To be fair, in most of these tests hyperthreading did provide a significant benefit (in the general CPU stress test, the hyperthreads increased performance by ~66%). It's just confusing that utilization metrics treat hyperthread usage the same as full physical cores.