Q: Anyone nuked a Sandybridge K CPU?
- Thread starter DB006
- Start date
DB006
New member
Had mine to 1.65v at benching since 6th Jan 11 and all is well, it seems there's only a handful of chips that have died out there.
Yes, common knowledge not to exceed 1.4v on a 24/7 basis, but it's heat kills cpu's not volts. Stick 1.7 through it it'll be ok as long as it's cool (DICE/LN2)
That's really not true. Volts can kill even if heat isn't an issue. If you look around you will see reports of people who ran above 1.4V and having their max clock be reduced over time. You won't see the damage right away, but over time you will.
See electromigration.
SieB
New member
Yeah, but using dice and ln2 is not normal use, if you are benching then short burst of running benches is ok but running at those kind of volts. On an air cooler for prolonged use 24/7 it is going to damage your chip though.
DB006
New member
Henrik
New member
DB006 is right in a sense.. it is heat that kills a CPU. But for us average guys its heat combined with voltage that kills/degrades a CPU. Theoretically you can push A LOT of volts through a CPU if it is cooled down so it becomes Superconductiv. Wich basically is a certain temperature (around -180c depends on the materiel) where the electrons line up with each other passing through the CPU creating zero resistance. Correct me if I'm wrong here but, that's also when you get the cold bug.. the electrons never stop circulating so the transistors can't turn on/off.
So if your CPU is close to the Superconductiv state the electrons moving through the CPU do not damage the electrical leads inside a CPU. Wich they would if the electrons are bouncing around inside the leads and the atoms inside the leads are moving do to heat, slow degradation will accrue.
If your interested and are absolutely bored have a good read through this
Supercunductivity: http://en.wikipedia.org/wiki/Superconductivity
Electromigration: http://en.wikipedia.org/wiki/Electromigration
So if your CPU is close to the Superconductiv state the electrons moving through the CPU do not damage the electrical leads inside a CPU. Wich they would if the electrons are bouncing around inside the leads and the atoms inside the leads are moving do to heat, slow degradation will accrue.
If your interested and are absolutely bored have a good read through this
Supercunductivity: http://en.wikipedia.org/wiki/Superconductivity
Electromigration: http://en.wikipedia.org/wiki/Electromigration
Not killed mine yet, but I have not got the warranty thing yet so not pushing it to far.
My chip will post at 5.5ghz but not load windows without stupid voltage and basically I want to get the warranty before I will try and push it.
Have fried one i7 2600k once.Tried to disable hyperthreading at 5.4ghz @ 1.565v LLC level9,rebooted and it died immediately.It was running 5.2ghz @ 1.545v LLC8 without any trouble.Temps were not the issue,as i use proper wc,and it never passed 71c at the hottest core.Was lucky to get a new 2600k.Had several 2600k/2700k that could do 5.4ghz,but only in a few benches.They are degrading fast if you push more than 1.55v through.My first 2600k was folding 24/7 4.8ghz @ 1.5v for nealy 6 months on water,and when i started benching it could still do 5.2ghz easily.Have a very strong feeling that its the volts that are killing more over heat.
db are you going to keep posting this shit on every forum you are on?
plenty have killed sandy k chips, i have a dead 25k here that looks like it has a dead imc, and its not heat that has killed them but it is voltage.
dropping to sub zero temps only makes it so you can run higher than normal voltage and stay with in working spec, it does not protect the ic from em.
plenty have killed sandy k chips, i have a dead 25k here that looks like it has a dead imc, and its not heat that has killed them but it is voltage.
dropping to sub zero temps only makes it so you can run higher than normal voltage and stay with in working spec, it does not protect the ic from em.
Henrik
New member
Well my point was that if heat would't be present, zero Kelvin (-273Celsius) "em" would not accrue.. so in a sense you could say heat kills.
