I tested with a project of mine that is physics intensive and I got a slightly more stable fps on Chrome. Also slightly better fps, but the precision is reduced and some objects penetrate each other more often. I perceived a general slower velocity in the movement and the distance spring joints seems slightly less stiff.
After reading Arima's post about the influence of the browser screen size in fps, I tried to test it and I confirm that the fps almost double when I resize the browser window to the game's size. On a second run the game stayed in a low fps no matter the size of the window nor the physics behavior version.
When I test physics for a long time I always get extreme slowdown after some time which makes it very inconstant and unreliable. Because it oscillates a lot the fps I'm not much confident in making a game that is heavily based on physics.