Can overclocking shorten a cards life-span?

[amenx]

I can overclock my XFX 6600gt to 580/1160 with no artifacts and with good temps (max 73c load). But can this affect the health of the card or shorten its life-span?

 

[Matthias99]

Of course. You're undoubtedly raising the temperature of the card more than normal, and that could definitely lower its usable lifespan.

However, it's very hard to know how long it would live to begin with, and it may outlive its utility anyway (that is, by the time it dies even with overclocking, it will probably already be obsolete).

 

[amenx]

But you see, 73c temp load is within the normal range of cards without OC, I can see your point if it was in the 90's. So if temps are the only factor that can shorten its life-span, than I'm better off than many other card owners who dont OC but have higher temps than mine. Would you agree with that?

 

[beserker15]

yeah, as long as you don't mess with volt mods and stuff, i doubt it'll really affect the health much. course, even if it does...you'll probably end up getting rid of your card long before that happens. my geforce 3 ti200 oced to ti500 is still perfectly alive, as with my 9500 oced to 9700pro+.

 

[Munky]

Actually, I'm sure even with a slight voltmod (<10% over volt) the card will stay alive for quite long - for example I've been Oc'ing cpu's for years and never had any die on me, but that's because I dont do suicide runs with voltages that would require phase change cooling. Of course over volting still shortens the life span, but as long as you keep the cooling adequate, and dont do voltmods, I see no reason why OC'ing would shorten a card's life span, which is already longer than you're likely to keep the card anyways.

But in regard to the OP, you're temps are a bit high, and will likely shorten the life of the card. I remember someone mentioning that for every 10 degrees C you raise the temp, you cut the life of the card in half. My card stays under 60C at load, so it's considerably safer than 70C or more.

 

[amenx]

I'm in agreement with you in principle that lower temps will result in a longer lasting card, but are not GPU's made to withstand much higher temps than CPU's? I've seen a lot of people with load temps in the mid-70's so I was thinking that was probably the norm. If yours is at 60 with no extra cooling, then you seem to be one of the more fortunate ones.

 

[Munky]

No, I put a zalman vf700 on mine. With the stock cooler it would also be around mid 70's, and while the card will not fry at 75C like it will at 120C, it will still shorten the life span. I think of it as a car engine analogy - the engine is built to withstand the stress of it's rated power output, but even so, if you always floor the accelerator and run it under full load, you will significantly reduce it's lifespan.

Also, I've seen shoddy cooling installed by the manufacturer, which would put the card under increasd temps even without OC'ing. My old saphire x800 ran ~72C under load at stock speeds - I removed the cooler and saw that they put a extra thick layer of themal paste that was decreasing the coolin efficiency.

 

[Demoth]

The answer is yes, but in most cases it dosen't matter anyways. Vid cards are not TVs or microwaves, they have a relatively short useable period. I have a voodoo card that still works fine, but the only reason I have it is because I have never gotten around to taking out of my junk drawer and chucking it. Most people's Geforce 2 cards work fine, but most will never use them again.

Of course, the newer top end cards may have shorter lifespans then these older cards by default because they draw more power and generate more heat, not to mention being more complex. Usually next gen mid and lower range like a 7800 GT or 6800 GS runs cooler and draw less power and probably could run moderately OCed for a decade with no problem. Rumor has it the X1800XT has an expected lifespan of only a few years (or even 6 months according to Nvidia)) at normal operating parameters, so something like that I'd think twice about OCing too heavily if you plan on using the card for several years.

 

[solofly]

Any overclocking to any component shortens life-span period. Another way to look at it, heat kills.

 

[rbV5]

Google "Electromigration". Its not just heat.

 

[51PEG]

Just wanted to add that XFX graphics cards has double life time guarantee....

But of course overclocking it violates the warranty terms(but how would they know?)

So I wouldn't be worried. I have the exact same card running at 570mhz and 1150mhz and i'm not even worried.

