TinyEngine NPU powers AI in TI MCUs
TI’s MSPM0G5187 and AM13E23019 MCUs integrate the TinyEngine NPU, enabling efficient edge AI in systems ranging from simple to complex. These latest additions to TI’s portfolio of AI-enabled hardware, software, and tools allow engineers to deploy intelligence anywhere. This announcement moves TI closer to its goal of integrating the TinyEngine NPU across its entire microcontroller lineup.
The MSPM0G5187 is powered by an Arm Cortex-M0+ 32-bit core operating at up to 80 MHz and includes 128 KB of flash. Its TinyEngine NPU is capable of running AI models with up to 90× lower latency and more than 120× less energy per inference than comparable MCUs without an accelerator. By performing neural-network computation locally, the NPU operates in parallel with the primary CPU running application code. Priced at under $1 in 1,000-unit quantities, the MSPM0G5187 brings edge AI to simpler, smaller, and lower-cost applications.
Aimed at real-time motor control, the AM13E23019 leverages an Arm Cortex-M33 32-bit core operating at up to 200 MHz and includes 512 KB of flash. It maintains precise real-time control loops for up to four motors while the TinyEngine NPU runs adaptive control algorithms. An integrated trigonometric math accelerator performs calculations 10× faster than coordinate rotation digital computer (CORDIC) implementations, enabling more responsive motor control.
The MSPM0G5187 is available now in production quantities on TI.com, while the AM13E23019 is currently available in preproduction quantities.
The post TinyEngine NPU powers AI in TI MCUs appeared first on EDN.
