How long will my CPU last if overclocked??

[octoberhungry]

Hi.

So let's say I overclock my i5 3570K to 4.4Ghz and my case is properly cooled, I also a 1000W PSU...Should my CPU last? What's the shelf life like with an overclocked CPU in this case?

Thanks!

 

[Pinhedd]

if you have to ask, not long enough

 

[wolfram23]

Depends on the voltage. I've been running my CPU (i5 750) OCed for... 2 years now. For more than a year it's been at 4ghz and water cooled. In that time I've had a bit of degredation which means I started to fail stress tests. Had to slightly bump the voltage, even though it was initially totally stable.

I know some SB chips can hit pretty high speeds without increasing the voltage at all, so it would last years and years with an OC.

I suppose the question is, how long do you want it to last? If you answer at least 1-2 years then no worries, you can push it pretty hard. If you answer 3-5 years or more, then you have to ask whether you need that much speed? For gaming, you probably won't see any benefit with an overclock so just maxing out at stock volts or maybe with a small increase will be sufficient.

 

[octoberhungry]

Depends on the voltage. I've been running my CPU (i5 750) OCed for... 2 years now. For more than a year it's been at 4ghz and water cooled. In that time I've had a bit of degredation which means I started to fail stress tests. Had to slightly bump the voltage, even though it was initially totally stable.

I know some SB chips can hit pretty high speeds without increasing the voltage at all, so it would last years and years with an OC.

I suppose the question is, how long do you want it to last? If you answer at least 1-2 years then no worries, you can push it pretty hard. If you answer 3-5 years or more, then you have to ask whether you need that much speed? For gaming, you probably won't see any benefit with an overclock so just maxing out at stock volts or maybe with a small increase will be sufficient.

Yeah I mean I'd like my CPU to last 3-5 years and I'll be primarily be using it for gaming. Thanks.

 

[Pinhedd]

Depends on the voltage. I've been running my CPU (i5 750) OCed for... 2 years now. For more than a year it's been at 4ghz and water cooled. In that time I've had a bit of degredation which means I started to fail stress tests. Had to slightly bump the voltage, even though it was initially totally stable.

I know some SB chips can hit pretty high speeds without increasing the voltage at all, so it would last years and years with an OC.

I suppose the question is, how long do you want it to last? If you answer at least 1-2 years then no worries, you can push it pretty hard. If you answer 3-5 years or more, then you have to ask whether you need that much speed? For gaming, you probably won't see any benefit with an overclock so just maxing out at stock volts or maybe with a small increase will be sufficient.

Bumping up the frequency increases dynamic power consumption which furthers electromigration. It is not necessary to increase the voltage in order to shorten the lifespan of any ASIC

 

[wolfram23]

Bumping up the frequency increases dynamic power consumption which furthers electromigration. It is not necessary to increase the voltage in order to shorten the lifespan of any ASIC

True, but as far as I'm aware the voltage is far more hurting to the lifespan.

 

[DJDeCiBeL]

This post is exactly why Intel offers the Performance Tuning Protection Plan http://click.intel.com/tuningplan/

$20 for the 3570K and it's absolutely worth it for an overclocker, IMO. OC all you want, and if you fry the CPU, they'll replace it (once), no questions asked. Something that the basic warranty DOESN'T cover. I got it for my 2500K, and whether I ever need it or not, it's nice to have the peace of mind that I'm safe to do whatever I want.

 

[octoberhungry]

This post is exactly why Intel offers the Performance Tuning Protection Plan http://click.intel.com/tuningplan/

$20 for the 3570K and it's absolutely worth it for an overclocker, IMO. OC all you want, and if you fry the CPU, they'll replace it (once), no questions asked. Something that the basic warranty DOESN'T cover. I got it for my 2500K, and whether I ever need it or not, it's nice to have the peace of mind that I'm safe to do whatever I want.

Oh good to know! Thanks. I'm actually getting my PC professionally overclocked but this will help if I start over-clocking myself.

 

[iam2thecrowe]

I have NEVER had a cpu fail, and I have overclocked every CPU I have ever owned. Get your PC "Professionally Overclocked"? does that come with some kind of warranty? otherwise its a waste of money.

