The SBC tier is the deployment class based on single-board computers, with optional attached accelerators.
The SBC tier targets fixed installations with a 5–25 W envelope. Select it from the deployment surface, then choose a runtime; the model alone does not determine the tier. section 11.4section 11.6
Deployment implications:
Use XNNPACK on the ARM CPU as the workhorse; optional Coral or Hailo acceleration brings a narrower operator table. section 11.6
A fixed installation with a Cortex-A78 host and a 5–25 W envelope is a selection case for this tier. section 11.4
How Quantization and Fast Inference covers it
How deep it goes
Explains tier selection and runtime pairing, but stops short of detailed conversion or tuning procedures.
Examples
Uses a fixed 5–25 W installation with a Cortex-A78 host and optional Coral or Hailo accelerator as the selection case.
What it assumes
Requires a defined deployment surface and power envelope; selects the tier from these constraints rather than model identity.
What you can do after
Choose the SBC tier for fixed 5–25 W installations and pair XNNPACK on ARM with optional Coral or Hailo acceleration.
How it connects
Connects SBC deployment to XNNPACK, ARM CPUs, Coral, Hailo, and deployment-surface-based tier selection.
This explainer was generated by Manning's knowledge graph from the text of Quantization and Fast Inference. The section links point to the parts of the book it draws on.