A close-up photo shows Loihi, Intel’s neuromorphic research chip. Intel’s latest neuromorphic system, Pohoiki Beach, will be comprised of 64 of these Loihi chips. (Credit: Tim Herman/Intel)
Intel unveiled its 8 million-neuron neuromorphic system codenamed Pohoiki Beach today at the Defense Advanced Research Projects Agency (DARPA) Electronics Resurgence Initiative in Detroit. Consisting of 64 Loihi research chips, Intel claims Pohoiki Beach can be up to 1,000 times faster and 10,000 times more efficient than CPUs for autonomous driving, robotics and other applications.
The 128-core, 14-nanometer Loihi neuromorphic chips, which Intel first detailed in October 2017, have a 60-millimeter die size and contain over 2 billion transistors, 130,000 artificial neurons, and 130 million synapses. Intel said Pohoiki Beach can help scale up neural-inspired algorithms such as simultaneous localization and mapping (SLAM) and indoor mapping for robots.
Intel defines neuromorphic computing as computing that emulates the neural structure of the brain, which can apply principles of common sense and context and deal with uncertainty, ambiguity and contradiction.
“We are impressed with the early results demonstrated as we scale Loihi to create more powerful neuromorphic systems,” said Rich Uhlig, managing director of Intel Labs. “Pohoiki Beach will now be available to more than 60 ecosystem partners, who will use this specialized system to solve complex, compute-intensive problems.”
Researchers are working to automate Western Sydney University’s foosball table under Loihi control, operating on visual input from event-based cameras. Foosball offers an excellent test for rapid closed-loop sensing, planning and control algorithms, a sweet spot for neuromorphic technology. (Credit: Sumit Bam Shrestha)
Some examples of how researchers are using the Pohoiki Beach include:
- Providing adaptation capabilities to the AMPRO prosthetic leg
- Object tracking using emerging event-based cameras
- Automating a foosball table with neuromorphic sensing and control
- Learning to control a linear inverted pendulum
- Inferring tactile input to the electronic skin of an iCub robot
“With the Loihi chip we’ve been able to demonstrate 109 times lower power consumption running a real-time deep learning benchmark compared to a GPU, and 5 times lower power consumption compared to specialized IoT inference hardware,” said Chris Eliasmith, co-CEO of Applied Brain Research and professor at University of Waterloo. “Even better, as we scale the network up by 50 times, Loihi maintains real-time performance results and uses only 30 percent more power, whereas the IoT hardware uses 500 percent more power and is no longer real-time.”
“Loihi allowed us to realize a spiking neural network that imitates the brain’s underlying neural representations and behavior. The SLAM solution emerged as a property of the network’s structure. We benchmarked the Loihi-run network and found it to be equally accurate while consuming 100 times less energy than a widely used CPU-run SLAM method for mobile robots,” professor Konstantinos Michmizos of Rutgers University.
Intel said it plans to introduce later in 2019 an even larger Loihi system named Pohoiki Springs. This version will build on the Pohoiki Beach architecture to deliver even more performance (up to 100 million neurons) and efficiency.
I would like to know how I can apply to Intel to be a research partner to experiment with the new Pohoiki Beach and the coming Pohoiki Springs hardware. I have 30 years of academic and industrial experience in machine learning, deep neural networks and deep reinforcement learning. I am extremely interested in neuromorphic approaches to greatly reduce energy needs. I would love to develop reinforcement learning algorithms to run on Pohoiki Beach.
Chuck, it may be too late for this round of testing, but I recommend reaching out to Intel directly.
I just want to know what is the price, how can be used with a computer
I am making an AI humanoid robot that I would like to try testing this loihi chip as part of it’s brain. This humanoid robot I have designed myself and she functions better than most of the robots out today. How can I get my hands on this chip?