
STMicroelectronics has developed a digital signal processor that combines signal processing and AI algorithms with a MEMS sensor for edge processing on a single chip
The Intelligent Sensor Processing Unit (ISPU) combines a Digital Signal Processor (DSP) suited to run binary neural network AI algorithms alongside a MEMS sensor on the same silicon.
The ISPU is a proprietary 16bit RISC DSP with a four stage pipeline running at 5MHz and full precision floating point processing. It operates from 16bit variable-length instructions and 20bit data, and includes a single-cycle 16-bit multiplier. All of this gives a faster interrupt than an ARM Cortex core at 4 cycles rather than 15. It occupies just 8,000 gates to minimise the amount of silicon area alongside the MEMS sensor, and uses just 40Kbits of memory storage.
The device is programmable in C and is extensible (in the chip-design phase) for dedicated instructions and hardware components. It also combines an accelerator for binary neural network models
- First self-learning edge AI sensor for wearables
- STMicroelectronics acquires Cartesiam for edge AI tool
ISPUs will be packaged in a standard 3mm x 2.5mm x 0.83mm package and will be pin compatible with previous ST devices for quick upgrades. ST says the integration in a single device is key, showing a 5-6x saving in power consumption over System-in-Package approaches in sensor-fusion applications.
“While technically challenging, integrating ST’s sensors on the same piece of silicon with our ISPU does improve sensor-based systems from an online experience to an Onlife one. It advances the sensor’s features to speed decision-making by reducing data transfers, enhancing privacy by keeping data local, while reducing size and power consumption, which cuts costs,” said Andrea Onetti, Executive Vice President, MEMS Sub-Group, STMicroelectronics. “Moreover, the ISPU is easily programmable with commercial AI models and can ultimately operate with all of the leading AI tools.”
Related articles
- Battle for the AI toolflow
- Schneider uses AI on ST MCU for people counter
- NanoEdge AI studio optimized for the STM32 dev boards
Other articles on eeNews Europe
- First quantum network for blockchain
- iSystem BlueBox adds RISC-V support
- RISC-V IP provider sees ISO 26262 ASIL-D certification at SGS-TÜV Saar
- First fully integrated SoC for Direct-to-Satellite IoT
