When I update to setZeroTimeout in those canvas vs svg demos, the drawing is so fast there is horizontal tearing and the frame rates are in the thousands.
Which means the FPS with setInterval is actually measuring how long it takes setInterval to fire, not how long it's taking to draw under the different methods.
The code in question is a benchmark of how fast two different render APIs work. The only way to do that reliably is to take the question of the timer firing rate out of the equation. I doubt you'd use setZeroTimeout in most realworld environments.
