dtourolle cbbe67fbd7 Let the probe's clock be its proof, not disable_cpu_ep_fallback ...

The strict flag refused the Hexagon over the ten quantise/dequantise
nodes at the graph's edges that QNN declines by policy, which cost
microseconds. A provider that hands real work to the CPU is slower than
the CPU floor and the timing already rejects it; the tablet measured
2.3 ms on the NPU against a 29.7 ms floor.

2026-09-19 16:13:19 +02:00

..

src

Let the probe's clock be its proof, not disable_cpu_ep_fallback

2026-09-19 16:13:19 +02:00

Cargo.toml

Add dr-inference-engine and route every model session through it

2026-09-19 16:02:37 +02:00