Lattice sensAI Ups Performance by 10X For Smart Edge Devices
July 10, 2019
Lattice is already known for delivering low-power programmable devices and continues that trend with this release while embarking down the coveted AI path.
If you haven’t heard about the push for AI in the embedded and industrial sectors, you’re either not paying attention or living under a rock. One of the latest vendors to enter the fray is Lattice Semiconductor, who recently developed its sensAI solution.
Lattice is already known for delivering low-power programmable devices and continues that trend with this release while embarking down the coveted AI path. While not officially a “new device,” this release adds performance and design flow enhancements to the current sensAI stack.
The sensAI stack provides a comprehensive hardware and software solution for implementing low-power (from 1 mW to 1 W), always-on AI functionality in smart devices that operate at the Edge of the IoT. Compute power at the edge continues to increase for obvious reasons—the ability to operate in real time, the security of keeping your data local, and the less (expensive) bandwidth that’s required.
A solution like sensAI lets OEMs more seamlessly update their existing designs with low power AI inferencing that’s optimized for their specific application. Enhancements to the stack include a 10x performance boost over the previous version, which is fueled by an updated CNN IP and neural network compiler with features like 8-bit activation quantization, smart layer merging and a dual-DSP engine.
It should also be noted that Lattice is growing its partner ecosystem around the sensAI platform. That includes design services and full product development.