Overclock.net banner
1 - 20 of 20 Posts

·
Registered
Joined
·
228 Posts
Discussion Starter · #1 · (Edited)
I am stress testing lots of programs just so I can stabilize my manual all P core frequency on air cooling.
Obviously you would not want to run it at 100C or above for regular usage. But there is a peaceful thing knowing it is fully stable no matter the work load and only thing holding you back would be too high of temps for consistent usage. Especially when real world usage would never hit those temps. As opposed to it is stable in real real work loads, but would crash under an extreme test even if temps were in check. This the temporary extreme high temps for ultimate proof of stability during some tests as I cannot temporarily switch to a super extreme custom loop and then back to prove stability if temps were kept in check.
So if I run a stress test and it hits 100C for a few seconds or even a few minutes, is it going to degrade permanently? Or would I have to run t hours at that temp or higher for hours or even days for permanent degradation??
And how hot would temp have to get for permanent degradation within a few seconds if 100C is fine for a few seconds??
 

·
LTSC for life crew
Joined
·
4,505 Posts
Generally you want to avoid high temperatures for sustained time periods. 100c on the CPU for 3 seconds isn't going to kill it though, 100c is stated in the Intel specs as absolute max allowable temp for the 12900K.


Also found this and it seemed to have some fairly interesting discussion going on.

https://www.reddit.com/r/askscience/comments/2lpus6
 

·
Registered
Joined
·
995 Posts
Not a ton of evidence on it, but Macbooks have been running processors to 100C for years. Obviously this can't be applied to the Alder Lake architecture, but most of the previous gen Intel processors are still alive and well. I personally have not experienced any degradation whatsoever from running Intel processors in this heat range for extended periods throughout my experience.

I wouldn't lose sleep on hitting 100C, but personally I'd be trying to get the temps at least within the 80's during everyday use, if for peace of mind and nothing else.
 

·
Registered
Joined
·
228 Posts
Discussion Starter · #4 ·
Not a ton of evidence on it, but Macbooks have been running processors to 100C for years. Obviously this can't be applied to the Alder Lake architecture, but most of the previous gen Intel processors are still alive and well. I personally have not experienced any degradation whatsoever from running Intel processors in this heat range for extended periods throughout my experience.

I wouldn't lose sleep on hitting 100C, but personally I'd be trying to get the temps at least within the 80's during everyday use, if for peace of mind and nothing else.

Thanks for the tip. And yes I am going to make sure I keep my temps in check for everyday usage. Just for stress testing to ensure stability temporarily during toughest parts to ensure it will pass.

Everyday usage temps are going to be much lower obviously as stress tests make temps much higher almost always.

Which CPUs from Intel have you ran around 100C range for extended periods and no degradation?? Was it Coffee Lake or Comet Lake or Rocket Lake??

And would you say Intel CPUs can handle higher temps than AMD without errors. Is that why Intel has 100C as throttle point where as AMD only has 90C as throttle point??
 

·
Registered
Joined
·
70 Posts
No need to worry at all, the throttling will kick in and should not damage the CPU. On the Ivy/haswell (3rd/4th gen) lots of people still used the stock intel HSF which on CPUs like the 3770K or 4790K would reach the throttling temp pretty often and they did just fine, albeit with a loss of performance of course.
 

·
Registered
Joined
·
7 Posts
Nothing stands against the trial of time, and everything degrades and eventually fails, including CPUs running at stock. It's just a matter of how fast; and generally speaking, higher voltages and higher temperatures accelerate degradation.

The reason why you basically never 'feel' the degredation at stock is that they have a pretty large safety margin built into the stock V/F curve to 'postpone' the felt effects of it. The margin usually on the scale of well over 10 years, and is long enough that it's way more likely for the CPU's VRMs on motherboard to age (degrade) and short like 12 volts into the CPU (or experience some other motherboard failure). This is a reason why dead GPUs are a more common event than dead CPUs.

Further, a big fraction of the performance or frequency harvest from an overclocking effort is about eating into that safety margin. In that an overclocker will usually experience the effects of degradation sooner and more frequently compared to someone running at stock, even if the rates of degradation are similar.

Degradation is not necessarily a catastrophic thing, as the response to it is often merely just bumping up the voltage by 5-20mV. Of course, doing so will accelerate it and the period/time between necessary bumps will narrow, but you should also consider its timescale and hw long you actually care about the CPU staying high-quality.
In a heavily-dramatised hypothetical, say in 4 years, you require an additional 15mV to stabilise at the given frequency, you might then need another 15mV bump in 3.5 years, and then a third in 3 years after that; at that point, you've had the CPU for over 10 years.

For a more in-depth response, check out this video
 

·
Registered
Joined
·
228 Posts
Discussion Starter · #8 · (Edited)
Nothing stands against the trial of time, and everything degrades and eventually fails, including CPUs running at stock. It's just a matter of how fast; and generally speaking, higher voltages and higher temperatures accelerate degradation.

The reason why you basically never 'feel' the degredation at stock is that they have a pretty large safety margin built into the stock V/F curve to 'postpone' the felt effects of it. The margin usually on the scale of well over 10 years, and is long enough that it's way more likely for the CPU's VRMs on motherboard to age (degrade) and short like 12 volts into the CPU (or experience some other motherboard failure). This is a reason why dead GPUs are a more common event than dead CPUs.

Further, a big fraction of the performance or frequency harvest from an overclocking effort is about eating into that safety margin. In that an overclocker will usually experience the effects of degradation sooner and more frequently compared to someone running at stock, even if the rates of degradation are similar.