Electromigration cannot be avoided in a computer processor. Whilst the best superconductors have a critical temperature of ~ -180 C (attainable with LN2) silicon has a critical temperature of ~ -269 C. That's 4 degrees above absolute zero. I work on semiconductors at low temperature and even my closed cycle liquid helium cryostat wont reach that temperature even in ultra high vacuum. Besides if it was superconducting then a normal transistor wouldn't work as a gate anyway.
Henrik
New member
Totally agree with you m8. Maybe I was a little careless when i stated that it would not damage your CPU. Because as you said it has to be superconducting and if it was it would not even work, it would just become a conductor. Assuming you could ever reach that temp. Maybe I'm wrong but is "em" so bad when going below -100?
That's what I meant with "in a sense heat kills" because I think I stated earlier that a transistor would not work at very low temps.
That's what I meant with "in a sense heat kills" because I think I stated earlier that a transistor would not work at very low temps.
Scoob
New member
Running "a bit hot", however you may personally define that, used to be a good warning that maybe you're pushing things too far. However, with Sandy B. being a generally cool chip and many using water cooling we no longer get that warning...potentially.
As an example, my friend bought a 2600k and it's been at 4.8ghz for over 6 months. To get this speed he needed to tread the fine line that is 1.4v, needing more vCore than most as he insisted on OCing with HT enabled. He needed less vCore with HT disabled incidentally, which isn't uncommon.
Anyway, he has a VERY effective WC loop, massive overkill, but that's what he likes - I helped him build it and it's fugly as hell, but most effective. As a result neither his CPU or GPU's for that matter have ever seen 50c under load - even more impressive when you consider he used two 480's at 850mhz core, and stresses his CPU using the rather harsh IBT. I mention this because I wanted to be clear that his system runs very cool even when being majorly pushed.
However, 9 months down the line and despite having hardly warmed up, his CPU will no longer hold 4.8ghz regardless of vCore. He's been forced to drop to 4.6ghz to have any degree of stability. This to me is the classic symptom of maybe having pushed the chip a little too hard for a little too long & running at near or over that critical 1.4 vCore. Electromigration as already mentioned. Well, more extreme electromigration that might otherwise occured over time at lower vCore draw.
My CPU, a 2500k, bought about a week earlier has been at 4.6ghz for 9 months. It was at 4.5ghz on day one. Now, I have inferior cooling in the form of a basic closed loop setup for just the CPU (Antec Kuhler 620) and I've touched near 80c during benching before refining my OC so I only see into the low-70's benching and mid-60's tops in typical gaming. During this I've been under the 1.4 vCore limit, with the odd blip over 1.4 while refining things. Now I can push my CPU as high as 5.0ghz with added vCore, but that pushes it over the 1.4 vCore "safe" limit under load. However I also get the early warning of temps hiting 80c - which is too high for me - due to my inferior cooling. That is what stopped me playing around, it was too hot for me, however my friend with his far superior cooling could of course continue to push on - NO warning & trying even higher vCore as he was still "safe" temperature-wise.
So, in summary, I have a CPU bought within a week of my friend, yet he pushed his a bit harder & now can no longer hold his overclock. Mine on the other hand, on far far inferior cooling is still going fine. Possibly a degree of luck there of course, we all know about the silicon lottery after all, but early on his CPU overclocked more readily than mine. However, he was willing to push the vCore harder than me and never broke 50c which encouraged him to continue to push.
I've posted it in a couple of topics now; with the advent of Sandy B. we've got some very cool running chips. People with good water loops, or massively over-engineered solutions like my friend - which incidentally was originally built for an older generation Q series CPU - seem to not be hitting those scary temps that might have put us off in the past, so continue to push into the realm where shear vCore is what's going to potentially do the damage. Even those on good air cooling can see this happening potentially, such is the cool nature of Sandy B. But I think water loops are where this is most relevant.
In summary, yeah, hot chips are bad (except with salt and vinigar when by the sea lol) but modern chips aint so hot so we really need to watch that vCore. In the early days we were all amazed at how well these chips worked, how cool they ran and how they equally seemed to be able to take high vCore and work fine. However I'm reading more and more about people whose chips have remained cool throughout their lifetime, but have pulled higher vCores and now seem to be having the odd stability issue.
