Sandy Bridge maximum vcore for 24/7 use

[Mathemagician]

I have done quite a bit of reading on this and there does not seem to be any definitive information regarding the maximum vcore for sandy bridge that is appropriate for 24/7 use. Opinions seem to vary pretty widely. On one hand there are "guidelines" stating to keep it below 1.35, and on the other hand there are technical reps from a motherboard maker which imply that 1.42 is a fine starting point and that up to 1.52 is ok so long as temperatures are under control. The same source stated that Intel told him/her that 1.57 definitely will cause degradation.

I was wondering if anyone has definitive information on this topic based on hard facts from a reputable source.

I apologize in advance if this issue has been previously settled in another thread.

[Metroid]

1.375v

[Falkentyne]

1.57v *will* definitely cause degradation. It's already happened to C_N and even the author of the UD7 overclocking guide. This is a 32nm chip you're talking about here... The only way you're going to avoid damage at such high voltages is with subzero or LN2.

1.35v is DEFINITELY the upper fully safe limit which shouldn't cause anyone any problems whatsoever, but the question is, is going up to 1.45v going to be safe long term or not. The chips are still too new (they've been out less than a month) to know the answer to that at the moment, so you just have to wait and see if vcores between 1.35-1.45v are going to be safe as long as you keep the temps down.

[Movieman]

1.375v

I'd agree with that..Maybe even 1.4..Maybe
I use 1.325vcore for 4634MHz 24/7 100% load on air with temps in the mid 50's

[deaffob]

Listen to the manufacturers. It's funny how people just come out with all sort of "max number" for each voltages without any proof. It was CPU PPL for 65nm and 45nm CPUs and the next was max i7 vCore and so on. Many experienced OCers proved them wrong by going really high values for months with any degradation with countless setups. Each major manufacturer has their lab and they have far better grasp and knowledge than average joes who blindly state numbers such as 1.35 as "max". Asus and Intel said if you go over 1.55, you'll start to see some degradation.

It's your choice to be over-cautious but just remember this, between 65nm, 45nm, and 32nm, there aren't any differences in max vCore recommendation from Intel. nm differences can't be measured by mere .01 volt.

[DAK1640]

I'd agree with that..Maybe even 1.4..Maybe
I use 1.325vcore for 4634MHz 24/7 100% load on air with temps in the mid 50's

Is that BIOS or load in windows?

[Movieman]

Listen to the manufacturers. It's funny how people just come out with all sort of "max number" for each voltages without any proof. It was CPU PPL for 65nm and 45nm CPUs and the next was max i7 vCore and so on. Many experienced OCers proved them wrong by going really high values for months with any degradation with countless setups. Each major manufacturer has their lab and they have far better grasp and knowledge than average joes who blindly state numbers such as 1.35 as "max". Asus and Intel said if you go over 1.55, you'll start to see some degradation.

It's your choice to be over-cautious but just remember this, between 65nm, 45nm, and 32nm, there aren't any differences in max vCore recommendation from Intel. nm differences can't be measured by mere .01 volt.

I would respectfully disagree. I remember well when the first 45nm Yorkies came out and people treated then the same as the 65nm Kentsfields and we had a rash of dead chips on the forum.
1.4vcore with 1.45VTT killed those chips.
It wasn't one or the other but a combination of both.
The numbers I post are yes, suggestions but they are safe ones and for air cooling. The last thing I want to do is to give info to some guy and see him lose a chip or anything else.
Yes, I'm conservative but the numbers I posted will give you a solid 4600+MHz for 100% load 24/7 use and he did ask for 24/7 numbers.

Is that BIOS or load in windows?

Bios settings..

[st0ned]

I would respectfully disagree. I remember well when the first 45nm Yorkies came out and people treated then the same as the 65nm Kentsfields and we had a rash of dead chips on the forum.
1.4vcore with 1.45VTT killed those chips.
It wasn't one or the other but a combination of both.
The numbers I post are yes, suggestions but they are safe ones and for air cooling. The last thing I want to do is to give info to some guy and see him lose a chip or anything else.
Yes, I'm conservative but the numbers I posted will give you a solid 4600+MHz for 100% load 24/7 use and he did ask for 24/7 numbers.

Bios settings..

Uhm I've run my i920 D0, and it was a crappy chip, with 1.45v vCore and 1.42 VTT daily use stable at 4472Mhz ( max it did... ) for like a year and a half ( on WC so temps where ~70ºC ) , that CPU is atm, stable in another build with 1.3v at 4.2Ghz on air, and it still's OK.

