i9 12900k temperature problem

[looniam]

So you are saying that a 10ms 100w spike can cause a 2000ms (that's 2 seconds btw...) 20-25c spike in temperatures? On a 360 AIO? Nah. I mean there is one way to test it, power limit it to 150w and run cbr20. But the op is gone so...can't do that.

on that POS AIO? feh like a higher end air cooler can't beat that. .
lesson for the day:
power fluctuates much much MUCH faster than temps

and we are done.

 

[JustBenching]

I'll have my ddr5 in 2 day and I'll test it, with no power limits there is no way in hell it's going to be hitting 90+c during gaming. Not even with an aircooler, let alone a 360 AIO. No matter how bad it is, it's still a truckload of rad space.

Btw you can make hwinfo poll every 50ms.

 

[plastiscɧ]

note to self:
ask the source of the graphic what they were playing while it was created.
could only be mahjong

let's be honest and realistic: nearly EVERYONE says the thing is running hot... why make such a scene?

I also don't understand why it has to be an i9 if it clocks the same as all the others. or does it run at 6Ghz?

 

[Psychoholic]

Anecdotal of course, but mine runs cool.. about 75C under full load (CB23 Loop) and this is on air.. noctua U12A.
By default motherboards overvolt the crap out of these, this has to be adjusted. Mine was pouring on 1.34v by default, i'm running it @ 1.24v (Still stock clocks) and its rock solid stable.

 

[JustBenching]

note to self:
ask the source of the graphic what they were playing while it was created.
could only be mahjong


View attachment 230422

He was playing cinebench r 20 I think.

 

[Vayra86]

He was playing cinebench r 20 I think.

No, Prime.

The point is, you have no control over when or whether the CPU will peak to 241W. It simply can as it is stock specced for PL1=PL2 and mobo vendors are not here to be kind on your cooling, they want to say they're faster with the same CPU. Remember past Intel gens and how vendors sneaked in OC settings. The CPU maxing at 150 ~W simply means it ran into constraints earlier but it sure as hell peaked to 241W briefly before landing there. Its how these turbo's work.

Three pages on, and we can conclude this *might* have happened on OP's CPU.
Somehow that wasn't true on the first pages, and now its a possibility.

You have to account for people being people, and oblivious to what they're actually doing, or what settings they're actually using. And the results simply line up with what reviews see. Its not hard and there's no need to present your own magical results that don't line up to show how much better you know things, unless it holds water for OPs case . The reality is, you have a different setup just as well even if you do have an ADL CPU.

 

[JustBenching]

No, Prime.

So it wasn't a game and it wasn't mahjong.

About the rest, supposedly tommorow theyll be shipping my ddr5, i guess ill have it sometime next week. Ill do my tests, but I can bet a paycheck the temps in gaming even with an overclocked cpu will not be anywhere near 90. And that's with an air cooler, not a huge radiator. Actually i dont think theyll even exceed 70,but ill have to see, ill keep you informed.

 

[Jeanbon]

Hey, just created an account to reply here and bring my experience with the 12900k !

First, my professional experience :
I work at the IT dep of a company that need a lot of compute power.
Recently, we built 5 custom workstation for our 3D Cloud point dep. We used 5 12900k on my recommendation. All cooled with a 360 AIO coller (MSI MEG C360)

What I can say with those CPU after 4 weeks of work (and a bit of gaming in our breaks but don't tell our boss ) :
CPU temp is stable at idle. (30°C)
On workload, some "spicks" can be seen around 85 to 90 for P core, but an avg of 80. 100% CPU use
On "Gaming", 65 to 70 avg BUT yes, we have spikes that reach 85... On the 5 Workstations. ~25-50% CPU use

Now what I can say about my personal experience :
With a custom loop, one 360 Rad, Velocity² cpu waterblock and around 0.98L (very good thermal inertia) of coolant only for the cpu, positive pressure case, good fan on every intakes...
I get 25C on long idle and 29-30 on short idle... nice
Ureal Engine 5 compilation (cpu at 100% for 30 min on big project) : little spikes, 70C avg... nice
Intel stress test : 80C... ok cool
Now gaming (on 4K monitor) :
MSF2020 : spikes 85C, avg 70 | DCS world : spikes 83C avg 73 | ready or not : spikes 85C avg 75| my game : spikes 80C avg 69

mmmmh, why does all the 6 CPU have those spikes ?! Here are my answers

1- The GPU... actually when I was playing, my GPU fans were not turning fast... but sometimes they kick off and go to 100%. Even with a really really good front to rear airflow in your case, you will have always hot air that goes to your radiator. (assuming that he is in the top of your case)
2- 12Gen CPU are spiking. I was really scared first because that's the first time that a cpu does it So much , but "new" tech, new problems and I check on all the internet to see that 95% of the 12900k owner have spikes
3- Thermal paste quality ? Actualy, we used 2 different paste in our build, but nothing changed so...

