How much does heat kill the life of your components?

[Stangs55]

I've got a mild 3% oc on my new FX55 + Zalman 7000Cu, but I don't want to decrease it's life at all. Will the small increase in heat decrease it's lifespan? What about oc'ing my 6800 ultra? At what temps should I back off to still be "safe" for both of these components?

Thanks in advance!

 

[imported_TimJ]

Overclocking will decrease the life of your components but the life span on them is still very long. Overclocked parts will become well outdated before they burn up.

 

[MysticX23]

as long as you aren't an extreme overclocker you don't have to worry.

 

[tk11]

You never can tell at exactly what point a chip will fail. My advice is if your going to overclock go big or go home. If you want to be conservative just don't mess with voltages. I've overclocked every cpu I've had (about 5) past it's limits to find how fast it would run and backed it down just enough to be stable and I've never fried anything.

 

[Duvie]

As long as you stay within the operating specs as listed y AMD there should be no degradation and certainly not within the likely use period of the average consumer for a cpu nowadays....it is when you deviate with vcore that I think most of the damage is done...

For example....

Let say a 2.4c P4 was produced in the same stepping batch and around the time the 3.2c was the flagship cpu of INtel. The fact is all cpus are made with the intent to be the fastest cpu to produce. the fact that so many of those become 2.4's, 2.6's, 2.8's and so on are more above marketing then bad yields resulting in down clocking. Many of the 2.4's of the D1 stepping can do 3.2ghz at default clock. therefore who is to say they were not qualified to have been 3.2's if Intel didn't need to bin them to 2.4's for reasons of market demand. as long as a chip can run at default vcore of the chip and not go past the operating temp as listed by AMD or Intel then the logevity should not be effected.....

The fact you go tht eflagship cpu means there is likely not as much headroom for ocing anyways. But if you can run it 5% and stiill not need more vcore or run at over 70c load I don't see where you are going beyond AMDs operating specs....

 

[caz67]

Overclock your cpu, heat won't matter at all, as the CPU will be obsolete way before it dies.

 

[pm]

Heat isn't too hard on a component unless it starts to get very high (>70C). The real killer is elevated voltages.

I don't agree with Duvie when he wrote:

the fact that so many of those become 2.4's, 2.6's, 2.8's and so on are more above marketing then bad yields resulting in down clocking.

If marketing ever causes a part to be down-binned, it is fairly uncommon. Frequency binning is almost always a Gaussian distribution, and that's why pricing falls a similar curve on the high end of the curve. In the event that there are plenty of (using Duvie's numbers) of 2.8GHz parts, then the pricing will be adjusted on the 2.8GHz parts so that they are priced the same or very closely with the 2.4GHz and 2.6GHz parts and that would very quickly take care of any demand for the lower frequency parts.

A Gaussian distribution in frequency is a fact of the semiconductor industry. If it somehow didn't exist, I'm sure that companies would just sell only top bins. It would certainly make things a lot less confusing.

 

[meltdown75]

Further to these questions...

As far as OC'ing the GPU, what is the best way to check the temp? Are there any internal temp monitors on videocards, or, software which can tell you? thanks

 

[Duvie]

Tha fact is Intel wants to command that top price for the flagship chip but needs to meet the demands of Dell and other OEMS for 1000 dollar budget machines that run the slower chips. You guys at Intel always fight this theory yet many times I have bought the lowest speed chip being built at the time with the same stepping as the flagship and get the chip to do the flagship speed at default vcore...Suspicious huh?????? I think not

I don't maybe you are right...Intel's yields may suck and they get all this fluctuation in parts. They certainly don't set out to make 2.4's, 2.6's and so on from different wafers. They are all meant to be the highest speed at the time. But like I said maybe so because Intel has really been sucking of late. The prescott line is a disaster IMO and a terrible transformation to the 90nm process (vs AMD).

Your pricing theory makes no sense cause ppl would never buy the cheaper part if they were all so close in cost...Intel does this to maximize cost and find that sweet spot to get the consumers money at every level of their demand. I cannot explain it perfectly but your theory lacks basic economic commmonsense....It makes sense for a company to do what I am stating. Intels cost is very minimal per each chip and no where near what the cost is for the top of line chips. So don't be fooled into thinnking they loose on selling lower priced 2.4's. If they sold them at all their truly benched and proven speeds the market wouldleap in bigger spurts and they couldn't command the huge premium for the latest on the top of the pile speed chips. Average JOe Schmo doens't want to lay down 400-800 on a cpu now for usually well more power then is needed. Average ppl want cheap systems and cpus for 200 or less and Intel just wouldn't make money at this level (or the amount they are accustomed to). We would still be back in the 2ghz range if this model was used cause it doensn't warrant much gain in speed. Sell all the chips at the same speed and make the money INtel likes or is accustomed to then they would sell them at around 400+ dollars. I think this is just enough to slow demand on PCs for average ppl. INtel knows this and their speed grades are for the different segments of users they see.