So, while we do hear reports of "dead" CPU's that have been killed from overclocking a little too much and pushing slightly more vCore than we now know to be considered "safe", the real issue, as much as there is one, appears to be a slight reduction in max overclock attainable. Look at it this way (in my friends example) he ran at 4.8ghz reliably for months & got great extra "free" perfomance. Now he's down at "just" 4.6ghz which hasn't impacted his gaming experience in the slightest, though his benchmark scores have dropped slightly.
Moving forward it will be interesting to see if he suffers any further degredation now he's reduced both his OC and vCore - I know for sure his chip still wont see the hot side of 50c regardless.
So, watch that vCore people! Short-term doubtless the chip will work fine, possibly for months, but more and more appear to be having the odd issue as time passes. I'd consider that a warning to take care, as has my friend.
Just to be clear, the odd blip over 1.4v while testing is liekly harmless, unless you are particularly unlucky. It's the sustained higher vCore draw over months that appears to be potentially damaging. I don't game that much at the moment, but I know a few people who will regularly do several hours a day with games that are fairly demading of CPUs. A recent example would be Skyrim, though they've patched the compiler bug that made a CPU work harder than it need now.
Apologies for the long post, back to my cave, see you next time
Cheers,
Scoob.
As an example, my friend bought a 2600k and it's been at 4.8ghz for over 6 months. To get this speed he needed to tread the fine line that is 1.4v, needing more vCore than most as he insisted on OCing with HT enabled. He needed less vCore with HT disabled incidentally, which isn't uncommon.
Anyway, he has a VERY effective WC loop, massive overkill, but that's what he likes - I helped him build it and it's fugly as hell, but most effective. As a result neither his CPU or GPU's for that matter have ever seen 50c under load - even more impressive when you consider he used two 480's at 850mhz core, and stresses his CPU using the rather harsh IBT. I mention this because I wanted to be clear that his system runs very cool even when being majorly pushed.
However, 9 months down the line and despite having hardly warmed up, his CPU will no longer hold 4.8ghz regardless of vCore. He's been forced to drop to 4.6ghz to have any degree of stability. This to me is the classic symptom of maybe having pushed the chip a little too hard for a little too long & running at near or over that critical 1.4 vCore. Electromigration as already mentioned. Well, more extreme electromigration that might otherwise occured over time at lower vCore draw.
My CPU, a 2500k, bought about a week earlier has been at 4.6ghz for 9 months. It was at 4.5ghz on day one. Now, I have inferior cooling in the form of a basic closed loop setup for just the CPU (Antec Kuhler 620) and I've touched near 80c during benching before refining my OC so I only see into the low-70's benching and mid-60's tops in typical gaming. During this I've been under the 1.4 vCore limit, with the odd blip over 1.4 while refining things. Now I can push my CPU as high as 5.0ghz with added vCore, but that pushes it over the 1.4 vCore "safe" limit under load. However I also get the early warning of temps hiting 80c - which is too high for me - due to my inferior cooling. That is what stopped me playing around, it was too hot for me, however my friend with his far superior cooling could of course continue to push on - NO warning & trying even higher vCore as he was still "safe" temperature-wise.
So, in summary, I have a CPU bought within a week of my friend, yet he pushed his a bit harder & now can no longer hold his overclock. Mine on the other hand, on far far inferior cooling is still going fine. Possibly a degree of luck there of course, we all know about the silicon lottery after all, but early on his CPU overclocked more readily than mine. However, he was willing to push the vCore harder than me and never broke 50c which encouraged him to continue to push.
