> Interestingly, the difference is very close to 1 full screen refresh. I don't know whether or not that's a coincidence.

Some sort of double buffering? To ensure it's consistently one frame, it would make sense to test on a slower computer via the same monitor.

There was some discussion on HN in 2017 regarding latency differences: https://news.ycombinator.com/item?id=15748146

As mentioned in those comments, it could be from the compositor