I9-12900KS reaches 100°C instantly????
-
Hello, friends. It's been a very long time since I've been here, and I'll tell you about the movie I'm suffering with my latest build.
I was coming from an Asus Prime X299 with an I9-7940XE which I had cooled with a Noctua NH-D15S
The thing is, a week ago I decided to upgrade my computer to the latest. Lian-li case, Asus Maximus Hero Z690 motherboard, and a piece of I9-12900KS.
To cool it, I thought I would use the noctua, but after the paste I had spent, I pulled the trigger and got an MSI AIO liquid cooler. Specifically the Corsair iCUE H150i ELITE with the little LCD screen. Just a whim.
I haven't updated the graphics card yet (I have a 1080ti but I'm not a gamer).
Anyway, I put everything together and did a stress test on the CPU.
Suddenly the temperature of all the cores went from about 35º to 100º in less than 1 second, I stopped the test and they went back from 100º to 35º in less than another second...Damn, I thought I hadn't mounted the liquid cooler properly and that it wasn't making good contact or something.
I put it back together. This time I used Noctua thermal paste that I had and things were exactly the same.
This time I left it for about 5 minutes and the temperature of the coolant went from about 35º to 45º but the CPU was at 100ºC SOLID.
Indeed, the air coming out of the radiator started to get hotter. Which means the liquid cooler was working.But as soon as I stopped the stress test, the CPU immediately went to 45º. Same as the coolant. I left it for a while to see if the temperature gradually went down and that's how it was.
Is this normal???????
That is to say, The CPU gets red hot every time you ask for the good stuff??? and then cools down practically immediately???? I don't see any sense in it. It can't be just thermal inertia that it heats up and cools down 60 or more degrees in less than 1 second.
I know these CPUs have a TDP that can reach 250W and they get very hot, but I just don't see it as normal...
-
@dixman said in I9-12900KS Instantly Reaches 100°C????:
Hello, friends. It's been a very long time since I've been here, and I'm here to tell you about the nightmare I'm having with my latest build.
I was coming from an Asus Prime X299 with an I9-7940XE, which I had cooled with a Noctua NH-D15S
The thing is, a week ago I decided to upgrade my computer to the latest specs. Lian-li case, Asus Maximus Hero Z690 motherboard, and a piece of I9-12900KS.
I was planning to use the Noctua for cooling, but after spending so much money, I decided to go all in and get an MSI liquid AIO. Specifically, the Corsair iCUE H150i ELITE with the LCD screen. Just a bit of a splurge.
I haven't upgraded the graphics card yet (I have a 1080ti, but I'm not a gamer).
Anyway, I put everything together and ran a stress test on the CPU.
Suddenly, the temperature of all the cores went from around 35°C to 100°C in less than 1 second. I stopped the test and then they went back from 100°C to 35°C in less than another second...Damn, I thought I hadn't installed the liquid cooler properly and that it wasn't making good contact or something.
I installed it again. This time I used Noctua thermal paste that I had, and everything was exactly the same.
This time I left it running for about 5 minutes and the coolant temperature rose from around 35°C to 45°C, but the CPU was at a SOLID 100°C.
Sure enough, the air coming out of the radiator started to get hotter. Which means the liquid cooler was working.But as soon as I stopped the stress test, the CPU immediately went back to 45°C. Same with the coolant. I left it running for a while to see if the temperature gradually decreased, and that's exactly what happened.
Is this normal???????
I mean, the CPU gets red hot every time you ask for some serious power??? and then cools down practically immediately???? I don't see any sense in it. It can't be just thermal inertia that causes it to heat up and cool down 60 or more degrees in less than 1 second.
I know these CPUs have a TDP that can reach 250W and they get really hot, but I just don't think it's normal...
Well, it seems that according to this:
https://www.google.com/amp/s/elchapuzasinformatico.com/2022/04/intel-core-i9-12900ks-review/amp/
You have a PC capable of frying an egg.
What you don't mention is how you configured the fans on the RL with iCUE, which has several options for both the pump and the fans, which basically are:
Low, normal, extreme.
Best regards -
@Dixman looks like it might be a bug in the temperature sensor or more likely a problem with the software that measures it. What is true is that the 12900KS is very hot, but it is a brutal processor in terms of performance.