 

[TSS]

OCing will always shorten the lifespan of a card. but if you keep it cooled sufficiently, there shouldnt be much of a problem. heck, i'll bet, if you card runs 70 load with the stock cooler at stock speeds and you change to a aftermarket cooler, which will run the card OCed at 65 load, it'll actually run longer then the stock card.

and its not just heat but thats only if you start messing with the voltages. with graphic cards people usually dont mess with the voltage since you'll have to volt-mod a card (or so i heard) while with CPU's you can control it in the bios.

just keep it cool, and everything should be fine. and yeas, 70's seem to be quite normal for graphics cards these days.

 

[nRollo]

Originally posted by: rbV5
Google "Electromigration". Its not just heat.

QFT

Here's the problem with this post:

You've got a pool of guys who aren't chip engineers answering it. You're literally asking a bunch of guys who have no clue whatsoever.

They want to believe OCing and volt modding doesn't affect the life of the cards, because that would lead to all sorts of problems like accepting financial responsibility for the part when it fails, or advertising the part accurately when they sell it and taking less money.

 

[Keysplayr]

Originally posted by: amenx
But you see, 73c temp load is within the normal range of cards without OC, I can see your point if it was in the 90's. So if temps are the only factor that can shorten its life-span, than I'm better off than many other card owners who dont OC but have higher temps than mine. Would you agree with that?

I haven't owned a 6600GT, but isn't 73C hot as hell for that card? Or is that normal?
I mean, it's only 8 pipes. My GTX has triple the pipes and runs 65 at load, 43 idle.

I know the 6600GT runs at 500MHz core, but thats only 70MHz faster with 1/3? the transistors. That 73C just seems pretty hot to me.

 

[BUrassler]

My 6600gt gets up to 75C upder full load :Q

I am thinking about grabbing a zalman cooler though.

 

[Bobthelost]

Ok, a graphics card may be desinged for a 10 year lifespan. In practice a card (let's call it dave) may live for 15 years before it dies of old age. But if you were to take Dave and overclock it (using stock voltage) you increase the temperature Dave runs at. Meaning it's lifespan drops to say 7.5 years (we're assuming an overclock that generates just enough extra heat to halve the lifetime).

However you don't know that any card in particular would live for 15 years, you only know that the cards are designed for to last for 10. So if you got one that would last 10 years (call it Benjamin) and overclocked it to the same extent then it'd live for 5 years in total.

What happens if the card isn't go great though, it's only going to live for 5 years because the quality control isn't as good as it should be? Result is 2.5 years of lifespan with the same overclock.

Fact is that overclocking increases heat, and heat reduces the lifespan of a chip. If a chip that's overclocked lasts more than it's designed lifetime then the odds are that i'd live even longer if you hadn't overclocked.

The cards may be rated to run at 80*C but it will mean that thier lifespan will be reduced compared to the exact same card running at 70*C.

Electro migration is much, much nastier, i'm not sure how significant it is for current GPU designs because it is very dependant on the technology and techniques used to make the chips. The old XP range were (apparently) pretty much uneffected by migration where the new SOI ones are supposed to be very suceptable to it. Suffice it to say that upping the Vcore should only be done if you are able to write the card off there and then (like all serious overclocking imo)

 

[apoppin]

Originally posted by: solofly
Any overclocking to any component shortens life-span period. Another way to look at it, heat kills.

and

Google "Electromigration". Its not just heat.

O/C'ing definitely shortens the HW's life [period]

take intel CPUs for an example . . . their rated life is [i believe] 6 years. IF your O/C shortens it by 2 years, so what?

it's a matter of "balance" . . . i tend NOT to O/C a part until it's near the end of its useful life - to squeeze more performance out of it. When i finally push my 2.80c to 3.6Ghz with 10-15% more voltage, i won't worry about "how much life is left" . . . when or IF it dies, it gets upgraded.

that said, IF you O/C , make sure you keep your temps as low or lower than stock .

Just be aware, you are shortening a parts life by o/c'ing and often you lose the warranty.

