This technology is a microcontroller unit (MCU) with an integrated in-memory computing accelerator that can be used to maximize acceleration while minimizing the required hardware.
Current tiny machine learning technology is focused on using microcontrollers to bring artificial intelligence to smaller devices. Microcontrollers are smaller and have higher processing capabilities, however, they do not possess enough energy efficiency, throughput, and robustness, which creates the need for an integrated RAM accelerator – taking up more space and increasing price. Although some microcontrollers possess high-computing capabilities, they use analog computing, which degrades their robustness. There are currently no low-cost (area-efficient) microcontrollers with robust integrated RAM accelerators.
This technology is a microcontroller unit (MCU) capable of artificial intelligence computations in tiny machine learning technology. The MCU integrates a RAM-based in-memory computing accelerator, which decreases the space taken up by the unit while also increasing its capabilities. The computing flow and customized software maximizes the acceleration of the MCU with need for less hardware.
This technology has been validated with a manufactured prototype.
Patent Pending
IR CU23146
Licensing Contact: Greg Maskel