Empirical measurements yield platform constants of approximately -20 μs on Jetson Orin Nano and -86 μs on Raspberry Pi 4 for GPIO-based hardware validation of edge ML inference timing, with 66 μs cross-platform asymmetry and greater day-to-day variability on the Pi.
Com- prehensive analysis of neural network inference on embedded systems: Response time, calibration, and model optimisation
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
eess.SY 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Per-Platform GPIO Overhead in Hardware-Validated Edge ML Inference Timing
Empirical measurements yield platform constants of approximately -20 μs on Jetson Orin Nano and -86 μs on Raspberry Pi 4 for GPIO-based hardware validation of edge ML inference timing, with 66 μs cross-platform asymmetry and greater day-to-day variability on the Pi.