on a "reversible computing" means as suggested by Ray Kurzweil? That method is supposed to generate zero net thermal loss. I can sorta-kinda envisage the logical operations involved - and wondered if perhaps the new memristor semiconductor flavor or the much in-the-news electron spin storage and other pending quantum computing widgets would get us any closer to that dream?
Reply | Report Abuse | Link to thisTo clarify the reference - quoting from his "The Singularity is Near"... Quote: "Rolf Landauer showed in 1961 that reversible logical operations such as NOT (turning a bit into its opposite) could be performed without putting energy in or taking heat out, but that irreversible logical operations such as AND (generating bit C, which is a 1 if and only if both inputs A and Bare 1) do require energy. In 1973 Charles Bennett showed that any computation could be performed using only reversible logical operations. A decade later, Ed Fredkin and Tommaso Toffoli presented a comprehensive review of the idea of reversible computing. The fundamental concept is that if you keep all the intermediate results and then run the algorithm backward when you've finished your calculation, you end up where you started, have used no energy, and generated no heat. Along the way, however, you've calculated the result of the algorithm." End quote.
He then goes on to inspiringly ask "How Smart Is a Rock?". Wondering if there any fundamental efforts towards changing the calculation process, not just the wiring and location? (This middle aged engineer needs a zetaflop before he turns 100!).