 

[wonder44]

Generally most intel CPU's will last 6-10 years as they have no moving parts. I think of it this way, if you leave a piece of paper sitting on a desk for 12 weeks and do not move it, it will be in the same condition. However if you carry it around with you for 12 weeks it is most likely to rip, bend ect... And overclocking will not harm your CPU in anyway unless it does not have suffecient cooling. I overclocked my old pentium 4 in 2006 from 2.4GHz to 3.0GHz and it is still running just as good except faster than what it was. The point is the CPU will be fine just make sure you try not to overvolt it too much and make sure it stays nice n cool. Hope i helped.

 

[xtreme5]

get a decent cooler stay under volt you're cpu would be healthy for about 5 or 6 years with overclocking..

 

[Pinhedd]

Generally most intel CPU's will last 6-10 years as they have no moving parts. I think of it this way, if you leave a piece of paper sitting on a desk for 12 weeks and do not move it, it will be in the same condition. However if you carry it around with you for 12 weeks it is most likely to rip, bend ect... And overclocking will not harm your CPU in anyway unless it does not have suffecient cooling. I overclocked my old pentium 4 in 2006 from 2.4GHz to 3.0GHz and it is still running just as good except faster than what it was. The point is the CPU will be fine just make sure you try not to overvolt it too much and make sure it stays nice n cool. Hope i helped.

Nonsense. All electrical circuitry have moving parts, they're called charge carriers. Putting more energy behind each charge carrier by increasing the source-drain voltage differential and increasing dynamic power consumption by bumping up the clock frequency will both increase electromigration and subsequently reduce the lifespan of the integrated circuit. Component binning by circuit manufacturers ensure that components have enough operational headroom that component decay from normal operation will keep the component stable and functioning at factory settings at least through the warranty period. Increasing the supply voltage or bumping up the frequency beyond what it was marketed at can reduce the the integrated circuit's lifespan to less than the warranty period. Comparing a CPU to a piece of paper is an absolutely horrible analogy

 

[octoberhungry]

I have NEVER had a cpu fail, and I have overclocked every CPU I have ever owned. Get your PC "Professionally Overclocked"? does that come with some kind of warranty? otherwise its a waste of money.

Yeah it comes with a 1 year warranty. I guess I'm just trying to get an idea of generally how safe it is to overclock as long as my PC is properly cooled. If cooled and setup properly should a CPU that's overclocked last as long as one that is not overclocked? Will my CPU last 5 years say?

 

[octoberhungry]

Nonsense. All electrical circuitry have moving parts, they're called charge carriers. Putting more energy behind each charge carrier by increasing the source-drain voltage differential and increasing dynamic power consumption by bumping up the clock frequency will both increase electromigration and subsequently reduce the lifespan of the integrated circuit. Component binning by circuit manufacturers ensure that components have enough operational headroom that component decay from normal operation will keep the component stable and functioning at factory settings at least through the warranty period. Increasing the supply voltage or bumping up the frequency beyond what it was marketed at can reduce the the integrated circuit's lifespan to less than the warranty period. Comparing a CPU to a piece of paper is an absolutely horrible analogy

You sound like you know what you're talking about....Do you think chip manufacturers deliberately 'under'-clock their CPU's because they are guaranteeing themselves to be able sell a longer lifespan of the CPU to consumers? They know their CPU could run faster but it's safer not too? I mean consumers wouldn't overclock otherwise no? Thanks.

 

[Guest]

This post is exactly why Intel offers the Performance Tuning Protection Plan http://click.intel.com/tuningplan/

$20 for the 3570K and it's absolutely worth it for an overclocker, IMO. OC all you want, and if you fry the CPU, they'll replace it (once), no questions asked. Something that the basic warranty DOESN'T cover. I got it for my 2500K, and whether I ever need it or not, it's nice to have the peace of mind that I'm safe to do whatever I want.

Agreed. I'm getting it for my 2500K as soon as it becomes available in my country.

