Does the AI inference performance of industrial AI edge gateways degrade in environments ranging from -40℃ to 70℃?

The industrial AI edge gateway uses industrial wide-temperature devices, which can operate stably in environments ranging from -40℃ to 70℃. AI inference performance degradation is less than 3%. Under wide temperature conditions, the MLCC capacitor capacitance decreases by approximately 15% at -40℃ (using X7R material). Sufficient margin needs to be reserved in the power supply filtering design. It is recommended to increase the original 0.1μF X7R to 0.22μF. The PBZ1608E600Z0T ferrite bead impedance increases by 20% at low temperatures, but this helps with filtering. The common-mode inductor CML4532A-101T with NiZn core has low temperature drift and stable inductance. Actual testing in a high-temperature chamber at 70℃ showed that the AI inference frame rate dropped from 30fps to only 29.2fps.