OT, but isn't the entire "per core" licensing scheme completely absurd for many products?
Since we know that in any significant system the effects of latencies in various cache hierarchies, buses, networks, storage, and cross process synchronization makes enormous difference in performance it's easy to see the CPU performance alone is not really a good predictor of system performance.
Thus, with a price calculated per CPU core the solution space narrows significantly, and experience tells us that solutions where per core licensing is involved tend to include expensive, hard to support networking and storage hardware which few would buy if not per core licensing had taken that option away.
If you want to charge per unit of work, then figure out how to do just that, and don't charge for what is effectively a theoretical peak unit of load?