ARM Cortex-M55
Realizes: deterministic low-power ML inference
ML-enabled Cortex-M55 with Helium vector extensions, embedded in STM32U5 and other advanced microcontrollers for energy-efficient inference.
Examples
🔗
STM32U5 Discovery kit ML inference
STM32U5 Discovery board pairs Cortex-M55 and Helium vector instructions to run deterministic ML workloads with low energy.
1) Compile TensorFlow Lite or ONNX model via STM32Cube.AI and emit Helium-optimized kernels; 2) Deploy the binary to STM32U5 SRAM/Flash; 3) Execute Helium-accelerated convolutional inference loops with deterministic scheduling; 4) Transfer results over USB while the core enters low-power modes between frames.
milliseconds
embedded
μJ