How Valuable Would Cutting Your ML Models Computation Time By 30% Be?

Hey IndieHackers!

For those of you who are deploying machine learning models to edge devices:

1. What kind of application do you deploy on IoT devices and is latency typically a problem?

2. How much time do you spend optimizing a model for computation time?

3. How valuable would cutting computation time by 30-50% be to you?