GitHub
// Learn

Tutorials.

Step-by-step walkthroughs with real hardware, real numbers, and no skipped steps. The written tutorial series is still in preparation — until then, the documentation covers everything from first boot to the test suite.

T·01 · Available now · in the docs

From zero to your first on-device inference

QEMU + FRDM-MCXN947 · ~30 MINUTES · NO PRIOR ZEPHYR EXPERIENCE

Set up the workspace, run Hello Inference on QEMU without any hardware, then flash the FRDM-MCXN947 over ISP and talk to the runtime through the syn shell.

T·02 · Coming with the Neutron SDK integration

Deploying your own TFLite model to the NPU

MODEL CONVERSION · TENSOR ARENA SIZING · eIQ NEUTRON

T·03 · In preparation

Profiling and optimizing an inference pipeline

syn prof · PER-STAGE TIMING · MEMORY PEAKS

T·04 · In preparation — Phase 3 shipped

Dual-core apps: inference on CPU0, logic on CPU1

SHARED-MEMORY IPC · MAILBOX · ASYMMETRIC MULTIPROCESSING