Originally Posted by d3v0
Awesome, should get a beer or something sometime.
And regarding that post, yeah thats exactly what prompted me to test with the low PLL voltages. Really hoping to have some luck in that 1.55-1.7v sweet spot. But I am doing due diligence and testing a large range of voltages.
I too, recently discovered the wonders of reduced PLL. I'm operating this 4.9 GHz (stable so far) OC with a PLL = 1.550V. It wouldn't POST at 1.500V but it's operating wonderfully at 1.550V. I previously had it at 1.700V and my thermals were higher at 4.9 GHz than I'd have liked. The 1.700V PLL worked great at my 4.7 GHz OC though.
I'm happy that I'm still learning new things about my system, even as I read these threads. :)
Originally Posted by am dew1
Originally Posted by madchemist83
Terrible choice. Sell all of ur ram and get a proper one, just two sticks. U know what's wrong? Voltage is. For sb u want 1.5v ram. Both of ur kits run at 1.65v. Also 2 sticks will run faster then 4.
Thanks for the feedback. I have read many conflicting things about 1.5v vs 1.65v DRAM voltage for these setups...not sure who to believe any more. In any case, I reduced DRAM voltage to 1.5 in BIOS and all seems stable so far and I am running Prime95 at this very moment.
As for populating all the slots, I'm sure you're probably right that having only 2 sticks will be faster, but I'm willing to sacrifice a tiny bit of speed for not having to worry about running out of RAM while editing video as I do. And again, there are a lot of theories about whether or not you will run out of RAM with 8GB except in very unusual circumstances. It's a wacky hi-tech world out there with lots of varying opinions and theories to say the least.
The Sandy Bridge design maximum voltage is 1.650V for RAM. Most manufacturers are making 1.500V DIMMs to cooperate more nicely with SB and there are even a few ultra low 1.250V solutions available.
There's nothing wrong with using 1.650V DIMMs in a Sandy Bridge in some situations, but the 1.50V is a more common alternative now that works well in nearly all situations.
According to Intel, V DIMM - VTT <= 0.500V is the design specification.
VDIMM = Memory voltage
VTT = VCCIO = QPI = Voltage for the integrated memory controller and PCI-E controller.
Intel's VTT Max value is 1.05 +/- 3% which yields a 1.080V absolute maximum, although some people report going as high as 1.250V safely. (Others have gone higher but some have reported long-term problems.)
So with the Intel default values of 1.050V for VTT, and with the delta between V DIMM and VTT needing to be 0.500V or less, you get VDIMM - 1.05 <= 0.500V Therefore VDIMM <= 1.550V This is where people come up with 1.50V for Sandy maximums.
So it's a bit more of a confusing answer. 1.650V is fine for DIMM as long as your VTT is 1.150V or higher. But Intel normally limits VTT to 1.050V, which results in memory limited to 1.550V and manufacturers remaining at 1.500V for a factor of safety.
I hope that helps clear things up.