Sandy Bridge Safe Voltage Confusion

[Chymerix]

I've been doing a lot of searching trying to find out if there's a confirmed "safe" 24/7 voltage for Sandy Bridge (2600K) and I haven't really found anything consistent.

There are reports that say exceeding 1.38 volts may kill the CPU, but I've also read a lot of threads where people are posting OC results with the voltage much higher than that in some cases.

Then there's my own experience with the Asus P8P67 Deluxe auto tune, which was only able to overclock my CPU to 42x100 before blue screening. Upon checking the voltages at this speed, I saw spikes up to 1.58 V at full load. Horrified, I immediately disabled the auto voltage feature and fixed the voltage at 1.35 (1.352) volts. With extreme LLC, optimized phase control, and 120% current capability, I was able to achieve a stable overclock of 46x100 (4.6 GHz) with the voltage spiking to a max of 1.368 volts at full load.

So my questions are - does anyone really know what the safe Sandy Bridge voltage is? Is the 1.38 V number accurate? If so, why would the Asus board overvolt the chip with its own auto tune which is supposed to be a "safe" and "easy" way to overclock? Am I safe with my voltages (1.352/1.368)?

Thanks

 

[Hogan773]

I've been doing a lot of searching trying to find out if there's a confirmed "safe" 24/7 voltage for Sandy Bridge (2600K) and I haven't really found anything consistent.

There are reports that say exceeding 1.38 volts may kill the CPU, but I've also read a lot of threads where people are posting OC results with the voltage much higher than that in some cases.

Then there's my own experience with the Asus P8P67 Deluxe auto tune, which was only able to overclock my CPU to 42x100 before blue screening. Upon checking the voltages at this speed, I saw spikes up to 1.58 V at full load. Horrified, I immediately disabled the auto voltage feature and fixed the voltage at 1.35 (1.352) volts. With extreme LLC, optimized phase control, and 120% current capability, I was able to achieve a stable overclock of 46x100 (4.6 GHz) with the voltage spiking to a max of 1.368 volts at full load.

So my questions are - does anyone really know what the safe Sandy Bridge voltage is? Is the 1.38 V number accurate? If so, why would the Asus board overvolt the chip with its own auto tune which is supposed to be a "safe" and "easy" way to overclock? Am I safe with my voltages (1.352/1.368)?

Thanks

Well there are lots of posts from the 5ghz club talking about voltage up or past 1.5....so we know that it probably doesn't AUTOMATICALLY kill the chip. But it could be damaging it such that its going to fail in 3 months, or 6 months, or who knows? I for one plan to be conservative in terms of voltage, as I want to keep this PC for years.

 

[Arkainium]

This is the source that came up with that figure:
http://forums.overclockers.co.uk/showthread.php?t=18227647

Whether they're credible or not I don't know, but just about every review of Sandy Bridge that I've read (including AnandTech) have used more than this "safe" voltage in their overclocking results. While I'm skeptical to say the least, I prefer to err on the side of caution since I haven't heard anything official yet and I cannot afford to replace my CPU if I'm wrong.

 

[eternalone]

We still have 2 more full months before the first reports of dead cpu's come rolling in from people not respecting the 1.35 max give it time. LOL. By the way from what I read 1.38v is (absolute) max for these chips, so I wouldnt consider 1.38v safe for 24/7.

 

[richierich1212]

SB is 32nm. 45nm could take over 1.4v easily.

 

[Chymerix]

Why would Asus engineers (who I'm sure are not incompetent) overvolt the CPU by such a large percentage if the maximum safe voltage is 1.38? That's what makes the least sense to me.

 

[bankster55]

Why would Asus engineers (who I'm sure are not incompetent) overvolt the CPU by such a large percentage if the maximum safe voltage is 1.38? That's what makes the least sense to me.

l'll take a guess at it, but dont quote me

The bios is quite complicated and all new, the flash chip is 32MB but the bios itself is only ~2.5MB. The older bios were around 1.25MB. Therefore I conclude the bios uses the traditional 1.25MB for the MS DOS MBR legacy and the other 1.25MB for the EFI GPT stuff. The remainder (the EFI shell) is written in C++ and has the AI functions.
(If anyone knows, would really like to find out if the GPT stuff is also written in DOS)

So the bios in non manual mode is doing all kinda things about power and speeds and voltages in real time - EVEN avail when in Windows.. So I THINK that the AI allows a max combo situation, sees it as bad and instantly self corrects. In other words on the fly config testing and rationalizing.

