Overclocking: Does it age the CPU faster?

[Haervii]

Will the cpu have a shorter life span after overclocking?

 

[spamboy]

Yes, the life will be shortened, and will be shortened even more the more you raise the voltage, but by the time it burns out you'll want a new one anyway.

 

[trmiv]

My parents are currently running my old Celeron 300A OC to 450, and that thing has been going strong for a long time now(I think close around a year and 8 months, maybe more). I personally have been running an Athlon 550 OC to 650 since last November, still going strong. I know it shortens the life a bit, but like spamboy said, you probably want a new one by that time anyway.

 

[HellRaven]

Yes it does shorten the CPU's life.

As others have already pointed out, the processor will be quite obsolete by the time it would become a problem.

From what I have read, modern (P3, Athlon) CPU's running non-overclocked could last 10-20 years. If you overclock significantly you could shorten the life to perhaps 5 years, but who would be using a processor from today 5 years from now? Not many people, and the ones that are won't be overclockers since we will have moved on to whatever the cool new shiznit is

 

[corkyg]

I'll go with the group . . . overclocking does stress the CPU. But . . . there is overclocking and there is overclocking. I never raise the core voltage . . . just adjust bus speed and multiplier if applicable. I have never noticed and shortened life because I upgrade about every 8 months to a new CPU and/or mobo.

All my "trickle down" castoffs are still running after several years.

I have never seen the degredation quantified. It seems to be a logical but subjective conclusion.

 

[Lord Evermore]

Overclocking the CPU does not shorten the lifespan, assuming it's stable, as the CPU came off the same wafer as CPU's that were clocked much higher from the factory. There is no physical difference between CPU's of different clock speeds, other than the quality of each die. Only raising the voltage will shorten the lifespan.

Same for memory running at higher than the rated speeds.

However hard drives and the like do seem susceptible to damage from higher FSB.

My first thought for a response: Yes, I've had to buy a walker for mine and it goes through prunes like mad these days.

 

[ToXiCRaGE]

O/Cing = Heat = Faster death for u CPU...... unless u have a damn good cooling device to take place b4 the heat buildup...... in theory with good heat reducing agent u can even prolong an O/C CPU life span

 

[MGMorden]

I agree with most people that by the time it becomes a problem most of use will have replaced the setup. I have my chip overclocked now (even with a voltage a lot higher than the default), but this chip will be replaced within the next 6 monts for sure, with a good chance it'll be replaced within the next 3 months, so the chip dying faster realy doesn't concern me.

 

[Hanpan]

Contrary to popular belief heat is not the number one enemy of your cpu. Heat may cause instability but it is voltage and frequency that lead to a prcessors death. MOst processors die because of electron migration. That is as elcetron goe through the cpu they slowly strip the surrounding material. Finally there is a break somewhere and you begin to see problems. As some of the modern cpu's have 22 = transistors one break may go unoticed but as more occur a cpu begins to die. The flow of electrons is increased by both raising the voltage and increasing the frequency. That's why you must sometimes rasie th voltage to attain a frequency becasue at a higher mhz the electrons are "used up" faster. I am not 100% certain about this but this is how it was explained to me. The best person to ask about this is pm.

 

[Bloodybrain]

Has anyone witnessed the death of a CPU because it was too old? Every other component in the computer will fail before your CPU will, even if it's overclocked. My Cel300A @504MHz is still powering my main box after two years... that's what I call a good purchase .

 

[nippyjun]

Yes, but there is no way of knowing how much shorter it will last. Typically the chip even if overclocked will outlast its need for use.

 

[AngelOfDeath]

Yes in most cases it will. I think - and that I'm not sure of, that the more easy they get there i.e. 700E@933, the longer it'll last. Andif the demand is high on 700Es they have put out every cpu they can to cover the demand and in this proces I think some of the cpus might have been 'underclocked' 933.

What's the point you might ask, well yes mostly it shortens the life, but because of the high demand on i.e. 700E, I think that, some of us might have got these u/c 933, which means that the life time isn't shortened.

