How can I elevate the model performance？

[ghost]

The inference cost time is more than 200ms in my device... while I use 4 threads to match the core num... (network input size is 640*480) However , when I use the inference in "SpeedTest", The forward time is less than 100ms, remains 4 thread... Could U share the SpeedTest's source code？ Thank you

[Charrin]

The inference cost time is more than 200ms in my device... while I use 4 threads to match the core num... (network input size is 640*480) However , when I use the inference in "SpeedTest", The forward time is less than 100ms, remains 4 thread... Could U share the SpeedTest's source code？ Thank you

How do you evaluate speed in your device?