My friend, thanks for clarifying my queries in my previous posts. I'd like to share a "strategy" that I have figured out for my motherboard when doing Offset overclocking.
So, to recap (and for other forum members who may not have caught on our previous conversation), my motherboard (Gigabyte GA-P67A-UD7-B3, with latest official BIOS of F7) is not able to use Offset Voltage Values and Load-Line Calibration Levels together. In short, If I want to use LLC, I need to go with Manual Overclock Method...if I want to use Offset Values (i.e. Offset Overclock Method), the LLC Levels get disabled.
- I have been playing around with a 4.5GHz overclock and comparing between an Offset Value of +0.035V (my previous official submission at Page 1026) versus a lower possible Offset Value (I have just finished a stress test with +0.025V).
- Between the two, the difference as far as "max recorded temperature on hottest core" is just one degree Celsius. The +0.035V Offset Value gave me 75 degress Celsius, while the +0.025V Offset Value gave me 74 degrees Celsius. By the way, for other readers, I live in the Philippines and it's the summer months right now. Average daily weather is between 33 to 35 degrees Celsius (yeah I know, not good weather for overclocking).
- As far as effort involved to yield that one degree of improvement (cost-vs.-benefit in terms of effort), the +0.035V Offset was more efficient for me in the sense that it was "more stable", i.e. off the bat I was able to complete the 12-hour Prime95 Regular Blend Test without BSODs.
- As far as Power Usage (I compared the data on recorded "CPU Package Power"), the difference between the two Offset Values is numerically insignificant.
My conclusion/inference: although the higher Offset Value gives a higher temperature, its impact would not be significant since in the first place, Offset Method of Overclocking is done because one would want the voltages to scale down when the computer is idle. At idle, I see the VCore below 1.0V. Plus, when the computer is on full load (specifically during prolonged stress-testing), the voltage drops at an average of 3-4 voltage steps as observed and already mentioned in previous posts.
I got the hang of it now. I really appreciate this forum and its very helpful members! I would be honest though, that I am going to revert back to a milder overclock (at most, my overclock will be 4.2GHz) as the power consumption (hence electricity bill) versus performance increase is not commensurate for my typical usage scenarios. In other words, my "net loss" will be greater than my "net gain".
But at the end of it all, I am happy and satisfied because the very premise of my venture into overclocking is to learn how this works. It is worth learning and knowing that if I intend to not upgrade for a long period of time, there is this avenue/option of overclocking so as to keep up with the demands of new technology/software (if ever). :)