State of std::intrinsics::unlikely

Hello, I want to ask if std::intrinsics::unlikely is working ?

I hace a C code that has a macro __builtin_expect((x), 0) and greatly improves performance.

Is unlikely working ? If not, any alternatives ?

Thanks !

The relevant issue is

I think it is stuck at the design/bike shedding phase -- there's no "obviously right" design (or at least it wasn't discovered so far), and it seems like no one is motivated enough to push some specific design to consensus.

Not the same but if you're after optimisations of that kind you could also try Profile Guided Optimization.

And what abount llvm ?

I tried with PGO but works worse than without this