Don't you need to initialise 12 instances for the i7? It seems that you only have 50% CPU utilisation going on in that test you posted. With speeds stable at 4Ghz (still well above the 3.2 base clock rating for this processor).
I suspect that with 12 threads, the frequency will drop.
Yes, you need 12 threads in a spin loop to fully load the i7. That's why his utilization was only 50%.