The object header overhead was always present and expensive in massive arrays - the classic example would be the Point class that has two "double x, y". Realistically the value classes make sense most (only) if they are placed in arrays. In order to use tons of points you'd end up having two arrays, double[] x, double[] y... or go even with direct buffers.

Personally, I don't care about the tiny optimizations possible in cases of just using few of them, e.g. Integer, as int[] is an option, and even writing custom maps where the keys are placed the said array ain't difficult.

As for performance, often times I had to PrintAssembly (the article mentions that at the bottom) to ensure the compiler did its bests, e.g. optimizing away boundary checks, inlining calls, etc.