Question i9-12900K Temp Question

Karadjgne

Titan
Ambassador
It's a 12900k and it's staying under 105°C. You're good.

Coretemp and other such software reports the hottest cores, and you have 16 of them, 8 P-cores and 8 E-cores. What they don't do is read the entire cpu temp. So even if you spike a couple of cores to 100ish °C, that's not cpu temp, that's just 2 core temps. Without looking at the entire list and making judgement on that, it's entirely possible that 2 cores in heavy usage can hit 100ish °C, 6 other P-cores are all sitting at 50-60°C and the E-cores are basically taking a nap at a cozy 40°C.

Which is an entirely different scenario to a Prime95 100°C full cores burn
 
Apr 7, 2022
6
0
10
0
It's a 12900k and it's staying under 105°C. You're good.

Coretemp and other such software reports the hottest cores, and you have 16 of them, 8 P-cores and 8 E-cores. What they don't do is read the entire cpu temp. So even if you spike a couple of cores to 100ish °C, that's not cpu temp, that's just 2 core temps. Without looking at the entire list and making judgement on that, it's entirely possible that 2 cores in heavy usage can hit 100ish °C, 6 other P-cores are all sitting at 50-60°C and the E-cores are basically taking a nap at a cozy 40°C.

Which is an entirely different scenario to a Prime95 100°C full cores burn
Ah ok I see. Thanks for the explanation! :D
 

Karadjgne

Titan
Ambassador
105C?
Why does the Intel Ark website say 100C?
So let's translate this. "Processor Base Power" is a vague "typical" power draw value at an undisclosed Intel workload that's definitely "optimized" to return the desired 125 W value. It's interesting that the i9-12900K, i7-12700K and i5-12600K all run at the same 125 W value in this special workload despite their completely different performance characteristics. "Maximum Turbo Power" is the real limit, the maximum amount of power the processor can draw at stock settings for an indefinite duration of time.

For the i9-12900K, this is 241 W. Unlike past generations of processors that were constrained by the Tau time value to hold maximum power draw, or PL2, Alder Lake processors now run at maximum power draw indefinitely if the load demands it and as long as the processor doesn't hit the thermal limit of 105°C. This is done without inventing a new system; Intel simply tweaked the PL1 and PL2 values and set them both to 241 W, which effectively means the processor can run at 241 W all the time as long as it doesn't overheat. The "125 W" limit now only exists on paper and in marketing documents.
(Techpowerup)

Throttles (Tjmax) at 100°C, will go to 105°C before shutdown is what I'm getting from that. It's not the first time I've seen it reported.

Intel made it absolutely pointless to OC a 12900k/s because they run max turbo on as many cores as are used by the application, without Tau.
 
Reactions: Phaaze88
(Techpowerup)

Throttles (Tjmax) at 100°C, will go to 105°C before shutdown is what I'm getting from that. It's not the first time I've seen it reported.

Intel made it absolutely pointless to OC a 12900k/s because they run max turbo on as many cores as are used by the application, without Tau.
What they say in the article is completely garbled up.
Intel still has pl1=125W and pl2=241W that's processor base power, and it doesn't matter what kind of workload you run or if it is disclosed or not, this is a ceiling, a on/off switch, after tau only that much power goes to the cpu no matter what the cpu or the software asks for, this is still the intel default setting.
The type of workload or disclosure is only relevant when associating a clock speed with the power draw, at 125W (any watt) cinebench will run different clocks from prime95 small fft.

PL1=PL2=241W is the max turbo power and just because all mobos come with this setting or even worse completely lifted power limits ( +MCE +whatever else) where you can get 300W+ it doesn't mean that it is standard, mobos just use it to get better reviews.


Also intel states max temp as tjunction (Junction Temperature is the maximum temperature allowed at the processor die) while most others including tools measure at the heat spreader which will be at a different temp and in general will be about 5 degrees cooler since some of the heat will already be dissipated.
 
Reactions: KyaraM

geofelt

Titan
I think you are good.

From a safety point of view, the cpu monitors it's own temperature and will throttle or even shut off if it detects a dangerous temperature.
That point is around 100c.
The cpu is designed to boost the best cores to a clock above what you might get with an all core overclock. That is what you want for games or other apps that can not load all cores 100%.
This happens very quickly and may not last very long.
Try this simple test.
Run the cpu-Z stress test.
You will likely see a couple of the cores hit 100c.
The good thing is that the cpu keeps chugging along and does not shut down.
In normal use, most workloads will not ever stress the cpu as hard.
In the process, look at the maximum clock rate for various cores.
You are likely to see the max rate on a couple to be surprisingly high. Perhaps 5.5.
 
Reactions: KyaraM

uWebb429

Respectable
May 22, 2020
515
150
2,440
98
most others including tools measure at the heat spreader
All Intel CPUs since the Pentium 4 era report the peak core temperature measured at the hottest spots on the core. Any temperature monitoring software can read this data. The Core Temp program is reading this data. This temperatures data is used to control thermal throttling and thermal shutdown.

Intel no longer publicly documents the shutdown temperature but it used to be set 25°C to 30°C beyond the throttling temperature. This helped prevent any "false trips" the moment the CPU reached 100°C.





As soon as one core hits 100°C, the CPU will start to thermal throttle and slow down. When this happens, the CPU only slows down as much as necessary to keep the CPU a hair under 100°C. Intel thermal throttling works great. Any slight loss in performance might not even be noticed while playing a game.

@bonbons
Will you hurt your CPU if it occasionally reaches the 100°C thermal throttling temperature? That would be highly unlikely. The CPU is always adjusting itself hundreds of times per second to keep the CPU on the safe side of 100°C.

If you need to run work loads that constantly peg multiple cores at 100°C for extended periods of time then you should consider getting a better cooler. You could also reduce the turbo power limits to control the amount of heat the CPU is putting out.
 
Noctua's D15 variants are very good, but, if power and boost duration limits are removed you can not expect temps any lower from an underload (200-240W) 12900K, IMO....

Is your heat sink only equipped with a single center fan?

From Noctua's website, and specifically from their NH-D15S info:
"Users who have sufficient room can also upgrade the cooler with either a 120mm or round 140mm fan on the front fin stack for further improved performance in dual fan mode. "
 

magbarn

Prominent
Dec 9, 2020
56
33
570
1
I had to set my PL1/PL2 to 190 watts on my 12900k and my NHD15 was much happier after that. Yes it's around 1-5% loss, but it's worth it as I game at 4K on my 3090 anyway.
 
Apr 7, 2022
6
0
10
0
Noctua's D15 variants are very good, but, if power and boost duration limits are removed you can not expect temps any lower from an underload (200-240W) 12900K, IMO....

Is your heat sink only equipped with a single center fan?

From Noctua's website, and specifically from their NH-D15S info:
"Users who have sufficient room can also upgrade the cooler with either a 120mm or round 140mm fan on the front fin stack for further improved performance in dual fan mode. "
Yes, I just have the single middle fan
 
Yep for gaming it does seem high...may just be a need to re-apply and reseat the CPU cooler. I run a 12700K overclocked with one core at 5.2GHz going down to 5Ghz all core on an adaptive voltage, and mid 50's to 60's on temps typically playing simulation games like AMS2, Assetto Corsa Comp, Race Room etc. though I am using a 360mm AIO. I know the 12900K does run hot but with some optimisation on vcore etc you should be running a bit cooler.
 

ASK THE COMMUNITY