My solutions : (if you think that you really have a problem)
Change the radiator place, put it in an intake of your case (but don't forget to always keep your pump as low as possible in your loop).
Blow air as much as possible from the front to the rear of the case.
Lower your cpu Voltage, intel extreme tuning software is realy cool for that.
Touch with your hand the inlet and outlet tube of your AIO and if there is no difference, then the radiator might be at fault.

Sry if I misspelled some word, I'm not very good in Eng writing.. I feel like I write like a child

 

[plastiscɧ]

Hey, just created an account to reply here and bring my experience with the 12900k !

First, my professional experience :
I work at the IT dep of a company that need a lot of compute power.
Recently, we built 5 custom workstation for our 3D Cloud point dep. We used 5 12900k on my recommendation. All cooled with a 360 AIO coller (MSI MEG C360)

What I can say with those CPU after 4 weeks of work (and a bit of gaming in our breaks but don't tell our boss ) :
CPU temp is stable at idle. (30°C)
On workload, some "spicks" can be seen around 85 to 90 for P core, but an avg of 80. 100% CPU use
On "Gaming", 65 to 70 avg BUT yes, we have spikes that reach 85... On the 5 Workstations. ~25-50% CPU use

Now what I can say about my personal experience :
With a custom loop, one 360 Rad, Velocity² cpu waterblock and around 0.98L (very good thermal inertia) of coolant only for the cpu, positive pressure case, good fan on every intakes...
I get 25C on long idle and 29-30 on short idle... nice
Ureal Engine 5 compilation (cpu at 100% for 30 min on big project) : little spikes, 70C avg... nice
Intel stress test : 80C... ok cool
Now gaming (on 4K monitor) :
MSF2020 : spikes 85C, avg 70 | DCS world : spikes 83C avg 73 | ready or not : spikes 85C avg 75| my game : spikes 80C avg 69

mmmmh, why does all the 6 CPU have those spikes ?! Here are my answers

1- The GPU... actually when I was playing, my GPU fans were not turning fast... but sometimes they kick off and go to 100%. Even with a really really good front to rear airflow in your case, you will have always hot air that goes to your radiator. (assuming that he is in the top of your case)
2- 12Gen CPU are spiking. I was really scared first because that's the first time that a cpu does it So much , but "new" tech, new problems and I check on all the internet to see that 95% of the 12900k owner have spikes
3- Thermal paste quality ? Actualy, we used 2 different paste in our build, but nothing changed so...

My solutions : (if you think that you really have a problem)
Change the radiator place, put it in an intake of your case (but don't forget to always keep your pump as low as possible in your loop).
Blow air as much as possible from the front to the rear of the case.
Lower your cpu Voltage, intel extreme tuning software is realy cool for that.
Touch with your hand the inlet and outlet tube of your AIO and if there is no difference, then the radiator might be at fault.

Sry if I misspelled some word, I'm not very good in Eng writing.. I feel like I write like a child

Thank u very much for the detailed testing experience!

that's exactly what we need in forums like this one.
i think that should clear the table of irregularities.

---

and something for you in a personal matter:

we Europeans have an excellent translator on our hands.
i also work with it in my administration because the pro version is also document proof.
feel free to try it out. forget google. it is nuts.

deepL>>>> translator

 

[Vayra86]

Anecdotal of course, but mine runs cool.. about 75C under full load (CB23 Loop) and this is on air.. noctua U12A.
By default motherboards overvolt the crap out of these, this has to be adjusted. Mine was pouring on 1.34v by default, i'm running it @ 1.24v (Still stock clocks) and its rock solid stable.

Hey, just created an account to reply here and bring my experience with the 12900k !