I'm going to stick with 1.425v for my 2600k that's the voltage for 5Ghz. Also with the new C1 state it will only achiev that under heavy load, so even safer.

But I see your conservative side of the question. I would also bet that voltagens under 1.4v would be, in any condition, safe as long as the temps are ok.

In to the nm question, I see plente of 980x runing at >1.4v, they also are 32nm and they seem ok.

2600k are a hard bone to crack, I can tell that from my own expertise.

[Movieman]

Uhm I've run my i920 D0, and it was a crappy chip, with 1.45v vCore and 1.42 VTT daily use stable at 4472Mhz ( max it did... ) for like a year and a half ( on WC so temps where ~70ºC ) , that CPU is atm, stable in another build with 1.3v at 4.2Ghz on air, and it still's OK.

I'm going to stick with 1.425v for my 2600k that's the voltage for 5Ghz. Also with the new C1 state it will only achiev that under heavy load, so even safer.

But I see your conservative side of the question. I would also bet that voltagens under 1.4v would be, in any condition, safe as long as the temps are ok.

In to the nm question, I see plente of 980x runing at >1.4v, they also are 32nm and they seem ok.

2600k are a hard bone to crack, I can tell that from my own expertise.

Best of luck to you!
Something that I should have added:
What we do ourselves just affects us but thats different than offerring advice to others.
When I tried to max the chip I used 1.5vcore but that was at a shot over 5GHz and MY chip and MY potential loss not someone elses.

[sin0822]

I say 1.45v.

[Falkentyne]

I have to agree with movieman.
I myself wasted hundreds of dollars degrading P4's (which were KNOWN to be very, very fragile), even running them as low as 1.625v on air. And everyone and their momma was saying that as long as you don't go higher than 1.7v (on air), you're fine. Yet just because *some* overclockers had no problems, many others had chips die or refuse to even work properly at stock or had degrading overclocks, even though the intel docs said ABSOLUTE MAXIMUM was 1.75v. So many people took absolute maximum to mean the same as max 24/7 air voltage...

And I remember how a few members on these forums (who don't post here anymore) mercilessly flamed and insulted me for being incompetent and stupid (when 1: I'm older than them, 2: I've been involved in this industry since 1994). And to make things worse, not only was I using high end air cooling (for the time) but also was not using any sort of vdroop mod.
People simply assume that just because THEY have no problems, that NO ONE ELSE will have problems either, and that's a common logical fallicy (That you usually learn about in school).

None of the people who insulted my "incompetence" offered to buy me another CPU. And it's very ironic that as the months went on, I was proven right by many other people who had these chips...others experienced degradation to various degrees as soon as they went over 1.55v.

I remember all too well what happened with the Yorkies. Dead chips, chips not stable at 3 ghz, refusing to POST in one board but posting in another...various things....and it always starts with 1 person claiming "omg it's safe", then another person trying the same settings and his chip dies...

I remember Benchzoner running yorks at 1.5vcore and 1.4 VTT 24/7 and not having them degrade or die, but that doens't mean that YOURS won't degrade or die...it's always YMMV, the instant you go out of specs.

And Movieman:
There were also people killing 980X's by treating them the same as D0 i7 920's...pumping 1.45v-1.5v into them and having them die...not many, of course..but it did happen...

[st0ned]

Best of luck to you!
Something that I should have added:
What we do ourselves just affects us but thats different than offerring advice to others.
When I tried to max the chip I used 1.5vcore but that was at a shot over 5GHz and MY chip and MY potential loss not someone elses.

Thank you

And I do agree with you, If I didnt show it I'll now. I do agree with your voltages, 1.35v must be totaly safe. This doesn't mean you can't go higher for a daily use.

With this chips and with >1.5v you will se some degradation, I'm sure of that, but also as Sin sad, 1.45v in a good water setup that could handle temps well seems a good clock/degradation ratio in my point of view.

My chip runing with my WC settup will hit ~70ºC with 1.424v, and I do have a good WC setup. So take care of those temps.

And Movieman:
There were also people killing 980X's by treating them the same as D0 i7 920's...pumping 1.45v-1.5v into them and having them die...not many, of course..but it did happen..

That might be me. From the 980X part I can't talk much because I cant afford one. Also about the chips beeing death Intel actual RMA policies are protecting us big time, hope they last...

[G H Z]

There isn't a single number that answers your question. Each piece of silicon is unique and will respond to voltage differently, that is why VID numbers exist and are given to each CPU. It would be more accurate to ask what is the percent of over-voltage that's safe for 27/7 use, and even then, it's not going to be exactly uniform between different chips.

