A new semiconductor product category has arrived on the electronics design landscape, and it’s termed as intelligent sensor processing unit (ISPU). It combines a DSP tasked with running artificial intelligence (AI) algorithms and a MEMS sensor on the same silicon. STMicroelectronics sums up its value proposition in four Ps: power consumption, packaging, performance, and price.
According to ST, besides putting the AI-centric decision-making on the edge, the ISPU reduces form factor and saves 80% power compared to system-in-package devices. Moreover, the ultra-low-power DSP can be programmed in C, familiar to many engineers. It also allows quantized AI sensors to support full- to single-bit-precision neural networks, boosting accuracy and efficiency in tasks like activity recognition and anomaly detection by analyzing inertial data.
Source: STMicroelectronics
The C-language-programmable DSP is an enhanced 32-bit RISC machine that is extensible for dedicated instructions and hardware components. It offers a full-precision floating-point unit, uses a fast four-stage pipeline, operates from 16-bit variable-length instructions, and includes a single-cycle 16-bit multiplier.
“The ISPU is easily programmable with commercial AI models and can ultimately operate with all of the leading AI tools,” said Andrea Onetti, executive VP of MEMS Sub-Group at STMicroelectronics. The device is packaged in standard 3 mm x 2.5 mm x 0.83 mm packages, and it’s pin-compatible with its predecessors to allow quick upgrades.
Related Content
0 comments on “DSP combines AI and signal processing on MEMS sensor”