I believe Intels yields (from talking to certain individuals) are much better then you seem to be giving them credit for...


History....

I have owned 7 different Intel cpus...

1.8a bought at time 2.2 was the best chip available....It did 2.2 ghz at stock easily. Did 2.4ghz with just about .05v increase.

1.6a bought at time the best chips was the 2.4a....It did 2.53ghz at default core. Did ultimately 2.73ghz with 1.71v

Another 1.8a at time of 2.4ghz chip. It did 2.4at default core and ulimately 2.61ghz with 1.6+ vcore

2.4b bought at the time 3.06b was the highest chip. This chip chip did 3.15ghz at default vcore and ultimately was ran at 3.24ghz with just a minor vcore boost.

My weakest chip was a 2.6c that would only do 3.2ghz with a .05v boost though 3.2c was the best available at the time.

My 2.4c at the time 3.2c was the best does 3.4c at default vcore and 3.5ghz with 1.58-1.6v

fairly uncommon huh??? I must be very lucky then....Through in Thugs, PastorJay and those guys who have seen their share of many high oc's.

 

[BentValve]

I recently had a perfectly fine running OCed system fail on my..it ran a solid year straight OCed about 30% total and temps were excellent.

I think the most vulnerable parts in an OCed system are the motherboard (chipsets getting hot) and the power supply..OCed Athlon XPs and P4s suck major amounts of juice.

 

[pm]

Fairly uncommon my arse!!!

What I'm saying can be found in any advanced semiconductor design book. At various times in the last decade, I have access to the binning information on various product lines at Intel. I know for a fact that what I'm saying is true. <shrug> Believe what you like.

As far as your claim that Intel needs to bin parts... think of any other industry that takes parts that come off the assembly line and bins them based on performance. Honda doesn't sell a 115hp Civic, a 100hp Civic, and a top of the line 135hp Civic. They could if they wanted too... but they don't. Everyone else in all of the different manufacturing industry comes up with different product lines into order to hit various price points... and Intel does this too. Celerons, Pentiums, Centrinos, Xeons, Itaniums, etc. The whole binning process just adds complexity and confusion to the process. Especially since the bins change on a quarterly or bi-yearly. If parts didn't come out as a Gaussian distribution, they wouldn't artificially create them.

 

[Duvie]

As much as the INtel marketing machine could sell anything, they couldn't sell to most consumers the number oc computer an frequent pc upgrades if everybody always bought the top of the line chip.

If binning doesn't exist then read my edit (qadded part above)...how could that possibly be in a enthusiast market like OCing where YMMV??? It happens now and probably more since with better processes, more expensive machinery (tools), etc are producing most better yields on the chips. I have had ppl tell me this. Yields are better then in the past. now that is with the northwoods and in size change like to .13 to .09 may have had its initial issues, but with later stepping or the last stepping Intel always seems to make rockets from the bottom of the production line to the flagship chip.

I still find this more then a coincidence...I can see where INtel would want to spin this so ppl wouldn't try to OC, perhaps....

 

[Duvie]

Originally posted by: meltdown75
Further to these questions...

As far as OC'ing the GPU, what is the best way to check the temp? Are there any internal temp monitors on videocards, or, software which can tell you? thanks

My 5800U has a temperature program. I think a lot of the newer cards do or you can find an 3rd party app for Nvidia or ATI had their enthusiast web pages....

 

[Duvie]

Everyone else in all of the different manufacturing industry comes up with different product lines into order to hit various price points... and Intel does this too. Celerons, Pentiums, Centrinos, Xeons, Itaniums, etc.

I don't think so when you at any one time INtel has had a n alomost 1ghz range from lowest produced chip to hihgest speed chip in one of those same categories. Often time that has been a difference of 33-25% in certain aplication. That opens up a huge field for ppl within that sumarket by itself. That same range can often be 140-800 dollars in price difference. Again that is too wide for one subset of consumers...

The whole binning process just adds complexity and confusion to the process.

How...If ppl don't OC and iNtel locks the multiplier then How would Joe schmo know??? I think Intel just fears ppl will see they are getting a lot more for what they paid for and start OCing in mass droves and that would sink the top of the line flagship chips. Intel is already trying to make moves on killing oc options in 3rd party chipset and mobo makers. They klnow. Give yourselves credit over there...we know you are smart!!!

It is so simple....Just a lock here on the multipler and whola!!! where is the complexity??? they have orders for 1000 cpus at the 2.4, 2.6, 2.8 , 3.0, and 3.2ghz range. They just fill them. Make enough chips for each order. I don't see the complexity. If they say no to the manufacturer we will only sell them at their rated speed and they were all so good then they would possibly have difficulty selling the 2.4 to 2.6 range which may account for the best selling chip in units. If they then dont' drop the price chances are demand may dry up, again since most do not want to pay 400-800 for a cpu. Intel then can't sell its flagship at the cost it wants and don't make as much money. Simple Economics here...