Would be also curious as to whether it happens with EIST turned off.

Thats my guess, open to other input.

 

[SlitheryDee]

Why would Asus engineers (who I'm sure are not incompetent) overvolt the CPU by such a large percentage if the maximum safe voltage is 1.38? That's what makes the least sense to me.

I wonder the same thing. In the "official" SB overclocking guide that someone here posted a link to, an Asus employee explicitly suggests voltages over 1.38 for certain overclocks.

Here's the link

Generally 4.8GHz stability can be achieved at 1.400 – 1.425 Vcore now with PLL enabled D2 parts exceeding their previous max multiplier the increase in voltage would be 1.425 – 1.450.

3. Increasing the range between 48 to 50x multiplier will generally require a voltage range between 1.40 to 1.500 with a LLC recommended setting of ultra high.
4. Increased range between 50 to 52 (52 generally considered peak max multiplier except for rare 54x parts) will generally require a CPU voltage range between 1.515 to 1.535V with LLC at Ultra High and potential fine adjustments to the CPU skew range.

Overall a key item to note is the best voltage to oc scaling range potential for the turbo multiplier is 1.400 to 1.425 vcore.

Personally, I'm going to err on the conservative side and keep my vcore around 1.30, but I can certainly see how there can be confusion about this. It doesn't help that SB doesn't really seem to have a fixed operational voltage at all. Rather, the voltage is constantly changing with the speed of the processor. It may have been this way with the 1st gen cores as well, but I skipped them so I wouldn't know.

 

[MrTransistorm]

I wonder the same thing. In the "official" SB overclocking guide that someone here posted a link to, an Asus employee explicitly suggests voltages over 1.38 for certain overclocks.

On the same page, someone asked him if 1.4V is safe. Here's what he said in reply:

Yes as noted in my guide. Ideally i would not recommend anything in excess of 1.425 for 24/7 operation while the overall temperature may still allow for additional voltage and even additional frequnecy scaling i do not think it is worth it.

Read into it whatever way you like, but I doubt Asus would allow a statement to be made by one of its reps that could potentially cause premature failures.

We still don't have a direct statement from Intel, and I doubt we ever will. You could probably ask 10 different Intel engineers and get 10 different answers.

Ultimately it comes down to how much you are willing to risk.

 

[Chymerix]

We still don't have a direct statement from Intel, and I doubt we ever will. You could probably ask 10 different Intel engineers and get 10 different answers.

Intel engineers should know to the micro volt how much these processors can take. If the results of people in the real world are deviating from Intel's tests then there's a bigger issue at hand. There's no good reason for the safe max voltage to be mysterious or ambiguous.

 

[Hogan773]

According to my 2600k box - changing multipliers or voltage may void the warranty. Yes I know its total CYA especially on a chip that is "Unlocked to the Max" or whatever the tagline is.......but for this reason I doubt Intel is ever going to come out and officially say "yes guys, you can OC all the way up to 1.5V and beyond that you'll kill your chips". Either the party line will be so conservative that it won't matter, or they'd be taking a big risk should a bunch of chips start blowing up a year from now.

 

[mclaren777]

What would be a good, safe voltage for a 100.0 BCLK with 42x Turbo multiplier?

Would I even need to raise the default value?

 

[Chymerix]

What would be a good, safe voltage for a 100.0 BCLK with 42x Turbo multiplier?

Would I even need to raise the default value?

Probably not.

 

[Castiel]

Clarkdale's didn't like anything over 1.45v so i can assume the same applies to Sandy Bridge. I think 1.4v is an acceptable daily voltage limit.

 

[VirtualLarry]

SB is 32nm. 45nm could take over 1.4v easily.

Not really. My friend's E5200 @ 3.75 degraded with 1.425v (BIOS) / 1.4v (CPU-Z load), in an IP35-E. Now it only runs at 3.625Ghz stable.

