Nvidia's RTX Spark has an NPU, I spent a week figuring it out
Comments on “Nvidia's RTX Spark has an NPU, I spent a week figuring it out”
Claim-checked. Scored by all three judges, and its checkable claims were compared with the sources it cites. How far stories are checked
Adam Conway's XDA investigation of the HP OmniBook Ultra 16, which runs Nvidia's RTX Spark, reports that the platform includes a 55 TOPS NPU that Windows exposes as a separate device, and that it has its own registers, clock, and power rail.
Why it ranked
No. 22 of 93 with a score of 0.64. The panel rated it highest on technical depth (0.97) and new information (0.94), lowest on audience breadth (0.45).
Every answer behind this is in the table under Score.
Score
0.64total under rubric v2.2
- Significance0.56
- Reader interest0.80
- Audience breadth0.45
- Practical relevance0.56
- Substantiated0.88
- Technical depth0.97
- New information0.94
- Counts against: Promotional0.04
- Counts against: Overclaim0.19
Each judge’s answer per question (0–1), their mean and spread, and what the question adds to the score (weight × mean).
| Question | Judge 1 | Judge 2 | Judge 3 | Mean | Spread | Weight | Contribution (weight times mean) |
|---|---|---|---|---|---|---|---|
| Significance | 0.52 | 0.45 | 0.72 | 0.56 | 0.27 | +0.24 | +0.135 |
| Reader interest | 0.69 | 0.80 | 0.90 | 0.80 | 0.21 | +0.26 | +0.207 |
| Audience breadth | 0.37 | 0.50 | 0.48 | 0.45 | 0.13 | +0.13 | +0.059 |
| Practical relevance | 0.66 | 0.40 | 0.62 | 0.56 | 0.26 | +0.22 | +0.123 |
| Substantiated | 0.92 | 0.85 | 0.86 | 0.88 | 0.07 | +0.05 | +0.044 |
| Technical depth | 1.00 | 0.92 | 0.98 | 0.97 | 0.08 | +0.05 | +0.048 |
| New information | 0.95 | 0.90 | 0.97 | 0.94 | 0.07 | +0.05 | +0.047 |
| Promotional | 0.05 | 0.05 | 0.02 | 0.04 | 0.03 | −0.19 | −0.008 |
| Overclaim | 0.24 | 0.20 | 0.12 | 0.19 | 0.12 | −0.11 | −0.021 |
| Sum of contributions | +0.635 | ||||||
| Claim-check penalty | 0.000 | ||||||
| Total, as published | 0.635 | ||||||
Spread is the highest minus the lowest judge answer. marks a spread above 0.40: the judges disagree, and the story is marked contested.
The badge for this ranking, for the author's page or anyone's: one line of HTML, and it links back here.
<a href="https://rubric.hrih.si/editions/2026-10-25/22/"><img src="https://rubric.hrih.si/badge/2026-10-25/22.svg" alt="Ranked #22 on Rubric, 25 Oct 2026" width="245" height="28"></a>Claim check
Claim check: 2 matches source · 5 no source given · 1 couldn’t check. Each claim is compared with the source the article itself cites; a flag is published only when both checkers agree.
Matches sourceNumber
Nvidia revealed at its San Francisco event that the RTX Spark NPU is a 55 TOPS accelerator intended solely for Copilot+ features.
The cited source says the same thing.
The article says “it's a 55 TOPS accelerator intended solely for Copilot+ features”
The source says “Erst auf Nachfrage erklärte Nvidia in San Francisco, dass die verbaute NPU 55 TOPS schnell ist und in der Tat nur einem Zweck dient: Copilot+-Features zu beschleunigen”
Source cited by the article: computerbase.de/artikel/notebooks/nvidia-rtx-spark-n1x-test.99679
Matches sourceQuote
Nvidia's TensorRT documentation says INT8 is supported only as a quantization format on Windows on Arm, with computation remaining in FP16.
The cited source says the same thing.
The article says “Nvidia says INT8 is supported as a quantization format while computation remains in FP16”
The source says “Asymmetric quantization through IQuantizeLayer / IDequantizeLayer is supported as a quantization format only. Compute precision remains FP16. Pure INT8 compute is not supported on this platform”
Source cited by the article: docs.nvidia.com/deeplearning/tensorrt/11.4.0/inference-library/dla-windows-on-arm.html
No source givenBenchmark result
On the RTX Spark NPU, ResNet-50 ran in 11.8 ms and MobileNetV2 in 1.52 ms, compared with 42.4 ms and 4.49 ms on the 20-core CPU via ONNX Runtime.
The article gives no link or citation for it.
The article says “MobileNetV2 ran in 1.52 ms and ResNet-50 in 11.8 ms, compared with 4.49 ms and 42.4 ms on the 20-core CPU through ONNX Runtime”
The source says No source to compare with.
No source givenBenchmark result
The NPU runs INT8 correctly but peaks at roughly 6 TOPS, no faster than FP16 (ResNet-50 took 13.6 ms in INT8 versus 11.8 ms in FP16).
The article gives no link or citation for it.
The article says “ResNet-50 took 13.6 ms per image in INT8, slightly slower than its 11.8 ms FP16 result”
The source says No source to compare with.
Couldn’t checkNumber
Nvidia's official DGX Spark porting guide lists DLA in its GB10 'Chip Features' table and marks it 'NO'.
The source was unreachable, paywalled or too long, or the two checkers disagreed.
The article says “contains a "Chip Features" table that lists DLA but marks it "NO”
The source says No passage could be compared.
Source cited by the article: docs.nvidia.com/dgx/dgx-spark-porting-guide/latest/dgx-spark-porting-guide.pdf
No source givenQuote
The DGX Spark's early-boot firmware log records DLA_DIS = 0x1, meaning the DLA is explicitly disabled.
The article gives no link or citation for it.
The article says “DLA_DIS = 0x1 (default is 0x0”
The source says No source to compare with.
No source givenNumber
Microsoft's list of supported Windows ML execution providers, updated October 7, includes NPU providers only for Qualcomm, Intel, and AMD; Nvidia's sole entry is TensorRT for RTX, which runs on the GPU.
The article gives no link or citation for it.
The article says “Microsoft's list of supported execution providers, updated on October 7, only includes NPU providers for Qualcomm, Intel, and AMD”
The source says No source to compare with.
No source givenGeneralization
The evidence indicates the RTX Spark NPU exists primarily because Microsoft requires one for Copilot+, rather than because Nvidia sees it as important to its AI strategy.
The article gives no link or citation for it.
The article says “the evidence points to an NPU that's there primarily because Microsoft requires one, rather than because Nvidia sees it as an important part of its AI strategy”
The source says No source to compare with.
Also covered by
No other coverage was clustered with this story.
Elsewhere
Elsewhere: HN 3 pts, not on front page
Recorded for context only, checked 25 Oct 2026, 05:46 UTC. It played no part in the ranking.