First, my professional experience :
I work at the IT dep of a company that need a lot of compute power.
Recently, we built 5 custom workstation for our 3D Cloud point dep. We used 5 12900k on my recommendation. All cooled with a 360 AIO coller (MSI MEG C360)

What I can say with those CPU after 4 weeks of work (and a bit of gaming in our breaks but don't tell our boss ) :
CPU temp is stable at idle. (30°C)
On workload, some "spicks" can be seen around 85 to 90 for P core, but an avg of 80. 100% CPU use
On "Gaming", 65 to 70 avg BUT yes, we have spikes that reach 85... On the 5 Workstations. ~25-50% CPU use

Now what I can say about my personal experience :
With a custom loop, one 360 Rad, Velocity² cpu waterblock and around 0.98L (very good thermal inertia) of coolant only for the cpu, positive pressure case, good fan on every intakes...
I get 25C on long idle and 29-30 on short idle... nice
Ureal Engine 5 compilation (cpu at 100% for 30 min on big project) : little spikes, 70C avg... nice
Intel stress test : 80C... ok cool
Now gaming (on 4K monitor) :
MSF2020 : spikes 85C, avg 70 | DCS world : spikes 83C avg 73 | ready or not : spikes 85C avg 75| my game : spikes 80C avg 69

mmmmh, why does all the 6 CPU have those spikes ?! Here are my answers

1- The GPU... actually when I was playing, my GPU fans were not turning fast... but sometimes they kick off and go to 100%. Even with a really really good front to rear airflow in your case, you will have always hot air that goes to your radiator. (assuming that he is in the top of your case)
2- 12Gen CPU are spiking. I was really scared first because that's the first time that a cpu does it So much , but "new" tech, new problems and I check on all the internet to see that 95% of the 12900k owner have spikes
3- Thermal paste quality ? Actualy, we used 2 different paste in our build, but nothing changed so...

My solutions : (if you think that you really have a problem)
Change the radiator place, put it in an intake of your case (but don't forget to always keep your pump as low as possible in your loop).
Blow air as much as possible from the front to the rear of the case.
Lower your cpu Voltage, intel extreme tuning software is realy cool for that.
Touch with your hand the inlet and outlet tube of your AIO and if there is no difference, then the radiator might be at fault.

Sry if I misspelled some word, I'm not very good in Eng writing.. I feel like I write like a child

@Zazii This is more evidence that your cooling is not the issue, but rather the voltage your motherboard is providing this CPU. Work from the BIOS/Software side to get this puppy under control.

Note the 1.34V and being hot. I have the same experience on a much older, yet similar architecture with my Coffee Lake CPU. Intel CPUs have been bursty bitches since a few years now, all to win those benchmark charts, screw user experience out of the box. Anything above 1.3V really becomes hot territory with Intel, except if you stick with lower core count CPUs combined with their (standard now) thinner IHS solutions. When I pushed 1.35V through on air I managed to run into thermal shutdowns during summer

Another note, if you live in a damp/hot country/environment, your ambient conditions are certainly going to be worse than what you see on Youtube videos. Youtube is like TV: All Show. Avoid that nonsense for truth seeking.

 

[plastiscɧ]

@Zazii This is more evidence that your cooling is not the issue, but rather the voltage your motherboard is providing this CPU. Work from the BIOS/Software side to get this puppy under control.

Note the 1.34V and being hot. I have the same experience on a much older, yet similar architecture with my Coffee Lake CPU. Intel CPUs have been bursty bitches since a few years now, all to win those benchmark charts, screw user experience out of the box. Anything above 1.3V really becomes hot territory with Intel, except if you stick with lower core count CPUs combined with their (standard now) thinner IHS solutions. When I pushed 1.35V through on air I managed to run into thermal shutdowns during summer

Another note, if you live in a damp/hot country/environment, your ambient conditions are certainly going to be worse than what you see on Youtube videos. Youtube is like TV: All Show. Avoid that nonsense for truth seeking.

that's right, it was the only way to deal with the hotheads of that time:

I could bring him down from over 90°C to 65°C under load.
today, most of them are soldered. it is now nearly mission impossible to do the delid without a VERY high risk to destroy them entirely.

 

[ir_cow]

Can we finally conclude most MBs have PL1=PL2 enabled by default. That allows the CPU to pull 240w. I noticed that some MBs will do the 4096w unlock when I have a the pump on the fan header.

You can enforce the 150 limit in the BIOs, but often it is hidden away.

 

[Psychoholic]

Can we finally conclude most MBs have PL1=PL2 enabled by default. That allows the CPU to pull 240w. I noticed that some MBs will do the 4096w unlock when I have a the pump on the fan header.

You can enforce the 150 limit in the BIOs, but often it is hidden away.

Yep, this also works, i have my PL2 set to 200w.
I did notice that if you set it to say 170w or so it will begin dropping below default all core load clocks where lowering the voltage keeps the default clocks.

Dropping PL2 from 241 down to 180-190 is almost ZERO difference in performance and shaves off 50-60w load power usage; of course this makes no difference in MOST games.

