Researchers at Sandia National Laboratories have demonstrated that neuromorphic computers that replicate the brain’s logic synthetically can solve more complex problems than those posed by AI.
In a newly published article in the journal Nature Electronics, the researchers detailed their findings which show that neuromorphic simulations using the statistical method called random walks can do all sorts of advanced computations like tracking X-rays passing through bone and soft tissue, disease passing through a population, information flowing through social networks and more.
Neuromorphic computers can even solve problems faster while using less energy than conventional computing in optimal cases according to Sandia theoretical neuroscientist and lead researcher James Bradley Aimone. This should be of particular interest to the high-performance computing (HPC) community as statistical problems aren’t really well-suited for either GPUs or CPUs.
Sandia engineer and the author of the new paper, Brian Franke provided further insight in a press release on how neuromorphic computers can be more efficient than GPUs in certain scenarios, saying:
“The natural randomness of the processes you list will make them inefficient when directly mapped onto vector processors like GPUs on next-generation computational efforts. Meanwhile, neuromorphic architectures are an intriguing and radically different alternative for particle simulation that may lead to a scalable and energy-efficient approach for solving problems of interest to us.”
Neuromorphic computing
In order to run their tests, the Sandia researchers used the 50-million-chip Loihi platform that they received a year and a half ago from Intel.
While neormorphic computing isn’t meant to challenge other computing methods, there are other areas in which the combination of computing speed and low energy costs make it a better choice according to Aimone.
At the same time, chips containing artificial neurons are cheap and easy to install unlike the difficulties posed by adding qubits to quantum computers. However, moving data on or off of neurochip processors can get expensive as the more data they collect, the slower a system using them becomes until it eventually won’t run at all. Sandia’s researchers were able to overcome this obstacle though by configuring a small group of neurons that computed summary statistics which were outputted instead of raw data.
Just like the human brain, neuromorphic chips work by electrifying small pin-like structures and adding tiny charges emitted from surrounding sensors until a certain electrical level is reached. Then the pin flashes a tiny electrical burst like a biological neuron.
Going forward, the next version of Loihi will increase its current chip scale from 128k neurons per chip to up to one million with large scale systems combining multiple chips to a board. Eventually, a technology like Loihi may find its way into a high-performance computing platform to help make HPC more energy efficient and environmentally friendly as well as more affordable.
