Walt: A device for measuring latency of physical sensors and outputs
github.com
github.com
In most cases it actually takes less time to send a network packet to a server thousands of miles away than it takes to change a single pixel on the local display. Wouldn't it be great if you could measure the display latency as easily as you can run 'ping'?
It may be counter intuitive, but touch-down latency and drag latency can be very different: on some hardware one may be faster or slower than the other. That's why both are measured.
Note: I worked with Mark to review most of his work on this project.
It seems crazy to me that similar things aren't more common. You can't optimize what you can't measure.