sizzling :
I have seen a 200mhz difference in stable OC on an Intel build just by using a top quality psu compared to a low end unit.
Except OP's question isn't so much about OCing as it is about why a 4.5% bump in frequency yielded a 1% net
decrease in benchmark results.
Yes, a better quality PSU may improve achievable stable overclock frequencies, but it should not affect benchmark results at a given frequency.