I wonder how this performs in different implementations of HTTP/2.<p>I'm happy to see the content side doing more of these kinds of tests. Browser developers can only handle so much and the content folks usually have a better idea of real world content than the browser makers.
I wonder how much of this performance delta is caused by the differences between the 1.1 and 2.0 specs, and how much is just a difference in the maturity of the implementation.