 

[Pinhedd]

You sound like you know what you're talking about....Do you think chip manufacturers deliberately 'under'-clock their CPU's because they are guaranteeing themselves to be able sell a longer lifespan of the CPU to consumers? They know their CPU could run faster but it's safer not too? I mean consumers wouldn't overclock otherwise no? Thanks.

They market chips according to a lot of factors but there are two key ones:

First, they set them according to what the fabrication process is capable of producing. It is well known that many top end chips can run between 33% and 50% faster than the speeds that they are marketed at. However, these speeds will not necessarily be achievable within target thermal output. The processors are designed to operate below a certain energy output (related to, but not the same as temperature) and exceeding this thermal output or allowing energy to build up without proper dissipation can cause the processor to degrade. Degradation causes manufacturing defects to become more apparent as well as new ones to appear. When Intel designs the chips they ensure that defects will not appear when operated within the thermal design power (TDP which you hear so much about). Having chips die a premature death when the customer is running it within operational speeds is bad business, so it's a very carefully calculated operating point.

So what does this mean exactly? When chips are fabricated they are not all fabricated equally. Chips are manufactured on what are known as 'wafers' (which is suitable because they are literally wafers of metalloids such as silicon at the start) and at the end of the process a certain percentage of them are simply not expected to work the same and many will simply not work at all. Each chip is tested to establish a quality level and identify any defects, they are then sorted. The chips that have too many defects are discarded. The chips that have some defects but can be made partially operational are separated from the chips that have no operational defects. The chips that have no defects and test the best will become the top of the line processors with all all operational components, the chips that have no defects and test slightly worse will become high end processors and so on and so forth. The chips that have defects but can be made operational but disabling part of the chip will be relegated to mid and low end processors. If market demand necessitates it, they will even take perfectly good chips and disable part of them, in fact this has become part of Intel's standard strategy to keep fabrication costs low.

Fun Fact: The famous 4/6 core Sandybridge-E processors are all actually fabricated 8 core processors with 20 Megabytes of L3 cache. Between 2 and 4 cores are disabled and up to 50% of the cache. The exact same chip is used for the Xeon E5 processors which have all 8 cores and up to 20 Megabytes of cache enabled.

Sometimes a manufacturing process yields are better than expected which results in a lower number of defects and/or overall higher performance. This is pretty typical of a maturing fabrication process which is why Intel's 'tick-tock' roadmap tends to have new architectures being more suitable for overclocking than die shrinks. When this happens, the binning process remains the same but there could be an excess of processors at one bin level and/or a shortage at another... which leads on to reason number two

Second, Intel prices and markets their processors in order to provide a very clear idea of performance per dollar. If every chip capable of being an i7 was packaged as an i7 then there would be a lot more i7s in the event of good fabrication yields. The converse is also true, bad yields will result in a shortage such as the one seen with TSMC's 28nm process which is currently hindering NVidia's 600 series. An excess of i7s would necessarily mean a shortage of i5s and i3s which upsets the careful price balance that Intel has established so as to differentiate their various products from each other. The overwhelming majority of sales come from the mid to lower end products, not the top end. So, they design their top end processor around around a particular performance point, standard warranty period, expected market demand at each price point, required yields and expected processor life. As long as the top offerings fit within this profile they can hack and slash at it to create inferior products and market them appropriately. Intel has not missed this mark in a very long time.

 

[jessterman21]

You sound like you know what you're talking about....Do you think chip manufacturers deliberately 'under'-clock their CPU's because they are guaranteeing themselves to be able sell a longer lifespan of the CPU to consumers? They know their CPU could run faster but it's safer not too? I mean consumers wouldn't overclock otherwise no? Thanks.

Not really, but yes they set a very safe clock for each CPU - that's why any CPU can be overclocked safely to some extent.

Keeping your CPU cool and at stock voltage while overclocking is the best way to guarantee the longest life.

