[BeeOnRope]

Uncontended CAS without carried dependendies on the result (almost always the case in this use case) are similar in performace to atomic add on most platforms.

The CAS is the price they pay for contention detection, though it would be interesting to consider solutions which usually use unconditional atomics with only the occasional CAS in order to check contention, or which relied on some other contention detection approach (e.g., doing a second read to detect when the value incremented by more than your own increment).

The solution looks reasonable to me given the constraints.

[o11c]

Part of my point was that "check for contention" is often a silly thing to do in the first place, when you can just avoid it entirely (and with simpler code too).

Admittedly, Java is fundamentally incapable of half of the solutions, but making a simple bump allocator (called once per statistic at startup) over per-thread arrays is still possible.