Voltage vs Heat; what nukes the cpu for good?

[gunnarhx]

Hello,
I have a very simple question; is heat the only thing that can permanently kill a cpu? Or if I crank up the vcore (whilst cpu temp is moderate), could it alone nuke the cpu by pulling too much current?

Background; I've got a phenom II x3 720 with a very stable OC at 3.6GHz, memory at 1066, NB at 2600.
Now I got my hands on a second GeForce GTX 260 and it seems that CPU bottoms out at some games.
So I am looking into ways to get the clock up to 3.7-3.8 (I know that it might not be possible based on other people's level of success with 720).
The case is a Cooler Master HAF, and I got a Thermalright black 130 extreme heatsink with two 2k fans (yes, noisy...) so I never had any problems with cooling. The CPU is < 40C on max load.

Now, the 3.6 OC is at ~1.536V. I haven't had any success with 3.7 even at 1.58V, but I could try some more if necessary, but then I get back to the question at top; if I maintain temp <50C, could I damage the CPU with voltage?

Thanks

 

[sid_nag17]

too much voltage will shorten the cpu's lifespan but as long as you have good cooling it should be fine.

why dont you try the aod profiles. it works like the intel turbo feature. there is an article on tms hardware about it chech it out.

u can crank up the cpu to the max only when you are playing games rest of the time keep the cpu at its default clocks.


or there is a very good software called 'Phenom msr Tweaker' download it.

it works as a custom cool n quiet .

 

[croc]

Hello,
I have a very simple question; is heat the only thing that can permanently kill a cpu? Or if I crank up the vcore (whilst cpu temp is moderate), could it alone nuke the cpu by pulling too much current?

Background; I've got a phenom II x3 720 with a very stable OC at 3.6GHz, memory at 1066, NB at 2600.
Now I got my hands on a second GeForce GTX 260 and it seems that CPU bottoms out at some games.
So I am looking into ways to get the clock up to 3.7-3.8 (I know that it might not be possible based on other people's level of success with 720).
The case is a Cooler Master HAF, and I got a Thermalright black 130 extreme heatsink with two 2k fans (yes, noisy...) so I never had any problems with cooling. The CPU is < 40C on max load.

Now, the 3.6 OC is at ~1.536V. I haven't had any success with 3.7 even at 1.58V, but I could try some more if necessary, but then I get back to the question at top; if I maintain temp <50C, could I damage the CPU with voltage?

Thanks

Voltage = heat. You can remove the heat faster with better cooling, but you are still generating heat in the chip by over-volting it. Also, there is the minor problem with electro-migration, where gate material migrates from the +VDC side of the gate towards the GND side of the gate. This process will eventually kill that gate, and your CPU.

 

[4Ryan6]

could I damage the CPU with voltage?

Absolutely

 

[gunnarhx]

Thanks for the responses.
Practically, it seems that you're saying that short term heat could be the killer, but even though temperature is under control, one will shorten the lifespan of the chip in the long run.
In my system, temperature looks to be under control, since I've got a lot of air moving in my case and dual heat sink fans.
I probably follow the suggestion to only bump up to max during games etc

Thanks again everybody

 

[yang]

both heat and voltage can damage your cpu. Because it's easier to reduce the heat rather than voltage in an overclock. Most likely your cpu will die more quickly by the high voltage you're pumping in

 

[overshocked]

I would say that voltage will kill it 10x faster than temps.

But as long as you keep the vcore within the cpu's specs, you should be pretty good.

I would say temp is the long term killer.

 

[4Ryan6]

I would say that voltage will kill it 10x faster than temps.

But as long as you keep the vcore in intels specs, you should be pretty good.

I would say temp is the long term killer.

Good explanation !

 

[richardscott]

yes but for every 10c cooler your chip is it doubles its life span so i rape my cpu's with high voltage but reduce the temps by like 80-90c to make up for the life span losses my e8600 has ran for a year as my main oc with 1.47vcore and have benched it upto 1.68v

 

[ubernoobie]

I would say that voltage will kill it 10x faster than temps.

But as long as you keep the vcore in intels specs, you should be pretty good.

I would say temp is the long term killer.

You mean amd's specs lol

 

[4Ryan6]

CPUs specs would be better

 

[gunnarhx]

Thanks for the continued info.

It seems that the prevailing opinion is that both voltage and temp will shorten the life expectancy of the CPU. So preferably keep the vcore within spec ( or close at least ) and keep a good look at the temp ( I guess, as low as possible).

