Offset Voltage...

[Idontcare]

That said I don't have to use that machine for professional use so I don't have to worry about silent data corruption even though I am yet to pick up any signs that would point towards it being a problem

The reality of silent data corruption is that it is not a risk uniquely specific to overclocking.

Undervolted systems are at risk as well.

And worse even still, every computer out there is subject to eventually degrading to the point where silent data corruption transpires even when the system is (and has always been) running at "stock"...and yet we lack the tools to know when this point has been reached.

It is surprising because when we look to where we store our precious data, the hard-drive or SSD, we have the SMART system keeping tabs on degredation, including the use of predictor models and so forth such that many types of imminent failures can be avoided with advance notice.

A consumer-grade CPU validation program really wouldn't be all the difficult to build.

Everest already has the latency tester program which is designed to query every instruction in the ISA. Combine that with a program like LINX running in the background so the CPU is good and hot, and have the instruction latency program also check/confirm the mathematical correctness of the output.

Maybe some company like AIDA64 will pull such an animal together someday.

 

[Puppies04]

Ehm, what I meant was if you're using a -offset you're obviously running a very modest overclock or even undervolting/underclocking in which case vdroop is never a problem because one doesn't use a high vcore in those circumstances.

But I didn't expect that a 4.4 Ghz oc would be (almost) possible with a -offset, although llc ups the vcore. I myself need +offset (without llc) for 4.2GHz and higher.

As I said I imagine it all comes down to the luck of the draw. I was disappointed with my first 2500K because it needed more voltage than a lot of the guys on this forum who had managed to get "golden" chips. As soon as I got to about 4.5-4.6ghz the difference between my chip and theirs became more than +0.1v.

That is why I jumped at the chance of getting hold of a newer 2500K to see if it behaved any better and the difference is night and day, the new one will run solid as a rock at 4.4ghz with a fixed vcore ( i ran it like this just to benchmark it) which would have caused the old chip to fall over at anything over 4.1ghz.

I have no idea if this was just luck or if the later chips were generally better but I would imagine there were better 2500Ks out in the wild than the "golden" ones on this board and I know there were people with chips that required even more voltage than my first one posting here which would point towards the potential difference in voltage required between the "best" and "worst" 2500Ks at even a modest overclock suck as 4.4Ghz-ish being a lot larger than I would have first imagined.

 

[supremor]

Don't mean to hijack the thread but I also have a question regarding offset voltage. I'm running a P8Z77V-Pro like the OP with a 3770k and until recently I was running a very tame 4.0 OC with everything on auto and an offset of -0.075 with load voltage around 1.1v. I decided to bump it up to 4.2ghz and I would consistently fail IBT maximum burn until I got the offset down to -0.04, anything higher and IBT will fail around test 5-10. Up to -0.05 or so it will simply tell me its unstable, any higher and it will reboot. LLC is set to OFF.

What I find odd is that the load voltage remains the same at 1.128v regardless of offset while idle voltage gets as high as 1.18v or so with the -0.04 offset needed for complete stability. I'll attribute this vdroop but it leaves me wondering why it won't pass IBT with a higher offset if it clearly only needs 1.128v for 4.2ghz. I should also mention that it was perfectly stable for daily use and a hefty play session of Far Cry 3 while at -0.055 offset I think it was.

Basically I'm hoping to get my idle voltage as low as possible, all the power saving features are on and it will clock down to 1.6 ghz @ ~0.920v or so but it keeps jumping up to 1.18v whenever it clocks back up to 4.2 in desktop use.

 

[Ed1]

Don't mean to hijack the thread but I also have a question regarding offset voltage. I'm running a P8Z77V-Pro like the OP with a 3770k and until recently I was running a very tame 4.0 OC with everything on auto and an offset of -0.075 with load voltage around 1.1v. I decided to bump it up to 4.2ghz and I would consistently fail IBT maximum burn until I got the offset down to -0.04, anything higher and IBT will fail around test 5-10. Up to -0.05 or so it will simply tell me its unstable, any higher and it will reboot. LLC is set to OFF.

What I find odd is that the load voltage remains the same at 1.128v regardless of offset while idle voltage gets as high as 1.18v or so with the -0.04 offset needed for complete stability. I'll attribute this vdroop but it leaves me wondering why it won't pass IBT with a higher offset if it clearly only needs 1.128v for 4.2ghz. I should also mention that it was perfectly stable for daily use and a hefty play session of Far Cry 3 while at -0.055 offset I think it was.

Basically I'm hoping to get my idle voltage as low as possible, all the power saving features are on and it will clock down to 1.6 ghz @ ~0.920v or so but it keeps jumping up to 1.18v whenever it clocks back up to 4.2 in desktop use.

If I follow your question , what happens is as you raised the multiplier the voltage that the offset is based off went up too (only way to stop is manual set voltage).
Since the idle is constant 1600 it went up with the less -offset .

Anyway your voltage seems fine, very good IMO , I would just leave it .

 

[2is]

I don't think manual voltage would stop that either. Doesn't manual voltage apply that volrage through out the frequency spectrum? That said, I agree with your last line, nothing wrong with the numbers he's through at us. Software readings aren't going to be 100% accurate, especially when we're dealing with voltage incriments of 1/100th of a volt or even 1/1000

 

[Ed1]

I don't think manual voltage would stop that either. Doesn't manual voltage apply that volrage through out the frequency spectrum? That said, I agree with your last line, nothing wrong with the numbers he's through at us. Software readings aren't going to be 100% accurate, especially when we're dealing with voltage incriments of 1/100th of a volt or even 1/1000

Yes, manual will give you one voltage setting no mater the multiplier, clock frequency .There might still be small variance from loads, vdroop .

My point and what I think he was asking is why the offset needs changing with higher multiplier . cause the base voltage will go up for what the CPU thinks it needs at higher OC's .

Thats why for low OC you can use auto w/o issue of high Vcore .

 

[supremor]

Yes I'm aware that my voltages are very low as is, It's not so much the CPU itself I worry about but the board. I have never killed a CPU with voltage but I have killed several boards after a few years of running extra voltage. I can't afford to upgrade as often as I'd like so I need this build to last a minimum of 2-3 years, preferably without the headache of warranty replacements and such.
It doesn't help that I'm the type of person who always thinks of the worst case scenario for everything

In any case I just found it peculiar that IBT would fail under different offset voltages while the actual load voltage was always exactly 1.128v which while being potentially inaccurate was consistent enough to make me wonder.

 

[2is]

You're OC'd to 4GHz which is less than what turbo ramps up to and you're running a negative offset. The load is quite minimal. That said, with such a mild overclock combined with your mind being not so at ease, why even bother with overclocking at all?

 

[Puppies04]

You're OC'd to 4GHz which is less than what turbo ramps up to and you're running a negative offset.

Since when was 4 less than 3.9?

 

[2is]

Since when was 4 less than 3.9?

It's not... My bad, i thought turbo was 4.1... That said, it's still a very mild OC and if he is that worried about it, running default is the best bet.

 

[supremor]

Oh I'm not worried, this is basically stock voltage for IB as is I was merely explaining why I was so interested in keeping voltage to a minimum even when it was so low to begin with.