Hailo’s New Chip Allows Faster, More Efficient Deep Learning

can be reached at meehna@gmail.com
can be reached at meehna@gmail.com
Share on facebook
Share on twitter
Share on pocket

In order for our increasingly autonomous cars to make crucial life and death decisions. engineers will rely on deep learning. Also known as artificial intelligence (AI), it’s a technology that learns by example, such that a driverless car will be able to recognize a stop sign or distinguish a pedestrian from a cyclist. Waymo just showed an amazing example of the technology in action.

  • Hailo’s new chip can process large amounts of information without using the cloud

  • The chip is smaller than a penny and costs less than current alternatives

  • Innovative design incorporates software developed in conjunction with the hardware

For deep learning to take place, large amounts of incoming data need to be processed. Using this data, models are trained through neural network architectures. With its small surface area, a computer chip has a finite capacity to do the job.

That’s where the cloud comes in. It has a massive capability to handle and process digital information for decision making. However, transferring data received from a car up to the cloud and back takes longer than if the data was crunched locally on the chip. Yes, the difference in time comes down to milliseconds, but those fractions of a second are huge. The result could be if you get in an accident or avert one.

Regular size CPU

Based in Israel, AI chipmaker Hailo developed what they claim is the world’s top performing deep learning processor. The Hailo-8™ processor can make those complicated computations for deep learning right on the chip. As a secondary advantage, by relying less on the cloud, the chip also lowers costs for automakers.

The Hailo-8™ processor features up to 26 tera operations per second (TOPS), on a chip smaller than a penny, including the required memory. Hailo’s innovative architecture relies on the core properties of neural networks and builds in a comprehensive Software Development Kit (SDK) co-developed with the hardware. ‘

Other advantages of the Hailo-8™ processor address the problem of heat. Because it operates with only a few Watts, there is no need for active cooling systems often needed in automotive applications.

Hailo plans to work with companies across many industries developing products for smart cities and homes. However, its focus is on automotive, and the company is already working with OEMs and tier-1 automotive companies


About the Author

can be reached at meehna@gmail.com
Close Menu