Sandy Bridge Processor Degradation

[Smartazz]

I currently have the 2500k in my sig which runs close to 100% 24/7. I'm assuming my settings at 1.3v 4.5GHz wont really degrade the chip, am I correct? Load temps are around 50c.

 

[Kenmitch]

Although there are no guarantees in life other than death and taxes, both your temps and vcore look to be on the conservative side. If it was me I wouldn't worry about the chip degrading over time. Are you hard locked or does your chip throttle down?

 

[Puppies04]

Running a chip at standard settings will degrade it over enough time, overclocking a chip will degrade it faster and running it hotter will degrade it even faster. The question I think you are asking is will running your chip @ 1.3v and 50c degrade it excessivly fast to which the answer is either "not really" or "there isn't enough data out there to answer the question properly"

The best answer i could come up with personally is that there are people out there pumping 1.4-1.5v through their chips at 5ghz and I haven't heard about any of them killing their chips yet so in theory you "SHOULD" be fine but any overclocking comes with an inherant risk of degrading the chip over a lot shorter lifespan than the manufacturer intended. Sorry I can't be more specific but you asked a very vague question and the data is just not avaliable atm.

 

[Puppies04]

Also can I ask what you are using it for to run it at "almost 100% 24/7"?

 

[Smartazz]

I should have clarified my situation more. I'm planning to replace this rig in 2 years with Haswell. I ran a 3ghz overvolted q6600 for several years and only experienced minimal degradation, it took 1.35v instead of 1.33v. Should I expect similar degradation from the 2500k. The q6600 also ran at 60c.

 

[IntelEnthusiast]

Running a chip at standard settings will degrade it over enough time, overclocking a chip will degrade it faster and running it hotter will degrade it even faster. The question I think you are asking is will running your chip @ 1.3v and 50c degrade it excessivly fast to which the answer is either "not really" or "there isn't enough data out there to answer the question properly"

I don't remember where I saw it but I think that the MTBF (Mean Time Between Failures) on our processors is something like 1,250,000 hours or somewhere over 146 years. So while running extreme in tempuratures or high voltages may shorten the life of your processor I would guess that the temps and voltages you posted wouldnt shorten it to the level that you see the difference.

 

[Smartazz]

I don't remember where I saw it but I think that the MTBF (Mean Time Between Failures) on our processors is something like 1,250,000 hours or somewhere over 146 years. So while running extreme in tempuratures or high voltages may shorten the life of your processor I would guess that the temps and voltages you posted wouldnt shorten it to the level that you see the difference.

Haha, that's well beyond the lifetime I expect from the chip.

I've been doing some searches on Google and came across a few threads where people report degradation at lower voltages like 1.3v-1.4v. However, others counter this claim by suggesting that the chips simply "settle in" early in their lifetimes. Can anyone comment on these claims?

 

[know of fence]

I remember reading about this here on AT (link), as the production process gets smaller CPUs get more ephemeral and susceptible to heat degradation.

http://www.anandtech.com/show/2468/6

 

[grkM3]

People get the cpu breaking in mixed up with degration.with every chip iv ever had it always clocked high for the first 2 weeks and then needed a voltage bump.after that first increase my chips run stable for months.

Like my 2600k its seeing 1.44 loaded and it sits around 1.4 idle and its been like this for almost a year.

 

[soccerballtux]

I should have clarified my situation more. I'm planning to replace this rig in 2 years with Haswell. I ran a 3ghz overvolted q6600 for several years and only experienced minimal degradation, it took 1.35v instead of 1.33v. Should I expect similar degradation from the 2500k. The q6600 also ran at 60c.

what motherboard did you have? Did you notice if the 1.35v was 1.35v in CPU-z, or just 1.35v in BIOS? A small voltage like that could be caused by the motherboard VRMs or whatever failing to produce the same voltage. My IP35-E experienced some degradation that would manifest itself in a way that looked like the CPU was doing the degrading-- but I started paying attention to the voltage in CPU-z and noticed it was inching lower from what it used to be. Eventually my 1.48v overclock was dipping to 1.46 under load, which was causing the error.

 

[soccerballtux]

I don't remember where I saw it but I think that the MTBF (Mean Time Between Failures) on our processors is something like 1,250,000 hours or somewhere over 146 years. So while running extreme in tempuratures or high voltages may shorten the life of your processor I would guess that the temps and voltages you posted wouldnt shorten it to the level that you see the difference.

unfortunately MTBF, by nature, doesn't tell you anything about the upper limit lifetime of a product though-- IE nobody has run a processor for 146 years especially with HDDs it's not a useful metric (IMHO) for the consumer. For someone running a server system it makes more sense because you can calculate how much the replacements are going to cost to keep you operating for one year.

 

[Puppies04]

People get the cpu breaking in mixed up with degration.with every chip iv ever had it always clocked high for the first 2 weeks and then needed a voltage bump.after that first increase my chips run stable for months.

Totally agree my 2500k was on for about a month (evenings and weekends) before it needed a slight vcore bump. It hasn't missed a beat since.

 

[Puppies04]

I don't remember where I saw it but I think that the MTBF (Mean Time Between Failures) on our processors is something like 1,250,000 hours or somewhere over 146 years. So while running extreme in tempuratures or high voltages may shorten the life of your processor I would guess that the temps and voltages you posted wouldnt shorten it to the level that you see the difference.

