Magnetic computing promises dramatic energy reduction potential
The findings, which have been published in the journal Science Advances, indicate that major reductions in power consumption are possible – as much as one-millionth the amount of energy per operation used by transistors in modern computers.
“We wanted to know how small we could shrink the amount of energy needed for computing,” said senior author Jeffrey Bokor, a UC Berkeley professor of electrical engineering and computer sciences and a faculty scientist at the Lawrence Berkeley National Laboratory. “The biggest challenge in designing computers and, in fact, all our electronics today is reducing their energy consumption.”
“Making transistors go faster was requiring too much energy,” said Bokor, who is also the deputy director the Center for Energy Efficient Electronics Science, a Science and Technology Center at UC Berkeley funded by the National Science Foundation. “The chips were getting so hot they’d just melt.”
Researchers have been turning to alternatives to conventional transistors, which currently rely upon the movement of electrons to switch between 0s and 1s. Partly because of electrical resistance, it takes a fair amount of energy to ensure that the signal between the two states is clear and reliably distinguishable, and this results in excess heat.
Magnetic computing emerged as a promising candidate because the magnetic bits can be differentiated by direction, and it takes just as much energy to get the magnet to point left as it does to point right.
“These are two equal energy states, so we don’t throw energy away creating a high and low energy,” said Bokor.
Bokor teamed up with UC Berkeley postdoctoral researcher Jeongmin Hong, UC Berkeley graduate student Brian Lambson and Scott Dhuey at the Berkeley Lab’s Molecular Foundry, where the nanomagnets used in the study were fabricated.
The engineers experimentally tested and confirmed the Landauer limit, named after IBM Research Lab’s Rolf Landauer, who in 1961 found that in any computer, each single bit operation must expend an absolute minimum amount of energy.
Landauer’s discovery is based on the second law of thermodynamics, which states that as any physical system is transformed, going from a state of higher concentration to lower concentration, it gets increasingly disordered. That loss of order is called entropy, and it comes off as waste heat.
Landauer developed a formula to calculate this lowest limit of energy required for a computer operation. The result depends on the temperature of the computer; at room temperature, the limit amounts to about three zeptojoules, or one-hundredth the energy given up by a single atom when it emits one photon of light.
The UC Berkeley team used an innovative technique to measure the tiny amount of energy dissipation that resulted when they flipped a nanomagnetic bit. The researchers used a laser probe to carefully follow the direction that the magnet was pointing as an external magnetic field was used to rotate the magnet from ‘up’ to ‘down’ or vice versa.
They determined that it only took 15 millielectron volts of energy – the equivalent of three zeptojoules – to flip a magnetic bit at room temperature, effectively demonstrating the Landauer limit.
This is the first time that a practical memory bit could be manipulated and observed under conditions that would allow the Landauer limit to be reached, the authors said. Bokor and his team published a paper in 2011 that said this could theoretically be done, but it had not been demonstrated until now.
Bokor noted the paper is a proof of principle and that putting such chips into practical production will take more time. But the authors noted in the paper that “the significance of this result is that today’s computers are far from the fundamental limit and that future dramatic reductions in power consumption are possible.”
Energy-efficient computing needs new devices, architectures says report
Single-electron transistor will enable extremely low power computing
Gaming PCs offer huge energy savings potential, says Berkeley Lab
Energy-saving algorithm taps existing sensors to reduce bills