Sometime later this year, Intel plans to release what it's calling "the industry's first silicon for neural network processing." Designed specifically for machine learning and artificial intelligence-focused computing, the Intel Nervana Neural Network Processor (NNP) will feature a new type of architecture compared to standard silicon chips. It's being built on the technology Intel acquired last year through its $408 million purchase of Nervana, a California-based startup.
Intel said the purpose-built NNP will support the development of "entirely new classes of applications" that can draw intelligent insights from large volumes of data. Those capabilities could enable better healthcare diagnoses, improve personalization on social media, speed up advances in autonomous vehicles, and provide more accurate weather and hurricane forecasts, according to the company.
Changing the Silicon Portfolio
Being developed in collaboration with Facebook, Intel's Neural Network Processor was unveiled yesterday at the Wall Street Journal's D.Live global technology conference in Laguna Beach, Calif. Intel CEO Brian Krzanich, who spoke at the conference, wrote afterward in a commentary that the chip will help drive progress in cognitive and AI technologies that will be transformative to the industry and the world.
"We have multiple generations of Intel Nervana NNP products in the pipeline that will deliver higher performance and enable new levels of scalability for AI models," Krzanich noted. "This puts us on track to exceed the goal we set last year of achieving 100 times greater AI performance by 2020."
In a separate announcement yesterday, former Nervana CEO and co-founder Naveen Rao added that the new NNP has been in development for more than three years, and features a design aimed at supporting more advanced machine learning and deep learning workloads.
"We've been listening to our customers and applying changes to Intel's silicon portfolio to deliver superior machine learning performance," said Rao, who is now vice president and general manager of Intel's AI products group. He added that the NNP was developed based on "the need for new thinking around the hardware required to support AI computations."
'Neuromorphic' and Quantum Advances on the Way
Intel is pursuing AI and cognitive computing advances on multiple fronts, and has recently acquired a number of other companies aimed at boosting its capabilities in machine learning, computer vision, and other technologies.
Last month, Intel introduced a self-learning test chip called Loihi that's designed to imitate the processes of the human brain. The "neuromorphic" chip, which is designed to make data-based inferences and become "smarter" over time, is expected to be shared with other AI-focused researchers in the first half of next year, according to Michael Mayberry, corporate vice president and managing director of Intel Labs.
Working with QuTech, a partner based in the Netherlands, Intel last week said it had come out with a 17-qubit superconducting test chip for use in quantum computing. Being developed for faster-than-silicon computing, quantum computers use qubits rather than binary bits for complex and data-intensive calculations.
Krzanich noted that Intel expects to come out with an even more advanced 49-qubit chip by the end of the year. Quantum computers offer the potential to "tackle problems conventional computers can't handle," he said. "We are well on our way to unlocking the potential of AI and paving a path to new forms of computing."