 

[plastiscɧ]

Can we finally conclude most MBs have PL1=PL2 enabled by default. That allows the CPU to pull 240w. I noticed that some MBs will do the 4096w unlock when I have a the pump on the fan header.

You can enforce the 150 limit in the BIOs, but often it is hidden away.

funny how intel leaves this mess to the mainboard manufacturers to smooth out.
the 241Watt crowbar only came to stay competitive against the RYZEN CPUs.
they couldn't just handle it another way.

 

[tox1c90]

Hey guys! I just registered to share some measurements I took during my first gaming sessions with my new 12900k build.

Baseline information (measured using hwinfo64):

• 12900k on MSI Z690 Tomahawk DDR5, Arctic Liquid Freezer II 360 (standard fans, MX-4 paste), Fractal Meshify 2 case, radiator outtake mounted at the top
• VCore Adaptive+Offset (-0.050V), PL1/2 = 241W, resulting in ~1.25V VCore during Cinebench R23 and Prime95 @ 4.9GHz all core
• 30 minutes Cinebench R23 -> max CPU package temp. 93°C, avg. power 242W
• 3 hours Prime95 30.7b9 blend -> max CPU package temp. 95°C, avg. temp. 89°C, avg. power 242W

Ok, now the gaming stuff. I played one hour of Battlefield 5 with the GPU (RTX 2080) being maxed at 99% usage all the time. I resetted the hwinfo64 log when I was ingame to accurately capture the average value.

• avg. CPU package temp.: 67°C, avg. power: ~100W
• max. CPU package temp.: 78°C, max power: ~165W
• min. CPU package temp.: 50°C

So it is obvious that it is oscillating a lot, which I could also observe looking at the individual core temperatures. The 78°C resulted from a pretty harsh temperature spike in a single core, usually it was not going above 72-74°C. The game really lets the CPU draw >150W from time to time.

I also have to say that I like to have it as silent as possible. That means up to 60°C I run pump and fans pretty much on minimum and let it ramp up to 50% @ 75°C and 100% @ 85°C. That means basically if there is a harsh temperature increase it takes some time until my cooling has ramped up.

Therefore, I assume the temps are pretty much in line to what has been observed in many other reviews? What's your opinion? I could start fiddling around with stuff like liquid metal paste and so on, but I am not sure if it's really worth it. I also think the cooler is mounted properly, because idle temps are perfect (around ~30°C or even slightly less at 22°C ambient temperature).

 

[plastiscɧ]

Hey guys! I just registered to share some measurements I took during my first gaming sessions with my new 12900k build.

Baseline information (measured using hwinfo64):

• 12900k on MSI Z690 Tomahawk DDR5, Arctic Liquid Freezer II 360 (standard fans, MX-4 paste), Fractal Meshify 2 case, radiator outtake mounted at the top
• VCore Adaptive+Offset (-0.050V), PL1/2 = 241W, resulting in ~1.25V VCore during Cinebench R23 and Prime95 @ 4.9GHz all core
• 30 minutes Cinebench R23 -> max CPU package temp. 93°C, avg. power 242W
• 3 hours Prime95 30.7b9 blend -> max CPU package temp. 95°C, avg. temp. 89°C, avg. power 242W

Ok, now the gaming stuff. I played one hour of Battlefield 5 with the GPU (RTX 2080) being maxed at 99% usage all the time. I resetted the hwinfo64 log when I was ingame to accurately capture the average value.

• avg. CPU package temp.: 67°C, avg. power: ~100W
• max. CPU package temp.: 78°C, max power: ~165W
• min. CPU package temp.: 50°C

So it is obvious that it is oscillating a lot, which I could also observe looking at the individual core temperatures. The 78°C resulted from a pretty harsh temperature spike in a single core, usually it was not going above 72-74°C. The game really lets the CPU draw >150W from time to time.

I also have to say that I like to have it as silent as possible. That means up to 60°C I run pump and fans pretty much on minimum and let it ramp up to 50% @ 75°C and 100% @ 85°C. That means basically if there is a harsh temperature increase it takes some time until my cooling has ramped up.

Therefore, I assume the temps are pretty much in line to what has been observed in many other reviews? What's your opinion? I could start fiddling around with stuff like liquid metal paste and so on, but I am not sure if it's really worth it. I also think the cooler is mounted properly, because idle temps are perfect (around ~30°C or even slightly less at 22°C ambient temperature).

it seems to match to the test done in #58. thank u as well.

I am currently very happy that my PC consists entirely of GEN4 and is allowed to age.
Based on your statements, I'm def from this step to go to always have the ABSOLUTELY latest on the table, although the old was not even close to 6 months OLD.

