Core i-920 C0/C1: 3.8 GHz overclock freezes but only with Prime95 - does it matter?

[Darren Hodgson]

I've been running my Core i7-920 @ 3.6 GHz for several months now with 100% stability using at first a BLCK of 180 MHz, a multiplier of 20x, a VCORE of 1.28750 V and a QPI/DRAM Voltage of 1.3 V then the same voltages but with a BCLK of 200 MHz and a multiplier of 18x due to having upgraded my memory from 6 GB of 1,333 MHz to 12 GB of 1,600 MHz G.Skill Ripjaws (the RAM requires only 1.5 V at its recommended 1,600 MHz speed and that is set manually in the BIOS). Load temperatures never exceeded 65 C during Acronis backups (the only application that stresses all eight cores to 1005) and gaming.

So I figured I'd push my Core i7-920 a little further on the same voltages by setting the multiplier to 19x to give a 3.8 GHz overclock and to my surprised it appeared to be stable. I ran Heaven 3.0 on maxed settings at 1920x1200 for 40 minutes without issue and 3DMark11 was fine as were the handful of games I tested. Feeling confident, I ran Prime95 v27.4 64-bit's default test but my PC appeared to freeze after 25 minutes during Test 8. The screen froze there was no error message or blue screen unfortunately.

Seeing as Prime95 pushes the PC to extremes that software doesn't, does it really matter when everything else appears to be fine? Obviously, my PC is not 100% stable in stress testing but does appear to be fine under normal use, even when I max all eight threads to 100% to using the MaxCPU tool.

I have bumped the VCORE up a little to 1.2950 V though and set the CPU PLL manually to 1.84 V for now and so far my PC has been stable. However, I have not ran Prime95 again.

Comments?