In my case, temp has never been a problem, so I will try to keep the voltage as low as I can for the frequency I want.

BTW, since we are on the topic; for a non-professional OC'er, who just try to bump up the speed to a 'respectable' level for purpose of video encoding, games etc. And assume that this person is challenging the specs for 1) voltage and 2) temp. What would the life expectancy be for both intel and amd chips?
Does anyone have experience in terms of years/months before the chip finally gives in?? Would be interesting to know...

Thanks

 

[overshocked]

You mean amd's specs lol

fixed lol

 

[richardscott]

if you keep both the vcore and temps below the cpu's specs then you can guarantee it will last 2 years or your money back

 

[CompuTronix]

gunnarhx,

I realize this is an AMD thread, however, the following very excellent AnandTech article explains processor degradation, which addresses the effects of excessive voltage and temperature. Although it concerns Intel, the principals also apply to AMD: - http://anandtech.com/cpuchipsets/intel/showdoc.aspx?i=3251&p=6

4ryan6,

Please excuse the slightly off-topic detour for a moment . While I agree with overshocked's statement, I would also like to take issue with it.

I would say that voltage will kill it 10x faster than temps.

But as long as you keep the vcore within the cpu's specs, you should be pretty good.

I would say temp is the long term killer.

overshocked,

I'm glad that you've commented on this point, because I've had a burning question to ask since you rolled out your Core i7 Overclocking Guide: Don't you think it's irresponsible to recommend to our readers temperatures and core voltages which exceed specifications?

From your Guide:

When at a full load the core i7 should not exceed 75c, 80c MAX.

From Intel's Processor Spec Finder - http://processorfinder.intel.com/List.aspx?ParentRadio=All&ProcFam=3052&SearchKey=

All Core i7 9xx variants:
68c Tcase Max (CPU temperature)

Since Intel's Thermal Specification is Tcase Max, add 5c to find the corresponding Tjunction (Core temperature) value, which is 73c, not "75c, 80c MAX". See the following Intel engineering document - http://arxiv.org/ftp/arxiv/papers/0709/0709.1861.pdf

From your Guide:

... max safe voltage for this chip is 1.375v

From Intel's Processor Spec Finder - http://processorfinder.intel.com/List.aspx?ParentRadio=All&ProcFam=3052&SearchKey=

All Core i7 9xx variants:
VID Voltage Range .80V-1.375V

... MAX voltage is 1.55 that means that up until that number your risk of damaging the chip is not that high...

I have never read any such specification in hundreds of pages of Intel Core i7 documents and papers. Perhaps I missed it? Would you care to share with us where Intel provides this information?

As I've responded to threads and provided help to our Forum Members throughout the past 3+ years, I'm always mindful that most consider the money they have wrapped up in their rigs as hard earned. Consequently, I've been very careful to not make any suggestions or recommendations to our readers in threads, or in my Core i7 and Core 2 Temperature Guide, which exceed manufacturer's specifications, or could result in someone toasting their transistors.

Aren't you contradicting yourself here in this thread? Would you consider correcting your Guide?

Respectfully,

Comp

 

[overshocked]

gunnarhx,

I realize this is an AMD thread, however, the following very excellent AnandTech article explains processor degradation, which addresses the effects of excessive voltage and temperature. Although it concerns Intel, the principals also apply to AMD: - http://anandtech.com/cpuchipsets/intel/showdoc.aspx?i=3251&p=6

4ryan6,

Please excuse the slightly off-topic detour for a moment . While I agree with overshocked's statement, I would also like to take issue with it.

overshocked,

I'm glad that you've commented on this point, because I've had a burning question to ask since you rolled out your Core i7 Overclocking Guide: Don't you think it's irresponsible to recommend to our readers temperatures and core voltages which exceed specifications?

From your Guide:

From Intel's Processor Spec Finder - http://processorfinder.intel.com/List.aspx?ParentRadio=All&ProcFam=3052&SearchKey=

All Core i7 9xx variants:
68c Tcase Max (CPU temperature)

Since Intel's Thermal Specification is Tcase Max, add 5c to find the corresponding Tjunction (Core temperature) value, which is 73c, not "75c, 80c MAX". See the following Intel engineering document - http://arxiv.org/ftp/arxiv/papers/0709/0709.1861.pdf

From your Guide:

From Intel's Processor Spec Finder - http://processorfinder.intel.com/List.aspx?ParentRadio=All&ProcFam=3052&SearchKey=

All Core i7 9xx variants:
VID Voltage Range .80V-1.375V

I have never read any such specification in hundreds of pages of Intel Core i7 documents and papers. Perhaps I missed it? Would you care to share with us where Intel provides this information?

