Perform latency benchmarks

[bpbradley]

Theoretically, this design should produce sub millisecond latency with 1000Hz+ refresh rates, and the host side should be the only limiting factor which would increase this (worst case device side latency should be 250us with the current configuration, near 0 device side latency is possible in the future with an interrupt based gpio driver instead of using the wii controller).

However it would be useful to design some benchmarking tests to compare actual measured latencies in practical environments compared to other solutions.

For instance, a program which listens for hid data and asserts a gpio when it receives a specific button press. The total signal propagation time could be measured (via oscilliscope or logic analyser) by also measuring the level change directly on the monitored button. Theoretically this would give a true measure of latency from the button press all the way to the host hid message unpacking, including latencies introduced on the host OS. I do not believe anyone has made a similar test which would produce empirical results; I have only seen theoretical figures which aren't especially useful in practice.

The raphnet adapter which is considered commercially best right now advertises 6ms latency by default, with as little as 3ms with tweaks. I am confident this solution beats that.