[Movieman]

I have to agree with movieman.
I myself wasted hundreds of dollars degrading P4's (which were KNOWN to be very, very fragile), even running them as low as 1.625v on air. And everyone and their momma was saying that as long as you don't go higher than 1.7v (on air), you're fine. Yet just because *some* overclockers had no problems, many others had chips die or refuse to even work properly at stock or had degrading overclocks, even though the intel docs said ABSOLUTE MAXIMUM was 1.75v. So many people took absolute maximum to mean the same as max 24/7 air voltage...

And I remember how a few members on these forums (who don't post here anymore) mercilessly flamed and insulted me for being incompetent and stupid (when 1: I'm older than them, 2: I've been involved in this industry since 1994). And to make things worse, not only was I using high end air cooling (for the time) but also was not using any sort of vdroop mod.
People simply assume that just because THEY have no problems, that NO ONE ELSE will have problems either, and that's a common logical fallicy (That you usually learn about in school).

None of the people who insulted my "incompetence" offered to buy me another CPU. And it's very ironic that as the months went on, I was proven right by many other people who had these chips...others experienced degradation to various degrees as soon as they went over 1.55v.

I remember all too well what happened with the Yorkies. Dead chips, chips not stable at 3 ghz, refusing to POST in one board but posting in another...various things....and it always starts with 1 person claiming "omg it's safe", then another person trying the same settings and his chip dies...

I remember Benchzoner running yorks at 1.5vcore and 1.4 VTT 24/7 and not having them degrade or die, but that doens't mean that YOURS won't degrade or die...it's always YMMV, the instant you go out of specs.

And Movieman:
There were also people killing 980X's by treating them the same as D0 i7 920's...pumping 1.45v-1.5v into them and having them die...not many, of course..but it did happen...

I remember, it's not that long ago..
Now this is just my gut feeling with no proof BUT to me it seems like the SB chips are "tougher" than the 980-990's( and those two are different from each other) The 980's seem to be very fragile IMHO
These SB 2600K's seem to do 85-90% of max with mild voltage( under 1.4) and if you have a good boiard, good memory and "know your stuff" then you may see the "Promised Land" of 5400+
Now I also need to add that I'm no bencher but I do have that "I wonder what this sucker will do" inside me but I don't spend the time that many will to get that last 100MHz..
I got to 5088MHz at 1.5vcore,ran SP1m, smiled and then spend time for 4600+ for 24/7 work.

[flesheatinvirus]

I would agree they do seem tougher then normal. I have been beating up my Sandy a lot and it still runs prime stable at 5G 1.38v and will bench at 5600 just fine. Cooling is everything and a max safe vcore can certainly be more then normal provided temps are carefully regulated.

[Movieman]

I would agree they do seem tougher then normal. I have been beating up my Sandy a lot and it still runs prime stable at 5G 1.38v and will bench at 5600 just fine. Cooling is everything and a max safe vcore can certainly be more then normal provided temps are carefully regulated.

What are you using?
Keep in mind all my numbers are for air..TRUE w/113cfm Delta in open case

[OC Nub]

I am going to play it safe for now (other than a few benches) and run 4.6ish. Anything above x46 multi and Internal PLL needs to be enabled to boot with this chip, until more is learned Im not comfortable enabling that for 24/7. If I feel like abusing a chip Ill get back on the x58 board and torture my 920 some more.

[Hondacity]

I say 1.45v.

how many chips have you used?

[deaffob]

I have to agree with movieman.
I myself wasted hundreds of dollars degrading P4's (which were KNOWN to be very, very fragile), even running them as low as 1.625v on air. And everyone and their momma was saying that as long as you don't go higher than 1.7v (on air), you're fine. Yet just because *some* overclockers had no problems, many others had chips die or refuse to even work properly at stock or had degrading overclocks, even though the intel docs said ABSOLUTE MAXIMUM was 1.75v. So many people took absolute maximum to mean the same as max 24/7 air voltage...

And I remember how a few members on these forums (who don't post here anymore) mercilessly flamed and insulted me for being incompetent and stupid (when 1: I'm older than them, 2: I've been involved in this industry since 1994). And to make things worse, not only was I using high end air cooling (for the time) but also was not using any sort of vdroop mod.
People simply assume that just because THEY have no problems, that NO ONE ELSE will have problems either, and that's a common logical fallicy (That you usually learn about in school).

None of the people who insulted my "incompetence" offered to buy me another CPU. And it's very ironic that as the months went on, I was proven right by many other people who had these chips...others experienced degradation to various degrees as soon as they went over 1.55v.