I've posted it in a couple of topics now; with the advent of Sandy B. we've got some very cool running chips. People with good water loops, or massively over-engineered solutions like my friend - which incidentally was originally built for an older generation Q series CPU - seem to not be hitting those scary temps that might have put us off in the past, so continue to push into the realm where shear vCore is what's going to potentially do the damage. Even those on good air cooling can see this happening potentially, such is the cool nature of Sandy B. But I think water loops are where this is most relevant.
In summary, yeah, hot chips are bad (except with salt and vinigar when by the sea lol) but modern chips aint so hot so we really need to watch that vCore. In the early days we were all amazed at how well these chips worked, how cool they ran and how they equally seemed to be able to take high vCore and work fine. However I'm reading more and more about people whose chips have remained cool throughout their lifetime, but have pulled higher vCores and now seem to be having the odd stability issue.
So, while we do hear reports of "dead" CPU's that have been killed from overclocking a little too much and pushing slightly more vCore than we now know to be considered "safe", the real issue, as much as there is one, appears to be a slight reduction in max overclock attainable. Look at it this way (in my friends example) he ran at 4.8ghz reliably for months & got great extra "free" perfomance. Now he's down at "just" 4.6ghz which hasn't impacted his gaming experience in the slightest, though his benchmark scores have dropped slightly.
Moving forward it will be interesting to see if he suffers any further degredation now he's reduced both his OC and vCore - I know for sure his chip still wont see the hot side of 50c regardless.
So, watch that vCore people! Short-term doubtless the chip will work fine, possibly for months, but more and more appear to be having the odd issue as time passes. I'd consider that a warning to take care, as has my friend.
Just to be clear, the odd blip over 1.4v while testing is liekly harmless, unless you are particularly unlucky. It's the sustained higher vCore draw over months that appears to be potentially damaging. I don't game that much at the moment, but I know a few people who will regularly do several hours a day with games that are fairly demading of CPUs. A recent example would be Skyrim, though they've patched the compiler bug that made a CPU work harder than it need now.
Apologies for the long post, back to my cave, see you next time
Cheers,
Scoob.
Running "a bit hot", however you may personally define that, used to be a good warning that maybe you're pushing things too far. However, with Sandy B. being a generally cool chip and many using water cooling we no longer get that warning...potentially.
As an example, my friend bought a 2600k and it's been at 4.8ghz for over 6 months. To get this speed he needed to tread the fine line that is 1.4v, needing more vCore than most as he insisted on OCing with HT enabled. He needed less vCore with HT disabled incidentally, which isn't uncommon.
Anyway, he has a VERY effective WC loop, massive overkill, but that's what he likes - I helped him build it and it's fugly as hell, but most effective. As a result neither his CPU or GPU's for that matter have ever seen 50c under load - even more impressive when you consider he used two 480's at 850mhz core, and stresses his CPU using the rather harsh IBT. I mention this because I wanted to be clear that his system runs very cool even when being majorly pushed.
However, 9 months down the line and despite having hardly warmed up, his CPU will no longer hold 4.8ghz regardless of vCore. He's been forced to drop to 4.6ghz to have any degree of stability. This to me is the classic symptom of maybe having pushed the chip a little too hard for a little too long & running at near or over that critical 1.4 vCore. Electromigration as already mentioned. Well, more extreme electromigration that might otherwise occured over time at lower vCore draw.
My CPU, a 2500k, bought about a week earlier has been at 4.6ghz for 9 months. It was at 4.5ghz on day one. Now, I have inferior cooling in the form of a basic closed loop setup for just the CPU (Antec Kuhler 620) and I've touched near 80c during benching before refining my OC so I only see into the low-70's benching and mid-60's tops in typical gaming. During this I've been under the 1.4 vCore limit, with the odd blip over 1.4 while refining things. Now I can push my CPU as high as 5.0ghz with added vCore, but that pushes it over the 1.4 vCore "safe" limit under load. However I also get the early warning of temps hiting 80c - which is too high for me - due to my inferior cooling. That is what stopped me playing around, it was too hot for me, however my friend with his far superior cooling could of course continue to push on - NO warning & trying even higher vCore as he was still "safe" temperature-wise.
