I answered them all with like 20 mins left at the end and got a 750. I was actually very displeased with this score, I did much better on practice tests, perhaps I would have did better had I slept at all the night before. I actually think that the test is so easy that 800 or so should be required just to get in to a phd program and that I should be forced to retake the test (at no charge of course) to go to grad school. By the way, when I read about people who didn't answer all the questions, I realize that virtually everyone who got a score similar to mine is probably much better than me at physics. Taking your time and leaving a few questions blank is a mathematically terrible strategy. A guess has an EV of 0, but it is actually better to guess and add some random variance to your result because the way the percentile ranks come in, getting an extra point is slightly better than losing an extra point is bad for those with average scores. But more importantly even questions about stuff that I had never heard of (and there were lots of these as I had only taken intro physics when I took the test, no quantum, e&m, or stat mech at that time) were usually such that you could eliminate two answers based on basic physics principles. This is worth 1/4 of a point, and an additional 3 or 4 points can be worth 10 percentile points in the middle range.
I don't see what the problem with time on this test is though, 2/3 of the problems require basically no calculations. I actually think the four older tests online were much more time-intensive than the one I took, I think they have dumbed down the questions a little to deal with the time problems. I have been told that the point of the test isn't 100% to see how good you are at physics, because honestly you can't do that in a 3 hour test, but simply to measure if you can think fast under pressure about problems related to physics. And with many of the problems, the point isn't to do long calculations but to see the simple trick/symmetry that allows you to quickly eliminate the other answers. Finally if you are going in to experiment there is probably little to no correlation between GRE performance and actual research performance, other than perhaps your ability to pass the quals at tougher schools.