The 2.3 seems to have something strange going on though. The distribution doesn't really overlap on the high end the way the 2.4s do. It really seems to me that the 2.4 behaves about as you'd expect under this theory, with a little bit more "mass" in the violin plot towards the bottom end, but with a higher median and similar peak. The 2.3s though are just lower, across the board.
Yeah, I lost my patience waiting for more benchmark data and just ordered the 2.4.