In addition i just ran the PPBM6 and kept an eye on temps, avg never exceeded 75 , individual never went over 78 , initially it actually had better temp readings than the benchmark before I OC'd , turing the MPE off it went higher by only 2-3 degrees which seems pretty good.
I am running the Noctua u12 to cool the 3930k.
One weird thing i came accross during this test , is a major difference in CPU temp between Speccy and the Asus Sensor app , here is a screenshot for reference ...
Is there a reason you use the 'strap' method instead of the multiplier method to overclock? Using the 'strap' method entails the risk that when you raise the BCLK, you also raise the PCIe bus frequency and that can be disastrous on additional cards.
The temperatures are pretty high for such a mediocre overclock. If you try the Stress Test from Prime95 you can expect temperatures that are in excess of 80° C and that is pretty high, even with only 1.305V for the CPU. 1.5V for memory is quite OK.
I would suggest you try a BCLK of 100 or maybe 101, and a multiplier of 42 with a VCore of around 1.32V. So no 'strapping'.
As I mentioned this was an automatic overclock which runs a bunch of tests and determines the maximum most stable settings, I only reason know basics of OC.
I changed settins to BCLK 100 and multi to 40 with Vcore at 1.3V .. clock sits right on 4000 give or take , though is this a worry , check out th eVcore sensor reading