Accelerate edge AI with graphics, multimedia and neural processing microcontroller
Based on an Arm® Cortex®-M55 core, with frequency up to 800MHz, the STM32N657L0 MCU from STMicroelectronics also integrates a Neural-ART accelerator with 600GOPS, 4.2MB SRAM, and NeoChrom graphics processing, to boost AI at the edge.
To drive up the performance of graphics and multimedia, the STM32N657L0 adds multiple graphics accelerators, in addition to a computer vision pipeline and a multimedia encoder/decoder. The core adds Arm Helium™ vector processing and enables energy-efficient digital signal processing, and also features a floating-point unit (FPU).
To enhance security, the MCU offers TrustZone®-aware support in addition to a memory protection unit (MPU) and a secure boot ROM for secure booting from external interfaces.
The Neural-ART accelerator runs at frequency up to 1GHz, and provides 600GOPS, which optimizes power efficiency with hardware units for deep-neural network (DNN) inferencing. Data flow is optimized and the internal buffer usage and power are reduced by dedicated streaming engines. The accelerator supports on-the-fly weight decompression as well as real-time data encryption and decryption.
Efficient 2.5D graphic processing is provided by the Neo-Chrom graphic accelerator for hardware acceleration for scaling, high-quality interpolation, free rotation, alpha blending, texture mapping, and perspective transformation.
Camera applications are supported by a parallel and CSI interface and an integrated hardware ISP.
Fast JPEG and motion compression and decompression are provided by a dedicated hardware accelerator.
Development is streamlined by the ST Edge AI Suite and by the TouchGFX advanced graphical framework.
The MCUs are offered in six VFBGA packages, ranging from 142 to 264 pins.