Deep Learning Processor Sets Performance Benchmark

Hailo unveils the Hailo-8TM, what the company is calling the world’s top performing deep learning processor. The chip is built with a unique architecture that enables edge devices to run sophisticated deep learning applications that could previously run only on the cloud.

Hailo sees key disadvantages existing in the current architecture of the embedded processing infrastructure, designed based on a 70-year-old underlying structure. Hailo’s holistic solution rethinks the existing pillars of computer architecture – memory, control, and compute – and incorporates a key, comprehensive Software Development Kit (SDK) co-developed with the hardware. 

The Hailo-8 processor features up to 26 terra operations per second (TOPS) at a size smaller than a penny, including the required memory. By designing an architecture that relies on the core properties of neural networks, edge devices can now run deep learning applications at full scale more efficiently, effectively, and sustainably than traditional solutions, while significantly lowering costs. According to preliminary results comparing Hailo-8TM to Nvidia’s Xavier AGX, which runs NN benchmarks such as ResNet-50, Hailo-8 consumes almost 20 times less power while performing the same tasks.

ResNet-50 Benchmark

For more details, visit Hailo.

Read more on

Suggested Articles

Applications for Toposens’ “Bat Vision” TS3 sensors include navigation systems that mimic the echolocation techniques employed by bats and dolphins.

DTS system stacks slices to process and record flight data from many sensors

Western Digital announced a strategic partnership with Platform IO Labs.