Degradation is not necessarily a catastrophic thing, as the response to it is often merely just bumping up the voltage by 5-20mV. Of course, doing so will accelerate it and the period/time between necessary bumps will narrow, but you should also consider its timescale and hw long you actually care about the CPU staying high-quality.
In a heavily-dramatised hypothetical, say in 4 years, you require an additional 15mV to stabilise at the given frequency, you might then need another 15mV bump in 3.5 years, and then a third in 3 years after that; at that point, you've had the CPU for over 10 years.

For a more in-depth response, check out this video

Well stock settings now a days almost auto overclock to worse thermals than mild/modest well tuned manual overclock.

But technically things are overclocked out of the box.

But will a CPU even degrade at the actual base frequency even with temps 60C or lower all the time and voltages well below safe limit. Or would it last forever with no degradation.

Its amazing to me how so many laptop CPUs I see from work actually run in mid to high 90s during a basic diagnostics and it is well within Dell or Lenovo spec. I am sure those will eventually degrade but have the safety margin you talk about built in.

But if it runs and never exceeds 60C, does it still degrade??
 

·
Registered
Joined
·
228 Posts
Discussion Starter · #10 · (Edited)
Is the 12900K know3n for getting hot? I remember the 11900K was toasty.

Yes it is. 11900K and Rocket Lake in general was even worse by quite a bit. And worse yet only 8 cores max.

With 12900K 16 good cores. 8 of the best in the world IPC and 8 still pretty good lower clocked Skylake/Zen 2 like IPC cores with no SMT/HT.
 

·
Registered
Joined
·
2,045 Posts
Well stock settings now a days almost auto overclock to worse thermals than mild/modest well tuned manual overclock.

But technically things are overclocked out of the box.

But will a CPU even degrade at the actual base frequency even with temps 60C or lower all the time and voltages well below safe limit. Or would it last forever with no degradation.

Its amazing to me how so many laptop CPUs I see from work actually run in mid to high 90s during a basic diagnostics and it is well within Dell or Lenovo spec. I am sure those will eventually degrade but have the safety margin you talk about built in.

But if it runs and never exceeds 60C, does it still degrade??
Degradation exists no matter what; it just quickens depending on how hard you push the chip.
But usually, CPU manufacturers design their chips to last long enough for its usefulness in the overall market, which is usually up to 5-10 years.
That's why they still provide three-year warranties as opposed to other hardware which only provide 1-2 years at most.
But most people tend to upgrade and ditch their chip within 3-5 years anyway.
 

·
Registered
Joined
·
1,929 Posts
Is the 12900K know3n for getting hot? I remember the 11900K was toasty.
12900K runs quite a bit hotter than the 11900K did. I saw negative voltage scaling once core temps hit 90C on the 11900K, the 12900K doesn't even show the same behaviour with the available temperature limits on the same MO-RA3 and CPU block.
 

·
Registered
Joined
·
1,259 Posts
12900KS is designed to give all performance it could at the cost of high power usage and temperature. Even high-end watercooling doesn't help much unless you delid the chip and get a very good paste spread etc.

Some stress tests are not too far off from real world usage, and if the temps are low now with some apps, surely apps or games could take more cores into use later during the years if you are keeping the chip, bringing higher temps too.
 

·
Registered
Joined
·
14 Posts
No problem. Here is my 2 hours of Prime95 blend. Manual OC All core P:50x E:40x. I ran it for 6 hours straight later also. Cooler Noctua NH-D15.
Rectangle Font Screenshot Parallel Pattern


Cinebench R23 for 1 hour stress test. Same OC.
Rectangle Computer Font Line Screenshot
 
  • Rep+
Reactions: z390e

·
Crayon Evangelist
Joined
·
7,756 Posts
there are countless laptops with intel silicon that have been bouncing off the thermal limit, some for as long as a decade, or longer. Thermal throttle is designed into the silicon to keep it alive for at least the useful lifecycle of the silicon. Dont worry about it. The reality is that you will have upgraded two or three times before degradation would actually be an issue.
 

·
Pole Vault Champ & Disco
Joined
·
1,606 Posts
No problem. Here is my 2 hours of Prime95 blend. Manual OC All core P:50x E:40x. I ran it for 6 hours straight later also. Cooler Noctua NH-D15.
View attachment 2560384

Cinebench R23 for 1 hour stress test. Same OC.
View attachment 2560385
I'm not familiar with this CPU, but it certainly goes well against my AMD chip which runs at stock - nice score, how many cores does it have?




 

·
Registered
Joined
·
14 Posts
I'm not familiar with this CPU, but it certainly goes well against my AMD chip which runs at stock - nice score, how many cores does it have?




The i9-12900K has 8 P-cores and 8 E-cores. With hyperthreading it has 24 threads. Here is my so far best score with my 24/7 mild OC.
Rectangle Font Screenshot Computer Technology
 

·
Registered
Joined
·
2,045 Posts
Is the 12900K a good overclocker? I remember the 11900K was hard to overclock without very high voltage and temps.
Insanely good. It just runs insanely hot as well. The limiting factor is heat, not voltage.
Your main issue will be binning chips. If you aren't buying a prebinned chip, you'll want either a KF or a KS. The former being the budget pick.
On a side note, memory overclocking is a bit of a hit-or-miss due to IMC lottery. It's easier on DDR5, but if you don't have strict demands, DDR4 is just fine.
 

·
Registered
Joined
·
1,929 Posts
Is the 12900K a good overclocker? I remember the 11900K was hard to overclock without very high voltage and temps.
The 12900K is easier to overclock, but runs far hotter than the 11900K did at similar power draw. Even if you add the E-cores, the 12900K will still draw less power than the 11900K.
 
1 - 20 of 20 Posts
Top