The IBM researchers in Zürich have revealed their prototype for what they see as the next step towards zettascale computing; they believe the answer is “right behind our eyes.” For years we have compared the power and efficiency of our computers to that of the human brain but it has always had an edge on our developments. Sure IBM’s cleverest computer in the world, Watson, beat Ken Jennings and Brad Rutter in the hit TV show Jeopardy; but Watson was using 85,000 watts to function where Jennings and Rutter were only using 20 watts, a power efficiency that they would love to reach. It is on this principle that IBM constructs the foundation of their attempt at a “paradigm shift in electronics” by using a ‘redox flow’ system pumping an electrolyte ‘blood’ carrying power/energy in and taking heat out.
“We want to fit a supercomputer inside a sugar cube. To do that, we need a paradigm shift in electronics – we need to be motivated by our brain. The human brain is 10,000 times more dense and efficient than any computer today. That’s possible because it uses only one – extremely efficient – network of capillaries and blood vessels to transport heat and energy – all at the same time.”
– Dr. Bruno Michel, IBM Zürich R&D team
Michel believes that energy efficiency, rather than raw computing power, is what will allow is to push past the restrictions set by silicone chips. He believes that we can achieve this by following a more bionic computing path, rather than spintronics, quantum computing, or photonics and that through this, by 2060 we can expect to see a one petaflop computer, that currently would fill half a football field, fit under your desk.
He plans on reaching this goal by hitting 3 core objectives. The first is 3D architecture, with chips stacked high, and memory storage units interwoven with processors. This seems like a very obvious idea when you think about it, so why haven’t we already implemented it? Well the problem with this design is that processors generate a ton of heat that you need to somehow dissipate and since you’ve stacked them all up, there’s a massive restriction in the amount of airflow you’re allowing to hit the surface area of each chip. IBM’s solution to this is the 2nd objective, to layer water pipes between each layer as demonstrated in SuperMUC, the German supercomputer which perversely harnesses warm water to cool its circuits allowing a 40% reduction of electricity consumption.
Finally we come to the 3rd objective, the ‘blood’ part that IBM will use to deliver power to and cool the processors. In the same way that blood gives glucose in one instance and takes heat with another, IBM is looking for a fluid that can multitask and their best candidate so far is Vanadium, a type of redox flow unit. Don’t let me mislead you, redox flow is not a new technology at all nor is it overly complex, it’s just that IBM is the first to try to implement that model to computing.
So yeah, there’s nothing to expect soon per say, but we can expect great things in the future.
“The big issue with 3D computing is getting the heat out – and liquid cooling could be very effective if integrated into 3D systems as proposed here,” but all of the above will not get electronics down to the energy-efficiency of the brain. That will require many more changes, including a move to analogue computation instead of digital. It will also involve breakthroughs in new non-Turing models of computation, for example based on an understanding of how the brain processes information.”
– Prof Steve Furber, Project leader for SpiNNaker Project