If you have a laser thermometer, check the temperatures a bit, or update the Bios, it is clear that even if it reaches 100º it is not possible that it reaches that temperature in an instant, just as it would suddenly drop to 35º... at least that is what I think, I don't know if it is physically possible.By the way, that 1080ti is a great card, even though I am not a gamer
I also have one, but if I am a gamer 
Regards!!
-
Well that's what I say. I think it has to be physically impossible.
The BIOS is updated to the latest version. It was the first thing I did after installing Windows.
I've even disassembled the RL thinking it wouldn't make good contact, but no, the RL was mounted perfectly and the thermal paste I used is NOCTUA NT-H2
Let's see if I record a video and upload it to YouTube so you can see what I'm saying.
-
Hello friends, how long has it been.
@Dixman, it is indeed strange, yes, if it were to go down slowly on a ladder we would doubt the thermal contact, but here it is stranger.
On the one hand, I would be encouraged to try putting on the noctua and contrast, while I analyze and gather information, you might not feel like it much but I wouldn't take the situation lightly.
On the other hand, I'll tell you about my own experience:
I'm also trying out new hardware for the first time in a long time, an R5 3600, and I've noticed that even with an increased heatsink, the behavior is to spike up and then slowly come down.
In my case, and with the heatsink blown but well, I see a lot of sudden heat and poor thermal contact.Yours is strange, even though it's much more micro, but sometimes having a well-cooled heatsink but making poor contact gives similar behaviors (not as exaggerated as it seems to be true). So analyze it well, put a program that gives you real-time graphs of all the values to analyze it well, and keep in mind, that if the heatsink is cold but the contact is a bottleneck, it can perfectly drop "all at once" but not have enough "heat flow" to remove the heat at the necessary rate, causing that spike to the top.
PD: I personally will do lapping as soon as I can, just like the last one, for me, the whole thing of improving and improving the cooling so that the heat keeps going "through a tiny hole" is like having a Ferrari with balloon wheels, and if I devalue it, I couldn't care less, in fact, maybe one day if I sell it, I'll even ask for more because it's lapped.
And turn up the flow to the max on the AIO, try everything, and keep getting details to see what happens. -
@dixman
Out of curiosity, since we both have a similar RL (mine doesn't have a screen, it has the typical leds)
Configured in icue with pump and fans on "normal "
When running the Aida64 stress test, it goes directly to 88 degrees (identical temperature in Aida and in icue) starting from a CPU temperature of 48 degrees.
When stopping the test, it takes about two seconds (I didn't time it) to return to the original temperature.
Regards
P.d I have a Rizen 5900X -
@clipper and with higher levels in pump and fan? What results do you get?
I have a similar problem as you, of course with fewer watts involved and by air, but I am seeing after a long time disconnected from the scene, that apparently it is not so easy to cool the current micros when they start to want to put voltage and frequency.
The i9 KS now I remember that is by itself one of the hottest that exist, already in the first tests, but of course 100 all at once huffff.
What do you want me to say, I have a very made idea of all this, but before I start ranting I'd better see around how people are doing with the issue.
-
How are you? Have you discovered something?
It is clear that modern micros increase voltage like a fool to achieve the frequencies, and that yours in particular is one of the hottest out there, but nobody would want those temperatures, and the truth is there is curiosity to see some graphs or something about how that increase develops.
-
I've been reading around a bit, my brand new r5 3600 is also leaving something to be desired in terms of heat dissipation.
I need to do more tests later, but for now, in the delid of my micro I've seen that they solder the his with gallium, and this not only has a lower thermal conductivity than hard solder, but it turns out that the layer of gallium is very thick!!, a youtuber after delidding and applying liquid metal, and having cleaned the resin perfectly, had to sand the lower edge of the his so that it made contact with the die, in the same photos you can see what looks like several millimeters of gallium between die and his.
This would perfectly explain what we observe.it seems that the new dies are more sensitive to pressure, which occurs with the expansions as well, and there they cure in health with a softer solder that accommodates the volume changes a bit, and with a thicker layer for the same purpose.
that's what I read earlier, and everything adds up, really.For now I'm not going to touch anything, when I get back from my week of vacation I'll get a laser thermometer and I'll come to more empirical conclusions.
I have more than acceptable temperatures, but I can't get over the idea of having banned access to all the performance that's left because of a "thermal bottleneck".If I discover something worthwhile I'll keep commenting.
