The Computex event has seen many announcements about AI technology from companies like AMD, Intel, Microsoft, and NVIDIA. But the Raspberry Pi Foundation is not lagging behind. They have introduced an external NPU for the Raspberry Pi 5, priced at $70.
The Raspberry Pi 5 now boasts 13 TOPS with the addition of this external AI module. This NPU, built with the Hailo-8L chip, offers impressive AI performance. While it falls short of the latest processors from Intel and AMD, it is on par with Intel’s Meteor Lake NPU, AMD’s Ryzen 8040, and Apple’s M1 chip.
The kit includes an M.2 HAT+ card with a Hailo-8L chip, priced at $78.60. You can also purchase the M.2 HAT+ card separately for $13.5. This provides flexibility for users who may only want the NPU card without the M.2 HAT+.
The Raspberry Pi 5 supports PCIe 2.0, while the Hailo chip utilizes PCIe 3.0. Users can modify a configuration file to enable PCIe 3.0 on the Raspberry Pi 5, as recommended for optimal performance. However, it is important to note that the Raspberry Pi 5 is not officially certified for PCIe 3.0, and stability issues may arise.
The software integration for the AI kit is seamless, allowing users to quickly develop AI applications with low latency and energy consumption. The kit offers full integration with the Raspberry Pi’s image management subsystem, making it easy to install and use AI demos in minutes. Hailo provides documentation and application examples on GitHub for further assistance.
This AI kit is compatible with official Raspberry Pi cameras and partner cameras, as well as pre-recorded videos. The Raspberry Pi Foundation emphasizes the simplicity and ease of use of this kit, making AI applications accessible to a wide range of users.