Is that at 100% load and "normal temps" i.e around 50 degrees(for 100% load)? That sounds a little far fetched but if you could link some data i'm sure it would clarify the point.

 

[know of fence]

Funny thing is that, power saving features on processors like variable frequency and voltage identification (VID) probably exist to extend the life of CPUs.

In that article Anand actually puts forth two numbers for a 45nm chip, that starts with 1.15V (Vcore) required to maintain a certain clock rate and degrades to 1.21 over the course of its 6 year life.

 

[Rvenger]

I don't remember where I saw it but I think that the MTBF (Mean Time Between Failures) on our processors is something like 1,250,000 hours or somewhere over 146 years. So while running extreme in tempuratures or high voltages may shorten the life of your processor I would guess that the temps and voltages you posted wouldnt shorten it to the level that you see the difference.

I assume you guys over at Intel are still stressing the old 486/386's and still didn't reach any degradation point?

 

[Smartazz]

what motherboard did you have? Did you notice if the 1.35v was 1.35v in CPU-z, or just 1.35v in BIOS? A small voltage like that could be caused by the motherboard VRMs or whatever failing to produce the same voltage. My IP35-E experienced some degradation that would manifest itself in a way that looked like the CPU was doing the degrading-- but I started paying attention to the voltage in CPU-z and noticed it was inching lower from what it used to be. Eventually my 1.48v overclock was dipping to 1.46 under load, which was causing the error.

Motherboard was a Gigabyte EP45-DS3L with 1.35v in the BIOS. CPU-z was typically lower than what I set.

 

[Smartazz]

Is electromigration what causes degradation?

 

[LOL_Wut_Axel]

Bought my 2600K about three weeks ago. It doesn't go over 70C in Prime95 and is running at a low voltage, so I'm hoping I don't have to change anything in three years at least. I don't think an OC in this range should result in much degradation. Even though a lot of people say under 1.4V is safe, I'm not taking chances. Besides, are higher temps, power consumption and degradation really worth 200-400MHz more when these are so powerful anyway? My 2600K only consumes around 110W when running Prime95.

 

[Smartazz]

Bought my 2600K about three weeks ago. It doesn't go over 70C in Prime95 and is running at a low voltage, so I'm hoping I don't have to change anything in three years at least. I don't think an OC in this range should result in much degradation. Even though a lot of people say under 1.4V is safe, I'm not taking chances. Besides, are higher temps, power consumption and degradation really worth 200-400MHz more when these are so powerful anyway? My 2600K only consumes around 110W when running Prime95.

My rationale is that it's not worth pushing so much harder, raising power consumption and heat output to get to 5GHz when it's only 11% faster. However, if you have a really good chip, I can understand wanting to hit the 5GHz barrier.

 

[FalseChristian]

I had a dud of of a CPU in my E8400. I needed 1.575v to get it to 3.825GHz. Now that is an insane amount of voltage and way over the recommended maximum of 1.36v and it ran fine for 3 years! I just replaced it yesterday with an i5 2500K at 4.5GHz using 1.32v.

 

[BrightCandle]

I ran an i7 920 at 1.35 volts vcore at 4 Ghz for nearly 3 years. About 2.5 years in it would not hold 4Ghz anymore at that voltage and I had to drop it to 3.8Ghz to return to full 24/7 stability. That was a 45nm chip.

At 32nm the default voltage dropped and so did the recommended maximum overclocking voltage. With 45nm Intel recommended 1.4V, with 32nm they have said 1.35V. So if you want about 2 years out of it you are probably OK, as people have already been running like that for a year and no signs of premature failures except for the lunatics.

 

[Smartazz]

I ran an i7 920 at 1.35 volts vcore at 4 Ghz for nearly 3 years. About 2.5 years in it would not hold 4Ghz anymore at that voltage and I had to drop it to 3.8Ghz to return to full 24/7 stability. That was a 45nm chip.

At 32nm the default voltage dropped and so did the recommended maximum overclocking voltage. With 45nm Intel recommended 1.4V, with 32nm they have said 1.35V. So if you want about 2 years out of it you are probably OK, as people have already been running like that for a year and no signs of premature failures except for the lunatics.

When did they say 1.35v recommended with Sandy Bridge? Btw, why do you run a custom watercooled rig at stock? I'm guessing for noise?

 

[Puppies04]

When did they say 1.35v recommended with Sandy Bridge?

They didn't as far as I know, the only voltage I saw intel allude to was the "max" voltage in the product data sheet for the SB cpus which was something like 1.52v. 1.35 is a number that is "assumed" safe by some overclockers, others say 1.4v.

 

[LOL_Wut_Axel]

They didn't as far as I know, the only voltage I saw intel allude to was the "max" voltage in the product data sheet for the SB cpus which was something like 1.52v. 1.35 is a number that is "assumed" safe by some overclockers, others say 1.4v.

It all depends on how much you want to use the chip before you have to bump up the voltage again or lower the clocks slightly. Given the manufacturing process I'd say 1.4V is fine for two-three years or so, and anything under 1.35V should be good for five years at least*. Higher degradation doesn't mean that your CPU will die, but it's good to not have a very high amount of it if you plan on keeping your chip for a long time and at a stable overclock.

*Given reasonable temperatures, of course. Higher temps also increase degradation.

 

[jimpatrick]

Btw, why do you run a custom watercooled rig at stock? I'm guessing for noise?

lol'd at this.


same temp on load,vcore and ghz as OP.wish my 2600k wont need any bump for atleast 1year