How much quicker does overvolting kill your CPU?

[Ed1]

Not sure if you were talking about me but I am using the Msi z77a-g45. My offset is 87%. How does that equate to voltage or am I misunderstanding that part?

Ah, sorry, I got mixed up with OP when I was replying to you .

 

[XiandreX]

Ah, sorry, I got mixed up with OP when I was replying to you .

how does the offset work?

 

[Ed1]

how does the offset work?

Hi . I not sure on your statement of 85% .
Offset is a voltage amount that either + or - to the VID voltage .
rough formula is Vcore - VID = offset voltage

to get VID value under load use core temp

example say you were testing under manual voltage and found at 4.5ghz your manual Vcore of 1.25 was stable . Now at load you found the VID was 1.20v

So in this case your offset would be 1.25 -1.20 = +0.05v

 

[zir_blazer]

Overvolting causes electromigration. The Processor could either degrade its maximum possible overclock, to not be stable even on default values, or worse, BSOD in every possible scenario if got irreparabily damaged.

 

[XiandreX]

Hi . I not sure on your statement of 85% .
Offset is a voltage amount that either + or - to the VID voltage .
rough formula is Vcore - VID = offset voltage

to get VID value under load use core temp

example say you were testing under manual voltage and found at 4.5ghz your manual Vcore of 1.25 was stable . Now at load you found the VID was 1.20v

So in this case your offset would be 1.25 -1.20 = +0.05v

Sorry for that, guess I worded it incorrectly.
I am talking about the Vdroop Offset Control. Mine is set at +87.5%.
How does that affect the CPU voltage if at all?

 

[Ed1]

Sorry for that, guess I worded it incorrectly.
I am talking about the Vdroop Offset Control. Mine is set at +87.5%.
How does that affect the CPU voltage if at all?

I am not familiar with Vdroop offset control , must be setting in MSI but not Asus .

Most MB you use LLC (CPU load line calibration ) , if your MB has different name my guess it will work the same as Vdroop is when you have say 1.250v set in bios under manual (to make example easy) , now in windows say you speed is 4.5mhz and when you load CPU to 100 you get 1.220 . So here your Vdroop dropped .030v .
This is done by Intel AFAIK to stop any load spikes which are bad for chips . So by raising the LLC value a bit you can limit the amount of Vdroop that happens .
Say you set to high (which on Asus is a mid value) it now may do 1.245v with 100% load .

One thing you don't want is to overshoot the voltage and cause spike .

Do search here there where some post on here with graph showing how Vdroop works (voltage to time) .
http://forums.anandtech.com/showthread.php?t=2318675&highlight=what+vdroop

PS: some MB you can add voltage to only the turbo boost speeds , so that another way to boost voltage w/o affecting idle voltage .

 

[Ratman6161]

Your CPU is more likely to become obsolete than suffer any ill effects from overclocking. Motherboards are another story. They definitely seem to die more quickly on me if I am overclocking them.

yup, I've got a number of old, perfectly good but also perfectly useless CPU's. Got a K62, a K63+, an old 1200 Mhz Thunderbird, a Barton 2500+, a single core athlon 64, two dual core Athlon 64's (you can see I was an AMD guy for quite a while) and some others. All of these were running just fine at the time I turned them off for the last time. Maybe they are "collectors items" now?

 

[Jacky60]

I have AMD Venice '3000', Barton 3200, Intel Pentium 4 and a Thunderbird 1Ghz waiting on a shelf for Armageddon so someone can bother reinstalling them. If you really try I guess you can cook CPU's but I still think they're pretty unlikely to die unless you go mad.

 

[blackened23]

I have AMD Venice '3000', Barton 3200, Intel Pentium 4 and a Thunderbird 1Ghz waiting on a shelf for Armageddon so someone can bother reinstalling them. If you really try I guess you can cook CPU's but I still think they're pretty unlikely to die unless you go mad.

Over-volting causes electromigration, of that there is no dispute. What typically happens when you push too much voltage too long, that a previously stable OC becomes unstable, and you have to either revert to stock or lower the OC slightly. And at that point, you're just playing the waiting game until your next, lower overclock becomes unstable again.

"Cooking" a CPU is rare. Killing a CPU is also rare - electromigration typically causes a chip to randomly BSOD and become unstable at a given OC clockspeed. And it usually takes a period of a year or more for this to happen, and you must push an excessive amount of voltage for it to happen. In my case, it took two years or more in every instance - but there's no set rule for this. Some folks may over-volt with no ill effects for 5 or more years. As I mentioned earlier, it's less of a concern these days since CPUs have improved idle states which allow you to use low voltage during idle operation, which typically is 80% or more of use. So if you use adaptive or offset voltage, that mitigates the risk a _lot_.

With all that said, i'm not trying to scare anyone into not overclocking. I'm just saying, the concern is there, so if you plan on using a chip for 3+ years - it's something to keep in the back of your mind. I still overclock all of my chips to this day, but i'm a little more cautious with regards to using too much voltage - I try to be more conservative in that respect, and I also take full advantage of offset voltage which helps mitigate the risks.