 

[DrCrap]

Originally posted by: keysplayr2003

Originally posted by: amenx
But you see, 73c temp load is within the normal range of cards without OC, I can see your point if it was in the 90's. So if temps are the only factor that can shorten its life-span, than I'm better off than many other card owners who dont OC but have higher temps than mine. Would you agree with that?

I haven't owned a 6600GT, but isn't 73C hot as hell for that card? Or is that normal?
I mean, it's only 8 pipes. My GTX has triple the pipes and runs 65 at load, 43 idle.

I know the 6600GT runs at 500MHz core, but thats only 70MHz faster with 1/3? the transistors. That 73C just seems pretty hot to me.

I have a 6600GT and though 73c can not be considered 'cool' temps it's deffinitly no where near "hot as hell" under manufacturer's specs, the card's normal operating temp under full load can be up to 90c-95c. So, though I wouldn't push it much further when running at 73c, I think you have nothing to worry about.
Another thing, as Rollo said, most people answering you here have no idea what they're talking about (and that icludes mighty Rollo himself). The fact is, that those cards are designed to take higher voltage, and run at higher speeds than the stock, the only reason stock speeds & volts are lower, is to gain "safety" margin, since not all computers mobo's power supply etc. emits the same charge, also while some people work in very cool/cold environmet, others work in very hot non ventilated rooms. So you see, those cards are meant to have flexibility. Naturally, if your room is cold, your mobo is of high quality etc. your card will be far from its limits, thus you'll be able to push it further (WITHOUT DAMAGING IT) than someone who has the opposite conditions.

 

[apoppin]

Originally posted by: DrCrap

Originally posted by: keysplayr2003

Originally posted by: amenx
But you see, 73c temp load is within the normal range of cards without OC, I can see your point if it was in the 90's. So if temps are the only factor that can shorten its life-span, than I'm better off than many other card owners who dont OC but have higher temps than mine. Would you agree with that?

I haven't owned a 6600GT, but isn't 73C hot as hell for that card? Or is that normal?
I mean, it's only 8 pipes. My GTX has triple the pipes and runs 65 at load, 43 idle.

I know the 6600GT runs at 500MHz core, but thats only 70MHz faster with 1/3? the transistors. That 73C just seems pretty hot to me.

I have a 6600GT and though 73c can not be considered 'cool' temps it's deffinitly no where near "hot as hell" under manufacturer's specs, the card's normal operating temp under full load can be up to 90c-95c. So, though I wouldn't push it much further when running at 73c, I think you have nothing to worry about.
Another thing, as Rollo said, most people answering you here have no idea what they're talking about (and that icludes mighty Rollo himself). The fact is, that those cards are designed to take higher voltage, and run at higher speeds than the stock, the only reason stock speeds & volts are lower, is to gain "safety" margin, since not all computers mobo's power supply etc. emits the same charge, also while some people work in very cool/cold environmet, others work in very hot non ventilated rooms. So you see, those cards are meant to have flexibility. Naturally, if your room is cold, your mobo is of high quality etc. your card will be far from its limits, thus you'll be able to push it further (WITHOUT DAMAGING IT) than someone who has the opposite conditions.

no matter how you spin it . . . the answer to the OP's question is "yes" - overclocking DOES shorten a cards life-span [compared to NOT o/c'ing it].

the 'details' are in the "how much" and "does it matter to me"

 

[nRollo]

Originally posted by: DrCrap
Another thing, as Rollo said, most people answering you here have no idea what they're talking about (and that icludes mighty Rollo himself).

:laugh: Thanks for the laugh, and you're right, I'd be the first to admit I'm not an engineer.

The fact is, that those cards are designed to take higher voltage, and run at higher speeds than the stock, the only reason stock speeds & volts are lower, is to gain "safety" margin, since not all computers mobo's power supply etc. emits the same charge, also while some people work in very cool/cold environmet, others work in very hot non ventilated rooms. So you see, those cards are meant to have flexibility. Naturally, if your room is cold, your mobo is of high quality etc. your card will be far from its limits, thus you'll be able to push it further (WITHOUT DAMAGING IT) than someone who has the opposite conditions.