 

[jessterman21]

They market chips according to a lot of factors but there are two key ones:

First, they set them according to what the fabrication process is capable of producing. It is well known that many top end chips can run between 33% and 50% faster than the speeds that they are marketed at. However, these speeds will not necessarily be achievable within target thermal output. The processors are designed to operate below a certain energy output (related to, but not the same as temperature) and exceeding this thermal output or allowing energy to build up without proper dissipation can cause the processor to degrade. Degradation causes manufacturing defects to become more apparent as well as new ones to appear. When Intel designs the chips they ensure that defects will not appear when operated within the thermal design power (TDP which you hear so much about). Having chips die a premature death when the customer is running it within operational speeds is bad business, so it's a very carefully calculated operating point.

So what does this mean exactly? When chips are fabricated they are not all fabricated equally. Chips are manufactured on what are known as 'wafers' (which is suitable because they are literally wafers of metalloids such as silicon at the start) and at the end of the process a certain percentage of them are simply not expected to work the same and many will simply not work at all. Each chip is tested to establish a quality level and identify any defects, they are then sorted. The chips that have too many defects are discarded. The chips that have some defects but can be made partially operational are separated from the chips that have no operational defects. The chips that have no defects and test the best will become the top of the line processors with all all operational components, the chips that have no defects and test slightly worse will become high end processors and so on and so forth. The chips that have defects but can be made operational but disabling part of the chip will be relegated to mid and low end processors. If market demand necessitates it, they will even take perfectly good chips and disable part of them, in fact this has become part of Intel's standard strategy to keep fabrication costs low.

Fun Fact: The famous 4/6 core Sandybridge-E processors are all actually fabricated 8 core processors with 20 Megabytes of L3 cache. Between 2 and 4 cores are disabled and up to 50% of the cache. The exact same chip is used for the Xeon E5 processors which have all 8 cores and up to 20 Megabytes of cache enabled.

Sometimes a manufacturing process yields are better than expected which results in a lower number of defects and/or overall higher performance. This is pretty typical of a maturing fabrication process which is why Intel's 'tick-tock' roadmap tends to have new architectures being more suitable for overclocking than die shrinks. When this happens, the binning process remains the same but there could be an excess of processors at one bin level and/or a shortage at another... which leads on to reason number two

Second, Intel prices and markets their processors in order to provide a very clear idea of performance per dollar. If every chip capable of being an i7 was packaged as an i7 then there would be a lot more i7s in the event of good fabrication yields. The converse is also true, bad yields will result in a shortage such as the one seen with TSMC's 28nm process which is currently hindering NVidia's 600 series. An excess of i7s would necessarily mean a shortage of i5s and i3s which upsets the careful price balance that Intel has established so as to differentiate their various products from each other. The overwhelming majority of sales come from the mid to lower end products, not the top end. So, they design their top end processor around around a particular performance point, standard warranty period, expected market demand at each price point, required yields and expected processor life. As long as the top offerings fit within this profile they can hack and slash at it to create inferior products and market them appropriately. Intel has not missed this mark in a very long time.

+1

That is one ridiculous rig, btw.

 

[mfleming001]

Fun Fact: The famous 4/6 core Sandybridge-E processors are all actually fabricated 8 core processors with 20 Megabytes of L3 cache. Between 2 and 4 cores are disabled and up to 50% of the cache. The exact same chip is used for the Xeon E5 processors which have all 8 cores and up to 20 Megabytes of cache enabled.

I just wanted to point out that the 4 core LGA2011s are actually a different die. My 3820 got its feelings hurt.

http://vr-zone.com/articles/intel-s-core-i7-3820-and-its-compact-die--truly-cheaper-socket-2011-/14893.html

http://news.softpedia.com/news/Intel-Core-i7-3820-Is-Based-on-an-Entirely-New-Sandy-Bridge-E-Die-243616.shtml

 

[octoberhungry]

They market chips according to a lot of factors but there are two key ones:

First, they set them according to what the fabrication process is capable of producing. It is well known that many top end chips can run between 33% and 50% faster than the speeds that they are marketed at. However, these speeds will not necessarily be achievable within target thermal output. The processors are designed to operate below a certain energy output (related to, but not the same as temperature) and exceeding this thermal output or allowing energy to build up without proper dissipation can cause the processor to degrade. Degradation causes manufacturing defects to become more apparent as well as new ones to appear. When Intel designs the chips they ensure that defects will not appear when operated within the thermal design power (TDP which you hear so much about). Having chips die a premature death when the customer is running it within operational speeds is bad business, so it's a very carefully calculated operating point.

