Qualcomm / Layer norm and RMS norm
RMSNorm, int8 and int16 on the NPU
Qualcomm QCS6490, Dragonwing RB3 Gen 2
Fail · Real chipRecorded result
Fail on per value correctness, as recorded. Latency and peak memory pass. int8 p50 402 us, int16 p50 446 us
Recorded 2026-09-30
Execution conditions
- Run through Qualcomm AI Hub, QNN SDK 2.50.0, Hexagon v68, OS build 1.6
- int8, 6 of 131,072 values outside tolerance. int16, 203 of 131,072
- Peak inference memory 2.33 MB int8, 4.35 MB int16
- Signed card with the fail recorded, certificate CLV-CDE43-651D0-CE5F1-3E721
Historical result; a new workload requires compatibility checking and fresh Verify.
Use this as your starting point
Save the reference, describe your workload and set the metric that matters. Your project keeps the selection and its recorded conditions together.
Evaluate for my workload →A historical result is not a compatibility guarantee. Your configuration needs a fresh Verify run.
Evidence summary imported from the local registry on October 2, 2026. A summary is not a newly issued proof pack.