I remember all too well what happened with the Yorkies. Dead chips, chips not stable at 3 ghz, refusing to POST in one board but posting in another...various things....and it always starts with 1 person claiming "omg it's safe", then another person trying the same settings and his chip dies...

I remember Benchzoner running yorks at 1.5vcore and 1.4 VTT 24/7 and not having them degrade or die, but that doens't mean that YOURS won't degrade or die...it's always YMMV, the instant you go out of specs.

And Movieman:
There were also people killing 980X's by treating them the same as D0 i7 920's...pumping 1.45v-1.5v into them and having them die...not many, of course..but it did happen...

Actually the minority here are the overclockers you've mentioned. Any OC community's data is limited, uncontrolled, and unreliable compared to a manufacturer's lab results. Asus stated that they tested hundreds generation 2 I7 for their new motherboard lineup. I'm not even going to mention how many did Intel test with, probably tens of thousand Sandy Bridges.

I simply stated that instead of trusting any one overclocker, it is better to trust what the manufacturers said. Asus and Intel said the degradation starts around 1.55. But if you look at their recommendation, it hasn't changed much since 65nm. I sincerely disagree with people saying that just because its processing tech is even more micro-scale, it could only take less voltage. This is false, as nm scale can not be measured with .01 voltage differences.

[Falkentyne]

I definitely agree with the SB being a tougher chip than the 980X. I mean, with LLC disabled, my SB was priming at 4.5 ghz (!) at auto voltages...I forgot what the BIOS setting was (I think auto gave a VID of 1.350?), but I think CPU-Z reported 1.26 under load; I know it was 1.2 something under load.

I mean, 1.2x for 4.5 ghz...
Yet the i7 980X 32nm required at least 1.4v for many people that reached 4.5 ghz on air...I guess that's logical since the 980X is simply a normal core i7 with different uncore ratio and 2 more cores on a process shrink...but that shrink did something to make the chips less durable I guess...no one had problems with 45nm i7's at 1.45, besides the temps...that's what held people back... this is still better than the tissue paper P4 days, but thank god I didn't buy the $1000 980x or 990x like I was originally going to do early this year....(to upgrade from the core 2 quad).

We'll see how things go long term. I'm at 4 ghz 1.26v, LLC off, which is good enough for 24/7... 5 ghz at 1.39v BIOS, while fully game stable (havent tried W7 SP1 yet) is just unnecessary at the moment...

I think people need to sit back and realize that we're at 5+ ghz on air now for some chips...not too long ago, 5 ghz on air was reserved for suicide shots...definitely fun times for upgrading...

The bad part is the various motherboard issues people are having...Gigabyte having cold boot loop+reinstalling devices issues, Asus having freezing problems with screeching on some boards...(haven't seen this happen on the Gigabyte side...)

[burebista]

I was wondering if anyone has definitive information on this topic based on hard facts from a reputable source.

I dunno if it's reputable or definitive but readme.

[Falkentyne]

I think the 1.52v the Intel rep is saying is that the chip can USE this vcore under specification, but you had better have some high end cooling to handle that. Perhaps they realize that people are using water and sub zero a lot now, and they think that the only people who will want these documents are people who actually build and bench systems.

I personally do NOT believe that 1.52v is the MAX SAFE vcore on AIR, as the temps just skyrocket. You need chilled water or subzero....

All he said was 1.52v MAX, not max ON AIR.
So in other words, if you stay under 1.52v, your chip will still remain under warranty. Doesn't mean you can RMA it to Intel if the OC degrades...

[CedricFP]

I dunno if it's reputable or definitive but readme.

Any dolt could read from a manual...

... I doubt that Dago has any more knowledge than a child would reading the manual, considering he quotes only the manual and doesn't even begin to discuss cooling, 24/7 or the fact that SB is on a smaller process.

[ugotd8]

Any dolt could read from a manual...

... I doubt that Dago has any more knowledge than a child would reading the manual, considering he quotes only the manual and doesn't even begin to discuss cooling, 24/7 or the fact that SB is on a smaller process.

What about the engineers that wrote the manual. I don't consider them dolts, as I'm sure no one else does either. They have engineered a marvel of performance/thermal efficiency that will go down as one of the best chips in Intel's history.

Intel has designed this chip to be overclocked. There are thermal limits built in, and none of us are coming close to the 90c throttle stop. They say 1.52 is safe, most are no where near that 24/7.

I for one, am overclocking on temps with this amazing platform.

And if it degrades or burns up, I'll send it back to Intel and tell them my grandmother overclocked it too far. ;-)

[JumpingJack]

I dunno if it's reputable or definitive but readme.

Reading through that exchange, I don't think the Intel Rep knew WTF he was talking about.