DEEPX DX-M1 M.2 LPDDR5x2 module: edge AI benchmarks
- Independent test Mario Bergeron
Key results
How to read power: results are compared only within the same model. By efficiency ranks only results whose source publishes a power figure; By speed ranks every result. Measured = measured for this workload. Vendor-measured (typical/avg) = the vendor’s own measurement at typical or average load, not an independent test and not a maximum, so fps/W looks better than at max TDP. Chip, Module or Board TDP = the vendor’s published maximum, not measured. Typical power, not max TDP = only a typical figure is published, so fps/W looks better than at full load. Scope says what the watts cover: Accelerator only (host not included), Module, or Whole system. Max of range = top of a published range, so fps/W is a floor (≥). “— not published” = the source gives no power; we never estimate one.
| Workload | Model · precision | Result | Power | Perf/W | Rank on this model | Grade · date | Source |
|---|---|---|---|---|---|---|---|
| object detection | YOLOv8mIQ8 (INT8) · batch unknown | 135 FPS | 2.76 WAccelerator only, host not includedBasis not stated | 48.91Basis not stated · Accelerator only | #1 of 5 by FPS/W #4 of 20 by speed | D undated | Vendor page · DEEPX AI Modules page, DX-M1 M.2 LPDDR5x2 "AI Performance" table (retrieved 2026-10-17)DEEPX AI Modules page, DX-M1 M.2 LPDDR5x2 "AI Performance" table (retrieved 2026-10-17) |
| YOLOv5mIQ8 (INT8) · batch unknown | 236 FPS | 3.16 WAccelerator only, host not includedBasis not stated | 74.68Basis not stated · Accelerator only | #1 of 3 by FPS/W #3 of 15 by speed | D undated | Vendor page · DEEPX AI Modules page, DX-M1 M.2 LPDDR5x2 "AI Performance" table (retrieved 2026-10-17)DEEPX AI Modules page, DX-M1 M.2 LPDDR5x2 "AI Performance" table (retrieved 2026-10-17) | |
| YOLOv8lIQ8 (INT8) · batch unknown | 93.00 FPS | 3.60 WAccelerator only, host not includedBasis not stated | 25.83Basis not stated · Accelerator only | #1 of 3 by FPS/W #3 of 3 by speed | D undated | Vendor page · DEEPX AI Modules page, DX-M1 M.2 LPDDR5x2 "AI Performance" table (retrieved 2026-10-17)DEEPX AI Modules page, DX-M1 M.2 LPDDR5x2 "AI Performance" table (retrieved 2026-10-17) | |
| image classification | ResNet-50INT8 (vendor quantized) · batch unknown | 1009 FPS | 4.70 WModule Measured | 215● Measured · Module | not in the default fps/W ranking #19 of 50 by speed | C 2026-06-10 (published) | Mario Bergeron, Edge AI Power BenchmarkingMario Bergeron, Edge AI Power Benchmarking (edge-ai-power-p05-deepx-m1), 2026-06-10 |
Ranks compare this board only with boards running the same model. Up to three models per workload; open a rank to see the full same-model table.
Show data — per-workload tables
object detection
| Like-for-like group | Best result | Model · precision · stack | Grade | Power | Source · date | Rank in group |
|---|---|---|---|---|---|---|
| medium (YOLO-m) · b=unknown | 236 FPS | YOLOv5m IQ8 (INT8) · DXNN SDK (Pro compilation) | D | 3.16 W Basis not stated, Accelerator only, host not included | DEEPX AI Modules page, DX-M1 M.2 LPDDR5x2 "AI Performance" table (retrieved 2026-10-17) undated | 2 of 7 |
| xlarge (YOLO-x) · b=unknown | 49.00 FPS | YOLOv8x IQ8 (INT8) · DXNN SDK (Pro compilation) | D | 3.17 W Basis not stated, Accelerator only, host not included | DEEPX AI Modules page, DX-M1 M.2 LPDDR5x2 "AI Performance" table (retrieved 2026-10-17) undated | 1 of 1 |
| large (YOLO-l) · b=unknown | 93.00 FPS | YOLOv8l IQ8 (INT8) · DXNN SDK (Pro compilation) | D | 3.60 W Basis not stated, Accelerator only, host not included | DEEPX AI Modules page, DX-M1 M.2 LPDDR5x2 "AI Performance" table (retrieved 2026-10-17) undated | 1 of 1 |
| unknown · b=unknown | 47.90 FPS | YOLOv9c IQ8 (INT8) · DXNN SDK (Pro compilation) | D | 2.83 W Basis not stated, Accelerator only, host not included | DEEPX AI Modules page, DX-M1 M.2 LPDDR5x2 "AI Performance" table (retrieved 2026-10-17) undated | 35 of 38 |
instance segmentation
No published instance segmentation results for DEEPX DX-M1 M.2 LPDDR5x2 module 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 DEEPX DX-M1 M.2 LPDDR5x2 module yet.
We only show cited public figures — nothing is estimated to fill gaps. Requests tell us what to source or test next.
image classification
| Like-for-like group | Best result | Model · precision · stack | Grade | Power | Source · date | Rank in group |
|---|---|---|---|---|---|---|
| mid-size CNN · b=unknown | 1009 FPS | ResNet-50 INT8 (vendor quantized) · vendor SDK (see source) | C | 4.70 W ● Measured, Module | Mario Bergeron, Edge AI Power Benchmarking 2026-06-10 (published) | 26 of 47 |
semantic segmentation
No published semantic segmentation results for DEEPX DX-M1 M.2 LPDDR5x2 module 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 DEEPX DX-M1 M.2 LPDDR5x2 module yet.
We only show cited public figures — nothing is estimated to fill gaps. Requests tell us what to source or test next.
LLM generation
No published LLM generation results for DEEPX DX-M1 M.2 LPDDR5x2 module yet.
We only show cited public figures — nothing is estimated to fill gaps. Requests tell us what to source or test next.
Grades: A audited MLPerf · B third-party · C community · D vendor. How we rank · JSON