So what does this mean exactly? When chips are fabricated they are not all fabricated equally. Chips are manufactured on what are known as 'wafers' (which is suitable because they are literally wafers of metalloids such as silicon at the start) and at the end of the process a certain percentage of them are simply not expected to work the same and many will simply not work at all. Each chip is tested to establish a quality level and identify any defects, they are then sorted. The chips that have too many defects are discarded. The chips that have some defects but can be made partially operational are separated from the chips that have no operational defects. The chips that have no defects and test the best will become the top of the line processors with all all operational components, the chips that have no defects and test slightly worse will become high end processors and so on and so forth. The chips that have defects but can be made operational but disabling part of the chip will be relegated to mid and low end processors. If market demand necessitates it, they will even take perfectly good chips and disable part of them, in fact this has become part of Intel's standard strategy to keep fabrication costs low.

Fun Fact: The famous 4/6 core Sandybridge-E processors are all actually fabricated 8 core processors with 20 Megabytes of L3 cache. Between 2 and 4 cores are disabled and up to 50% of the cache. The exact same chip is used for the Xeon E5 processors which have all 8 cores and up to 20 Megabytes of cache enabled.

Sometimes a manufacturing process yields are better than expected which results in a lower number of defects and/or overall higher performance. This is pretty typical of a maturing fabrication process which is why Intel's 'tick-tock' roadmap tends to have new architectures being more suitable for overclocking than die shrinks. When this happens, the binning process remains the same but there could be an excess of processors at one bin level and/or a shortage at another... which leads on to reason number two

Second, Intel prices and markets their processors in order to provide a very clear idea of performance per dollar. If every chip capable of being an i7 was packaged as an i7 then there would be a lot more i7s in the event of good fabrication yields. The converse is also true, bad yields will result in a shortage such as the one seen with TSMC's 28nm process which is currently hindering NVidia's 600 series. An excess of i7s would necessarily mean a shortage of i5s and i3s which upsets the careful price balance that Intel has established so as to differentiate their various products from each other. The overwhelming majority of sales come from the mid to lower end products, not the top end. So, they design their top end processor around around a particular performance point, standard warranty period, expected market demand at each price point, required yields and expected processor life. As long as the top offerings fit within this profile they can hack and slash at it to create inferior products and market them appropriately. Intel has not missed this mark in a very long time.

Wow! Thanks for the detailed answer! I just learn a lot! Cheers!

 

[octoberhungry]

Not really, but yes they set a very safe clock for each CPU - that's why any CPU can be overclocked safely to some extent.

Keeping your CPU cool and at stock voltage while overclocking is the best way to guarantee the longest life.

Thanks..yeah it's going to be liquid cooled so I'm hoping i wont have any problems. Dumb question perhaps - when you say 'stock voltage' do you mean not increasing the voltage into the CPU over what it is designed for? Thanks.

 

[jessterman21]

Thanks..yeah it's going to be liquid cooled so I'm hoping i wont have any problems. Dumb question perhaps - when you say 'stock voltage' do you mean not increasing the voltage into the CPU over what it is designed for? Thanks.

Right - not increasing the voltage at all. There's usually some good headroom without bumping the voltage.

 

[octoberhungry]

Right - not increasing the voltage at all. There's usually some good headroom without bumping the voltage.

Gotcha. Thanks!

 

[saint19]

All those answers for something easy?

The CPU will die when HAVE to die, if you kill it is your own responsibility.

 

[J_E_D_70]

Great conversation! But I'd ask what are you using it for, exactly? Are you computing pi 24/7 for fun or playing games? That makes all the difference in the world.

As long as you leave speedstep on, your 4.4 is actually running at 1.6 at idle and for things like web surfing. If you play something like Battlefield 3 multiplayer, you'll hit roughly 3.7 across the cores here and there so you still won't be at 4.4.

It's not like the old days when a 4.4 OC was running at that speed the entire time the PC was on.