 

[Castiel]

Not really. My friend's E5200 @ 3.75 degraded with 1.425v (BIOS) / 1.4v (CPU-Z load), in an IP35-E. Now it only runs at 3.625Ghz stable.

Over how long a period larry?

 

[VirtualLarry]

Over how long a period larry?

After about a year or two. Temps might have had something to do with it, as the HyperTX2 was caked with dust too, we cleaned it out, and I re-tested with Prime95 to re-certify the OC, and that's when we discovered that it had degraded.

He had been getting BIOS beeps while gaming, which is an indication of the overtemp alarm going off.

 

[Castiel]

After about a year or two. Temps might have had something to do with it, as the HyperTX2 was caked with dust too, we cleaned it out, and I re-tested with Prime95 to re-certify the OC, and that's when we discovered that it had degraded.

He had been getting BIOS beeps while gaming, which is an indication of the overtemp alarm going off.

An Overclocked E5200 lasting that long is well worth it

 

[The Sauce]

I have been overclocking and overvolting for years - probably more than 16 systems and not one CPU failure, ever. Current proc is I5-750 @ 4.2 w/ 1.4V. So if this decreases your processor life, look at it this way, if (and that's a major if) this decreases the life of your proc to 2 years, by then the replacement proc will be <$100 and will overclock even better. That's a chance I'm always willing to take.

 

[nyker96]

I've heard most mention max 1,35 for 24/7 work. some say max 1,3. so I guess 1,3-1,35 is ok. but definitely not 1,4 or over.

 

[smakme7757]

So far i've been benching at 5.5Ghz with 1.55v. My 24/7 overclock is 4.5Ghz with 1.265v.

Intels datasheet says 1.52v is MAX although max is not exactly safe i would presume. Soi don't know, the official datasheet sayd 1.52v and some other non Intel people are saying 1.35v. But i presume they are speculating purely on the whole 'It's a smaller die therefore it takes less voltage'.

So who do we believe? After all Intel has changed datasheets in the past

 

[Castiel]

Ivy Bridge is 11 months away.. My processor only has to last that long

 

[pm]

Intel engineers should know to the micro volt how much these processors can take. If the results of people in the real world are deviating from Intel's tests then there's a bigger issue at hand. There's no good reason for the safe max voltage to be mysterious or ambiguous.

As an Intel engineer (but not in any way am I a company spokesperson), I have to say that I appreciate the implied compliment, but the real world is not so clear. Quality and Reliability Engineering (QRE) is a difficult statistical job and Intel does the best to set limits which will ensure a long and reliable life for it's products, but even so every CPU is different. While you could do a statistical analysis and determine a median value down to a microvolt, in the real world it's unclear where any individual part would fall in that broad statistical spectrum beyond a probability.

All that said, I look at people who claim to know the maximum safe voltage for a processor who have no large dataset of information or long term analysis and I personally think these "experts" have no idea what they are talking about.

* Not speaking for Intel Corporation *

 

[smakme7757]

All that said, I look at people who claim to know the maximum safe voltage for a processor who have no large dataset of information or long term analysis and I personally think these "experts" have no idea what they are talking about.

* Not speaking for Intel Corporation *

+1

 

[SHAQ]

As an Intel engineer (but not in any way am I a company spokesperson), I have to say that I appreciate the implied compliment, but the real world is not so clear. Quality and Reliability Engineering (QRE) is a difficult statistical job and Intel does the best to set limits which will ensure a long and reliable life for it's products, but even so every CPU is different. While you could do a statistical analysis and determine a median value down to a microvolt, in the real world it's unclear where any individual part would fall in that broad statistical spectrum beyond a probability.

All that said, I look at people who claim to know the maximum safe voltage for a processor who have no large dataset of information or long term analysis and I personally think these "experts" have no idea what they are talking about.

* Not speaking for Intel Corporation *

Isn't this variation the reason there are different VID's for each chip? My quick and dirty OC is .1v-.15v over VID. There are rarely any good returns over that level and it is specific for each CPU. No need to worry about any absolute values that way and should be very safe. Actually most of the overclock a user will get is the first few hundred Mhz at stock voltage.