Manual voltage X Offset voltage.

Drangueos

Limp Gawd
Joined
Feb 4, 2012
Messages
231
I have an I7 3770K @4.5Ghz with manual voltage 1.28V - 24/7

If I want to use Offset voltage the peek voltage to be stable is around 1.32v, if I set lower than this sometimes my system crash.

so what is better for the processor, manual 1.28V 24/7 or Offset from 0.95v to 1.32v???


My PLL voltage atm is manually set to 1.45V, under 1.43V it gets unstable, auto voltage was 1.8v and I7 doesn't need anything near that, so I changed 1.5v for most of people is enough.
 
I would take the offset route myself - considering the chip is probably idle a significant percentage of the time it is in use I'd rather have the lower voltage there. 1.32V is not too high for Ivy, so nothing to worry about there.
 
Agreed, I would go the offset route. I'd also not set my PLL just barely above where you get instability. It makes a very marginal impact in temps and power consumption and I wouldn't go lower than 1.6 personally.
 
Agreed, I would go the offset route. I'd also not set my PLL just barely above where you get instability. It makes a very marginal impact in temps and power consumption and I wouldn't go lower than 1.6 personally.

why wouldn't you go under 1.6? any reason for that?
 
Last edited:
For something like PLL which makes almost no difference in temps, I see no reason to run it that far out of spec.
 
I keep reading on the forums about offset voltage and stories about people endlessly trying to get their systems stable with just the right amount of voltage and I keep asking myself; why?

Why go to all this hassle when you could use a fixed turbo multiplier and a fixed voltage, enable the C3 and C6 core sleep states and you are done tweaking. When a core enters C6, the core voltage drops down to nearly zero which is far lower than any value reported by most monitoring software. At the desktop, an overclocked, idle Ivy Bridge CPU can have all 4 cores spending over 99% of their idle time in C6. Ivy Bridge CPUs come out of the deep sleep states quicker than previous generations so there is no noticeable loss in performance going this route.

Some power consumption testing at idle would make for an interesting review to debunk some of the offset voltage myths.

If you want to see how much time your cores are spending in C6 then check out the latest version of RealTemp and grab yourself a cheap Kill-a-Watt meter next time you are at the hardware store. Learning how to minimize power consumption will reduce your idle temperatures and ensure a long life for your CPU.

RealTemp T | I Edition
http://www.overclock.net/t/1330144/realtemp-t-i-edition

Here's a paper from some smart guys at Berkeley that explains the best way to operate a CPU to maximize performance while minimizing power consumption.

Power Optimization – a Reality Check
http://www.cs.berkeley.edu/~krioukov/realityCheck.pdf
 
I keep reading on the forums about offset voltage and stories about people endlessly trying to get their systems stable with just the right amount of voltage and I keep asking myself; why?

Why go to all this hassle when you could use a fixed turbo multiplier and a fixed voltage, enable the C3 and C6 core sleep states and you are done tweaking. When a core enters C6, the core voltage drops down to nearly zero which is far lower than any value reported by most monitoring software. At the desktop, an overclocked, idle Ivy Bridge CPU can have all 4 cores spending over 99% of their idle time in C6. Ivy Bridge CPUs come out of the deep sleep states quicker than previous generations so there is no noticeable loss in performance going this route.

Some power consumption testing at idle would make for an interesting review to debunk some of the offset voltage myths.

If you want to see how much time your cores are spending in C6 then check out the latest version of RealTemp and grab yourself a cheap Kill-a-Watt meter next time you are at the hardware store. Learning how to minimize power consumption will reduce your idle temperatures and ensure a long life for your CPU.

RealTemp T | I Edition
http://www.overclock.net/t/1330144/realtemp-t-i-edition

Here's a paper from some smart guys at Berkeley that explains the best way to operate a CPU to maximize performance while minimizing power consumption.

Power Optimization – a Reality Check
http://www.cs.berkeley.edu/~krioukov/realityCheck.pdf

thx for that, ill have a look.

I already have something like this kill-a-watt to monitor my power consumption
 
Back
Top