News

Kneron, a neural processing specialist backed by Qualcomm, is preparing to begin sampling an automotive-grade neuron processing unit (NPU) with an integrated image signal processor (ISP). The KL730, the new device, features a peripheral interface that can connect various digital signals such as image, video, audio, and millimeter waves.

Compared to previous Kneron devices, the KL730 claims to deliver a significant improvement in energy efficiency, with a 3 to 4 times increase. Designed for edge-AI applications, the chip supports compute power ranging from 0.35 to 4 effective TOPS, enabling it to support advanced lightweight GPT large language models like the nanoGPT.

One of the key features of the KL730 is its ability to address security in the AIoT (Artificial Intelligence of Things) space. It allows users to run GPT models either partially or fully offline, providing an added layer of privacy and control over their data. The chip utilizes Kneo, Kneron’s private edge AI network, enabling AI to reside on users’ edge devices.

This feature is particularly advantageous for engineers who can now design integrated circuits (ICs) without having to share confidential data with large cloud companies that operate data centers. By keeping the AI processing closer to the edge devices, users have more control over data privacy and can avoid relying on external cloud servers.

For more information about Kneron and its products, visit their website at kneron.com.

The post Kneron to Sample Automotive-Grade Neural Processing Unit for Edge AI appeared first on satProviders.