As I've responded to threads and provided help to our Forum Members throughout the past 3+ years, I'm always mindful that most consider the money they have wrapped up in their rigs as hard earned. Consequently, I've been very careful to not make any suggestions or recommendations to our readers in threads, or in my Core i7 and Core 2 Temperature Guide, which exceed manufacturer's specifications, or could result in someone toasting their transistors.

Aren't you contradicting yourself here in this thread? Would you consider correcting your Guide?

Respectfully,

Comp

No offense comp, but i think there is to much hype over "specs".....

Lets give this a little thought, overclocking is out of spec to begin with. I encourage people to rock on, and crank up the voltage a little out of spec [:lectrocrew:6] . I dont think its irresponsible at all.

I have run cpu's into the dirt, pushing 1.9 vcore and going down to -120c. Witch are both WAAAYYYY out of spec. and they keep on rocking.

24/7 i like to keep it at -45c and with 1.6vcore. I have never had a hiccup.

These users (the ones that will overclock) traditionally are looking to update there hardware within 3 years. Wich is about 1/10 of how long intel says a chip will run if its in spec.


Id say to hell with intels spec sheet. [:jaydeejohn:4] Overclocking is awesome and have some fun with it.
[:thegreatgrapeape:3]

Cheers

 

[peacemaker-]

I don't argue that having one's guides be accurate is important. My comment is in regards to 73*C vs 80*C core temps. I agree that the Intel spec of 73*C should be stated. Then I have no problem with offering a personal opinion that 80*C for full load is still within a personal comfort range. And then I personally would like the reasoning for that comfort range of 80*C.

It just so happens that my 4.0 overclock took me to 80-81*C during testing. Certainly I was uncomfortable and decided that 4 hours of Prime95 at these temps was enough "stability" for me (along with 1 hour of OCCT, 50 passes of Intel Burn Test and 3 passes of memtest86+). I have read more than one site which has stated 80*C is ok, without specifics to back it up. I personally am comfortable knowing that I will never be running at 80*C during my normal computing and hopefully, never again. It is ironic that my testing alone subjects my chip to the worst beating it will take. But I justify this for the sake of trust in the systems "stability".

The fact that Intel claims that the core i7 920 will start to shut down at 100*C, and the fact that I haven't seen reports of fried chips from 80*C use, I made my own deductions. If there is evidence to the contrary, someone please chime in.

 

[overshocked]

I don't argue that having one's guides be accurate is important. My comment is in regards to 73*C vs 80*C core temps. I agree that the Intel spec of 73*C should be stated. Then I have no problem with offering a personal opinion that 80*C for full load is still within a personal comfort range. And then I personally would like the reasoning for that comfort range of 80*C.

It just so happens that my 4.0 overclock took me to 80-81*C during testing. Certainly I was uncomfortable and decided that 4 hours of Prime95 at these temps was enough "stability" for me (along with 1 hour of OCCT, 50 passes of Intel Burn Test and 3 passes of memtest86+). I have read more than one site which has stated 80*C is ok, without specifics to back it up. I personally am comfortable knowing that I will never be running at 80*C during my normal computing and hopefully, never again. It is ironic that my testing alone subjects my chip to the worst beating it will take. But I justify this for the sake of trust in the systems "stability".

The fact that Intel claims that the core i7 920 will start to shut down at 100*C, and the fact that I haven't seen reports of fried chips from 80*C use, I made my own deductions. If there is evidence to the contrary, someone please chime in.

Yeah man, for 100% load it is fine to go over intel specs.

Hence why i state that the absolute max is 80c.

On extreme air benching i have gone up to 107c with the i7, and it doesnt shut down. Im sure it was scaling back though. [:mousemonkey:4]

 

[one-shot]

75C scares me a little. With the warm weather and lack of AC, I had to back off my overclock.

 

[CompuTronix]

Id say to hell with intels spec sheet. [:jaydeejohn:4] Overclocking is awesome and have some fun with it.[:thegreatgrapeape:3]

overshocked,

I'm all for fun, but that's quite a flippant attitude when it involves other people's money. Suit yourself; I still think it's irresponsible. Regardless, you haven't answered my question concerning your claim that "The MAX voltage is 1.55". I find it difficult to believe that you haven't encountered any processor degradation while running "cpu's into the dirt"... or have you?

