With site isolation, EQT will improve, but not necessarily in a way that visibly improves user website. For example, consider a sample website that has a tiny iframe that does a very limited amount of work. If the EQT without site isolation is x, then the average EQT for the two renderers under site isolation will be x/2 because the renderer for the OOPIF is considered equally weighed as the renderer for the main frame. In reality, though, most if not all of the user input will be processed by the main frame, so the reported EQT should still be x.
Comment 1 by bugdroid1@chromium.org
, Apr 5 2018