2 things must happen now: x86 will die. there will have to be a solution for this and the future PC hardware must be more energy efficient with smaller architecture.

before that, i won't lift a finger.

 

[tox1c90]

I managed to bring it down by 3-4°C. I read the reports about bent sockets that can occur with certain boards which can lead to a slightly bent heat spreader when the CPU is inserted into the socket, so I wanted to check if this is the case for me and maybe decrease my cooling performance.

Fortunately, both the copper plate of the cooler as well as the HS of the CPU is as flat as I was able to measure with the tools I have. Although the thermal paste looked as it had proper contact over the whole surface, it looked like there was a tiny bit of a ramp from left to right. So this time I took more care than ever while tightening the cooler to make sure it has perfectly even contact.

Now I have 89°C max., 87°C avg. @ 242W in CB R23 and 80°C max., 77°C avg. with power consumption capped @ PL1/2=200W. Seems to be the best I can achieve.

 

[caroline!]

Overclocked with AI overclock, temps jumped up to 86 even.

Makes sense since crap auto-OC tools pump a lot of voltage to circumvent the silicon lottery and make all chips overclock the same. Don't use them.

Like the rest said, temps are normal considering your cooling. AIOs like that are on par with midrange air coolers.

 

[outpt]

Don’t know anything about alder lake. Since the system is brand new I would look at the cooler. When I built my 5800x everything was alright considering these cpus run hot anyway. Temperature started going all over the place. One drt cooler. That’s my 2 cents

 

[GerKNG]

You dont cool 150W easily on any air tower at all. Im not sure you know what you are saying here. Detached from reality it surely is though. When I push over 140W sustained on a Dark Rock Pro Im seeing 80+C just the same.

that's absolutely no Problem.

My NH D15 has zero problems cooling my 10900k at 5.1 Ghz (that pulls almost 260W in R23)
and my 12600k at 5.1 Ghz (with a way higher thermal density) runs at around 90°C because of the almost 1.45V

cooling the 12600k at a reasonable 4.9 GHz (around 150W... what you said is "not easy on any air cooler at all") sits in the LOW 60s even after a 30 Minute loop and barely audible fans set to a bit above 1100 RPM.

Edit: 90°C... not 80.

 

[Deleted member 24505]

that's absolutely no Problem.

My NH D15 has zero problems cooling my 10900k at 5.1 Ghz (that pulls almost 260W in R23)
and my 12600k at 5.1 Ghz (with a way higher thermal density) runs at around 80°C because of the almost 1.45V

cooling the 12600k at a reasonable 4.9 GHz (around 150W... what you said is "not easy on any air cooler at all") sits in the LOW 60s even after a 30 Minute loop and barely audible fans set to a bit above 1100 RPM.

The D15 is probably the best air cooler there is though.

 

[ir_cow]

My NH D15 has zero problems cooling my 10900k at 5.1 Ghz (that pulls almost 260W in R23)

Can you sustain that though? I had a D15 on my 10900K as well at first. It couldn't handle R23 or Prime95 at stock for more than 3 minutes without throttling. Put a waterblock on it and 5.1 @ 1.275v was the best. Never breached 80c again. Now if you are just playing games or boosting 1 core to 5.1, it isn't a issue using a air-cooler. Even for Alder-Lake.

 

[GerKNG]

Can you sustain that though? I had a D15 on my 10900K as well at first. It couldn't handle R23 or Prime95 at stock for more than 3 minutes without throttling. Put a waterblock on it and 5.1 @ 1.275v was the best. Never breached 80c again. Now if you are just playing games or boosting 1 core to 5.1, it isn't a issue using a air-cooler. Even for Alder-Lake.

i never do anything else except all core OCs and no AVX Offsets.

i have a well ventilated case and that's it.
i talk about R23. not games. and it was my 10850k not 10900KF. but of course your 10900k throttles in P95 but in R23? either you have very high ambient temps, bad case airflow or non just untouched fan speeds.

5.1 on the 12600k is barely stable except for light all core load or gaming. but running R23 once or twice is always below 90°C (btw there is a typo in my comment above)

 

[mrthanhnguyen]

A guy who delid my 12900k told me the stock ihs is not completed contact with the die. Thats why some Alder Lake chip has very high temp.

 

[GerKNG]

5 and 5.1 Ghz OC (1.36V and 1.46V)
NH D15 Chromax Black in a Meshfiy S2 with 4 Arctic P14 Case Fans.
20C Ambient.