NXP Semiconductors has announced a new family of microcontrollers, with which it hopes to take on the challenge of edge AI and on-device machine learning: the NXP MCX range.
"“As we approach the milestone of 75 billion connected devices, we are entering a new era of edge computing, requiring us to fundamentally rethink how to best architect a flexible MCU portfolio that is scalable, optimized and can be the foundation for energy efficient industrial and IoT edge applications today and in the decades to come," claims NXP's Ron Martino. "By building on our strong legacy in MCUs, this new portfolio will offer the performance and integration needed to address the real-time workloads for the next wave of innovation."
The portfolio in question is split across four key model families. The MCX N Advanced series is the headliner: High-performance processing with dedicated neural processing unit (NPU) coprocessor offering a claimed thirtyfold performance gain for inferencing workloads and an EdgeLock security subsystem with side-channel protected cryptographic accelerators. The MCX A Essential range, meanwhile, offers optimizations for cost reduction and advanced analog capabilities.
The MCX W Wireless chips add in low-power radio connectivity, including BLuetooth Low Energy (BLE), with a focus on connected battery-powered devices. Finally, the MCX L Ultra-Low Power range does exactly what it says on the tin: Reduces the power draw down to what NXP claims is "one of the industry's lowest static and dynamic power consumption" with a view to boosting battery life.
Precise specifications vary by device: All models are built around an Arm Cortex-M33 processing core, but running at different speeds: The MCX N Advanced family runs between 150MHz and 250MHz; the MCX A Essential at 48MHz to 96MHz; the MCX W Wireless at 32MHz to 150MHz; and the MCX L Ultra-Low Power at a surprisingly speedy 50MHz to 100MHz with optional boost mode for a fifty per cent clock speed gain.
Models will be available with up to 4MB of on-chip flash and up to 1MB of static RAM (SRAM), the company has confirmed.
NXP is keen to point out that the new chips aren't a replacement for its existing Kinetis and LPC ranges, which it says "are still relevant devices in our industry." Instead, the new parts will be sold alongside NXP's older designs — and will use the same MCUXpresso toolchain, with the differences constrained to the hardware abstraction layer (HAL) to make migration easy.
More details on the new MCX range are available on the NXP website; the company has not yet publicly announced pricing.