PCIe card brings edge AI acceleration to developers
Powered by BrainChip’s AKD1500 edge AI coprocessor, the AKD1500 PCIe development card enables developers to run models built with existing AI frameworks. The card plugs into a standard PCIe slot in a desktop, workstation, industrial PC, or single-board computer, providing access to the AKD1500’s self-learning capabilities for edge AI development.
The AKD1500 coprocessor chip uses the Akida neuromorphic processing engine to deliver up to 800 effective GOPS at <1 mW/GOP for low-power AI neural network acceleration. Its built-in capacity for on-device learning allows for secure application personalization without needing a cloud connection or extensive retraining.
The PCIe development card complements the AKD1500’s other deployment options, including an M.2 module, packaged and unpackaged silicon, and licensable IP. Developers can test their own models with streaming data on a PC, then deploy the validated models on AKD1500 hardware without modification.
Available through BrainChip’s online store, the AKD1500 PCIe development card is priced at $149.
The post PCIe card brings edge AI acceleration to developers appeared first on EDN.
