HWBenchedge-AI hardware data, cited & graded

Radxa Dragon Q8B (as tested by Jeff Geerling): edge AI benchmarks

(independent test)independent test system · no photo yet
No licensed photo available — placeholder drawing.
Vendor (independent test)Form factor independent test systemCompute CPUIndicative price unknownPower envelope unknownPublished rows 1
LLM generation12.30 tok/sC 0.506 tok/s/W measuredundated

Best published result per workload, across model sizes — open “Show data” for like-for-like groups.

Show data — per-workload tables

object detection

◌

No published object detection results for Radxa Dragon Q8B (as tested by Jeff Geerling) yet.
We only show cited public figures — nothing is estimated to fill gaps. Requests tell us what to source or test next.

instance segmentation

◌

No published instance segmentation results for Radxa Dragon Q8B (as tested by Jeff Geerling) yet.
We only show cited public figures — nothing is estimated to fill gaps. Requests tell us what to source or test next.

pose estimation

◌

No published pose estimation results for Radxa Dragon Q8B (as tested by Jeff Geerling) yet.
We only show cited public figures — nothing is estimated to fill gaps. Requests tell us what to source or test next.

image classification

◌

No published image classification results for Radxa Dragon Q8B (as tested by Jeff Geerling) yet.
We only show cited public figures — nothing is estimated to fill gaps. Requests tell us what to source or test next.

semantic segmentation

◌

No published semantic segmentation results for Radxa Dragon Q8B (as tested by Jeff Geerling) yet.
We only show cited public figures — nothing is estimated to fill gaps. Requests tell us what to source or test next.

speech-to-text

◌

No published speech-to-text results for Radxa Dragon Q8B (as tested by Jeff Geerling) yet.
We only show cited public figures — nothing is estimated to fill gaps. Requests tell us what to source or test next.

LLM generation

Like-for-like groupBest resultModel · precision · stackGradePowerSource · dateRank in group
LLM ≤4B · b=112.30 tok/s CPU-onlyLlama 3.2:3b
as in linked issue (Ollama default tags are 4-bit) · Ollama / llama.cpp (see linked issue)		C24.30 W measuredJeff Geerling ai-benchmarks README (Findings tables)
undated		56 of 64
How to read these numbers
Every number is labelled. Confidence: A MLPerf closed, audited · B documented third-party method · C community / review · D vendor-published, not independently verified · E estimate. Power basis: measured with the result · spec vendor spec/envelope · power n/a not published.

Machine-readable: /find.json (see llms.txt); full tables are in “Show data”. Confidence grades: A audited MLPerf closed · B documented third-party · C community · D vendor-published · E estimate. Unknown values are shown as unknown, never estimated. Generated 2026-10-16 from public sources.