So, in summary, I have a CPU bought within a week of my friend, yet he pushed his a bit harder & now can no longer hold his overclock. Mine on the other hand, on far far inferior cooling is still going fine. Possibly a degree of luck there of course, we all know about the silicon lottery after all, but early on his CPU overclocked more readily than mine. However, he was willing to push the vCore harder than me and never broke 50c which encouraged him to continue to push.
I've posted it in a couple of topics now; with the advent of Sandy B. we've got some very cool running chips. People with good water loops, or massively over-engineered solutions like my friend - which incidentally was originally built for an older generation Q series CPU - seem to not be hitting those scary temps that might have put us off in the past, so continue to push into the realm where shear vCore is what's going to potentially do the damage. Even those on good air cooling can see this happening potentially, such is the cool nature of Sandy B. But I think water loops are where this is most relevant.
In summary, yeah, hot chips are bad (except with salt and vinigar when by the sea lol) but modern chips aint so hot so we really need to watch that vCore. In the early days we were all amazed at how well these chips worked, how cool they ran and how they equally seemed to be able to take high vCore and work fine. However I'm reading more and more about people whose chips have remained cool throughout their lifetime, but have pulled higher vCores and now seem to be having the odd stability issue.
So, while we do hear reports of "dead" CPU's that have been killed from overclocking a little too much and pushing slightly more vCore than we now know to be considered "safe", the real issue, as much as there is one, appears to be a slight reduction in max overclock attainable. Look at it this way (in my friends example) he ran at 4.8ghz reliably for months & got great extra "free" perfomance. Now he's down at "just" 4.6ghz which hasn't impacted his gaming experience in the slightest, though his benchmark scores have dropped slightly.
Moving forward it will be interesting to see if he suffers any further degredation now he's reduced both his OC and vCore - I know for sure his chip still wont see the hot side of 50c regardless.
So, watch that vCore people! Short-term doubtless the chip will work fine, possibly for months, but more and more appear to be having the odd issue as time passes. I'd consider that a warning to take care, as has my friend.
Just to be clear, the odd blip over 1.4v while testing is liekly harmless, unless you are particularly unlucky. It's the sustained higher vCore draw over months that appears to be potentially damaging. I don't game that much at the moment, but I know a few people who will regularly do several hours a day with games that are fairly demading of CPUs. A recent example would be Skyrim, though they've patched the compiler bug that made a CPU work harder than it need now.
Apologies for the long post, back to my cave, see you next time
Cheers,
Scoob.
Your posts are always informative and awesome mate
I have seen a few bad Sandy Bridge CPUs. One of them was just unstable, no benching or overheating, the dealer exchanged it and now the workstation works fine. Another CPU died when a friend of mine gave it close to 1,8v under benching. Maybe it was already bad, or maybe he killed it. My CPU has been on 1.7xv many times (since January 2011), due to high LLC. On daily basis my CPU gets 1,465v (100@ x52, x51, x50, x50) and I havent noticed any degrading. Over the last 2-3 weeks alone my CPU has been benched at 1,61v several times (5-6 hours of benchmarking), and 100% LLC let it get 1.63+ running it at (103@ x53, x53, x53, x53).
GPU@1.348Mhz - 1,3v
GPUMEM@1.724MHz - 1,7v
CPU@5.456MHz - 1,6v
CPUMEM@2.194MHz 8-10-7-24-1T - (Dram 1,75v, VCCIO 1,3v, VCCSA 1,29v)
all voltages are in the red zone, but I am on water and can keep it all cool.
Link
GPU@1.348Mhz - 1,3v
GPUMEM@1.724MHz - 1,7v
CPU@5.456MHz - 1,6v
CPUMEM@2.194MHz 8-10-7-24-1T - (Dram 1,75v, VCCIO 1,3v, VCCSA 1,29v)
all voltages are in the red zone, but I am on water and can keep it all cool.
Link
