Multi-core MCU includes neural networking accelerators for high AI performance
The Infineon PSOC™ Edge E84 features a sophisticated compute architecture with both high-performance and low-power blocks, both backed by dedicated artificial intelligence (AI) acceleration hardware and fast memory.
The PSOC Edge E84 MCU further extends the broad PSOC portfolio of microcontrollers, which features hardware-assisted machine learning (ML) acceleration for use in next-generation responsive compute and control applications.
The PSOC Edge MCUs use the capability of AI to improve end-user experiences by enabling end products to be more intelligent and intuitively usable, lowering the barriers to human-machine interaction and adding contextual awareness. They do so while providing robust privacy and safety protection through embedded Infineon Edge Protect technologies.
The PSOC Edge E84 is based on a power-optimized multi-core architecture. A high-performance block features an Arm® Cortex®-M55 CPU core, including Helium DSP support, paired with an Arm Ethos™-U55 neural processing unit (NPU). A low-power block combines a Cortex-M33 CPU paired with the ultra-low-power Infineon NNLite hardware accelerator for neural networks in always-on ML and AI applications. PSOC Edge family MCUs feature advanced memory resources, with up to 6 Mbytes of system SRAM and 512 kbytes of non-volatile RRAM memory shared across the high-performance and low-power blocks.
The MCU provides an always-on acoustic activity detection capability which enables the operation of a human-machine interface (HMI) with low active and standby power consumption. It also includes support for advanced graphics, voice, and audio, as well as standard communication and timing peripherals suitable for a variety of consumer and industrial applications.
Software enablement is based on the ModusToolbox™ development software suite, including integration with the DEEPCRAFT™ AI Suite solution with off-the shelf DEEPCRAFT™ Ready Models. ModusToolbox is a comprehensive collection of multi-platform tools and software libraries which includes board support packages, a peripheral driver library, and middleware libraries.