The poor core utilization stems from Bulldozer/Piledrivers shared FP units and an average cache performance. It has already been proven that OS coding (core parking W7 vs. W8) makes virtually no difference in the FX series performance. ...and if you run Cinebench then the FX makes a compelling choice- unfortunately, very few people use Cinebench, and even fewer use Cinebench in isolation. Any meaningful real-world comparison needs to take into account multiple apps running concurrently (see my earlier post for the bit-tech testing). fwiw, the FX 6300 is a better bet (performance per $) than the 8350.