I think you're leaving something out of this though: the fact that the quality of the silicon on the chip itself varies card to card, so one guy in a cool room with a 6600GT might not get 10MHz and his neighbor in a hot room might get 75MHz.

OCing is highly YMMV, and as much as it pains me to say it, I agree with Apoppin' that the only "good" time to do it on high end parts is at the end of their life, when the loss of $ is irrelevant.

I don't agree with all this "Yippie skips! I OCd my already bitchin fast X1800XT and got 236 more 3DMarks!" stuff.

 

[NoStateofMind]

So would it be possible to keep temps lower than stock and have it overclocked to last the normal life span?

 

[nRollo]

Originally posted by: PC Surgeon
So would it be possible to keep temps lower than stock and have it overclocked to last the normal life span?

I think ONLY a chip engineer can answer this. If the threshold thermal migration occurs at lowers with increased voltage in the chip, the answer would be "No". I also don't know how the variables of the silicon quality and frequency itself factor into the equation.

It's one possibility that lower than stock temperatures are the key, but we need more scholarly advice than "It worked for me and some other guys".

 

[CKXP]

http://forums.anandtech.com/messageview...atid=28&threadid=1773169&enterthread=y

please read pm's post

 

[apoppin]

Originally posted by: Rollo

Originally posted by: PC Surgeon
So would it be possible to keep temps lower than stock and have it overclocked to last the normal life span?

I think ONLY a chip engineer can answer this. If the threshold thermal migration occurs at lowers with increased voltage in the chip, the answer would be "No". I also don't know how the variables of the silicon quality and frequency itself factor into the equation.

It's one possibility that lower than stock temperatures are the key, but we need more scholarly advice than "It worked for me and some other guys".

from the link above - thenks CKXP . . . IMPORTANT! [and pm does work for Intel - this time he IS speaking for them -and knows his stuff]

Originally posted by: pm

As far as the original question:

There's a linear dependence on mean-time to fail (MTTF) on a CMOS part to temperature, there's a square dependence on voltage to MTTF.

In non-statistics speak, increasing the temperature a bit will make your chip a little more likely to die. Increasing the voltage a little bit will have a much bigger statistical liklihood to kill your chip.

In the "old days" - prior to 0.18um process technology - the dominant reliability failure mechanism was electromigration... although this depended a fair bit on the design and the design rules used by the manufacturer. In electromigration, higher temperatures and higher voltages reduce the average time to failure. Because electromigration is dependent on current density, increasing voltage is worse.

Once the industry switched to dual-damascene copper technology, electromigration failures were relegated back to design-related mistakes. Instead the dominant failure mechanisms became time-depenedent dielectric breakdown (TDDB), PMOS BTI (although this is pretty much accounted for in manufacturer burn-in), and NMOS hot-electron gate-impact ionization (NMOS hot-e). In all three of these small increases in voltage can result in large reductions in operational lifetime, and small increases in temperature, result in generally small reductions in operational lifetime - in fact, in the case of NMOS hot-e, it gets worse with lowered temperature.

All three of these are quantum mechanical effects. If anyone is really curious, I can go into more details to explain exactly what is happening. For more details, you can read through this (rather condensed, somewhat esoteric but fundamentally correct) notes page http://www.eie.polyu.edu.hk/~ensurya/lect_notes/Reli_Fail/Reli_Fail_notes.htm. Although the whole page is an interesting reference, the part relevant to this discussion starts with "Properties of Metal-Oxide Silicon (MOS) System". Or Google, "NMOS hot-electron", "PMOS Bias Temperature Instability" and/or "time-depenedent dielectric breakdown", I can also give more detailed IEEE journal publications too if anyone wants them.

Patrick Mahoney
Enterprise Processor Division
Intel Corp.

We used to discuss this in GH for years . . . 'search' using pm and you will really learn something.

 

[nRollo]

Originally posted by: CKXP
http://forums.anandtech.com/messageview...atid=28&threadid=1773169&enterthread=y

please read pm's post

That post is good I bookmarked it. Thx.