So, where did you come up with 1.55?

Comp

 

[overshocked]

overshocked,

I'm all for fun, but that's quite a flippant attitude when it involves other people's money. Suit yourself; I still think it's irresponsible. Regardless, you haven't answered my question concerning your claim that "The MAX voltage is 1.55". I find it difficult to believe that you haven't encountered any processor degradation while running "cpu's into the dirt"... or have you?

So, where did you come up with 1.55?

Comp

I know youd love for me to say that I pulled it out of my arse, but nope.

Strait from the DATA sheet itself.


Processor Absolute Minimum and Maximum Ratings



Symbol------------------Parrameter--------------------------------- max
VCC-----Processor Core voltage with respect to VSS------- 1.55 V

Happy? [:lectrocrew:2]

 

[CompuTronix]

1.55 it is indeed. Sir, I stand corrected. Shame on me for not recalling my notes more carefully.

However, in all fairness, and to offer perspective, the following is for everone's benefit: Intel® Core™ i7 Processor Extreme Edition and Intel® Core™ i7 Processor, Datasheet, Volume 1 - http://download.intel.com/design/processor/datashts/320834.pdf


"2.10 Absolute Maximum and Minimum Ratings

Table 2-6 specifies absolute maximum and minimum ratings, which lie outside the functional limits of the processor. Only within specified operation limits can functionality and long-term reliability be expected.

At conditions outside functional operation condition limits, but within absolute maximum and minimum ratings, neither functionality nor long-term reliability can be expected. If a device is returned to conditions within functional operation limits after having been subjected to conditions outside these limits, but within the absolute maximum and minimum ratings, the device may be functional, but with its lifetime degraded depending on exposure to conditions exceeding the functional operation condition limits.

At conditions exceeding absolute maximum and minimum ratings, neither functionality nor long-term reliability can be expected. Moreover, if a device is subjected to these conditions for any length of time then, when returned to conditions within the functional operating condition limits, it will either not function or its reliability will be severely degraded."


Have you encountered any processor degradation, or are Intel's engineers mistaken?

 

[overshocked]

Im sure i have but i have been running my cpu's for years like this and nothing noticable has happend yet....

 

[CompuTronix]

With respect to the title of this thread - Voltage vs Heat; what nukes the cpu for good?, and with respect to your previous statment:

I would say that voltage will kill it 10x faster than temps.

But as long as you keep the vcore within the cpu's specs, you should be pretty good.

I would say temp is the long term killer.

I would say that processor degradation is an undeniable issue in 45 nanometer technology, which shouldn't be dismissed, and will become more pronounced as die-shrinks evolve. The AnandTech article I posted above - http://anandtech.com/cpuchipsets/i [...] i=3251&p=6 - offers a clear explanation. If you follow the Real Temp thread at ExtremeSystems, a very astute Forum Member, rge, who has assisted Real Temp's developer, Kevin Glynn, with much of his research, has documented processor degradation due to overvolting. This is certainly not an isolated instance.

Additionally, an anonymous Intel engineer stated on the Real Temp thread that the VID Voltage Range value shown in the Processor Spec Finder should not be exceeded, because there is very little upward tolerance before degradation occurs.

My point is that since Intel clearly states that operating a processor at "Absolute Maximum" will result in degradation, and it's been documented by others, including myself, do you really think it's wise to suggest that it's OK to ramp the Vcore up from 1.375 to 1.55? That's a 12.73% increase. Don't you think that's pushing toward the extreme for the novice overclocker who might not understand the potential consequences?

 

[overshocked]

Of course every di shrink is more proned to degregation than the size before.

But be that as it may, I did (and continue) to suggest that people have fun with there chips. I usually warn people and say that degradation does occur.

At that point i bestowe the resposobility apon them to do more research and find out how much is to much. It is beyond the outlook of my guide to tell them.

Why is this beyond the outlook of my guide?
Becuase there is some risk involved in overclocking at all, they should do what they feel safe doing.
Hence why i dont call my guide (and nor should anyone call their guide) the bible of overclocking. It is simply an "i7 overclocking guide".

Personally I like to put myself under potential "risk" and slap a phase change on my system or an LN2 pot and crank up the voltage all the way.

The outlook of my guide however IS to show what I have expierienced as well as show HOW to overclock. [:lectrocrew:6]