Well maybe this is crap, but I don't think so

AoD

 

[Lord Evermore]

Hanpan, electron migration occurs with any processor, even at stock frequency, and the difference between different speeds of the same core is so little with just a frequency overclock as to be unmeasurable. Voltage modifications are more damaging, but electron migration really is still not an actual issue.

 

[Subversal]

I've got a Celeron 366@550/2.2v and it is still running strong after almost 2 years. I agree with everyone... by the time the problem arises you won't even be using the CPU.

 

[Zenmervolt]

O/Cing does shorten the life of a cpu, sometimes significantly. Also, Evermore, PM, who is an engineer for Intel, has stated in several O/C threads that increasing the frequency, and especially the voltage, increase the occurance of electron migration by amounts that can reduce the life of a processor significantly. In addition, that "underclocked" 950 you mentioned will have a significantly greater mean time to failure if it is running at 750. Intel, and most chip companies, determine which chips are set at what MHz by testing and calculating the MTTF for each chip at a givent MHz, so a chip marked 950 and a chip marked 750 will have approximately the same MTTF if run at their advertised frequencies. Because the 950 can take the extra stress that its higher speed produces, it will have a greater MTTF if it is underclocked to 750. Basically, the SpeedStep chip in my laptop would have a greater life expectancy if I were to leave the computer in batttery mode all the time (700MHz) instead of leaving it plugged in so I could crack RC5 at 850MHz. At least, this is how I recall it to have been explained to me by PM.

Zenmervolt

 

[pm]

Increasing the frequency reduces the life of a part. A higher frequency results in higher current and more heat. The effects of electromigration (which is the migration of metal atoms in a wire and is a real problem but not usually the real problem) are directly proportional to increased heat (which increases the vibration of the lattice), and current squared (i^2). If you are running the processor at a higher frequency, the current is higher (since you are charging and discharging wires more frequently) and the heat is higher. Hot-e depends strongly on voltage (V^3, IIRC), current and temperature. PMOS BTI depends strongly on temperature and voltage. These are the primary failure mechanisms.

Intel specs it's CPU's for a minimum of 7 years of reliable operation.

Increasing the voltage has a huge impact on processor lifetime (voltage increases result in an exponential decrease in MTTF).
Increasing the frequency has a substantially smaller effect.
Decreasing either voltage or frequency will extend the MTTF of the part.

Whether or not you care because you will upgrade the chip long before it will die is a whole different arguement and involves a whole bunch of subjective points.

Saying that they all come from the same wafer and thus will all run for the same period of time is incorrect. PVT and other effects across the wafer typically create a Gaussian distribution of expected operating life (once you eliminate early marginal failures from the graph) in addition to the typical Gaussian distribution of frequency. As far as how Intel tests it's chips and determines what will still be running at 850MHz and what won't be in 7 years, I don't think I'll comment on this in order to avoid "negative management attention".

 

[zippy]

In response to pm's post I just have the following to say:

Oh, cool.

Hehe, just kidding. Good info there- heh, it would have been really funny if you didn't work at Intel and were a cashier at a grocery store or something and were reeling off all that good, detailed info.

 

[pm]

Yeah, I've actually been fooling you guys. I'm really a shelf stocker at Walmart.

One more thing on the subject: the 7 year spec number is running under worst-case conditions (I believe it's at 80C for Pentium III's), which none of us are.

 

[TravisBickle]

Yah, don't worry about it. old CPUs are THE most disposable things in existence.
maybe chasing girls gives you a heart attack sooner, but do you want to stay at home shaking your banana?

 

[hagbard]

PM - so, given what you said about the spec being for 80c, what is the value in using anything other than the stock Intel HS/F to bring down temps? Especially for those who plan on replacing the CPU within a year or two anyway? And another question, if I may. Would the damage from overclocking kill the processor outright, or would it produce corrupt data?

...hag

 

[Haervii]

I'm getting the general consensus that my Duron 700@950@1.8v will last a year . BTW, Lord Evermore, are you gay? And if you are, why don't you stop being bashful and ask "Nathan" out?

 

[Wingznut]

I can vouch that pm works at Intel. When I was there the other day, I saw him working as hard as usual, dishing up grub for the techs in the cafeteria.