Hitting 90-100C at stock with a 13900K in CB23 and an AIO is sadly normal yes. 88C in a game is more of a concern. Have you set your AIO pump and fans to max during Cinebench?

 

Your Intel Core i9-13900K reaching 100°C in Cinebench and 88°C in gaming is worrisome. First, confirm the AIO cooler is mounted correctly and thermal paste is applied evenly. Optimize fan configurations for better airflow within the case. Experiment with unde revolting and BIOS settings to strike a balance between performance and temperature. Consider upgrading to a more efficient CPU cooler if needed. Check for any software tracking issues causing excessive CPU usage. Regularly monitor temperatures to ensure they stay within safe operating limits.
I hope it will help.

 

Yea, cinebench temp is ok, but gaming idk why is hitting so much, is it only in this game?
Your CASE fans are front intake and exhaust rear, and tell us rpm on aio and fans in case how fast they spin. Are u used thermal paste for cpu the one that comes with AIO?
This score in cinebench is normal for stock 13900k.

 

MSI MAG AIOs have problems with clogging. Your higher temps may be an early warning before clogging gets worse.

 

This here. Known issues with msi aio

 

If that cooler is running perfect it still isn't going to keep up with a 13900k. The best AIO's struggle with it.

 

Apparently even custom loop does too, did you see pipes' post above?

Anyway, best way to fix it is to delid and direct die mount - but then that will void warranty.

 

it seems I'm not the only one having the 100 degree problem in chinebench, I also notice my loop made up of 2 360 radiators, 9 fans to cool the liquid, a dcc3.1 pwm pump and a full-nickel EK supremacy . I have also used the termalright retention system, which seems to get worse, as far as I understand it seems. hey there are CPUs that have concave ihs or some sort of problem

Swapping the 3.1 out for the 4.2 may help. The 3.1 was dogging down my loop as well. It was ok with 1 rad but anything more and the pressure just wasn't enough. Just my 0.02

 

100C is perfectly fine. Even 110C-115C with constant 350W is nothing to be worried about. And the chip can draw over 500W as long as you can cool more.

These silicon has been improved. It's way more important to keep the ample voltage so you don't see typical unstable 13900K crashes all over the places in this forum.

 

Even 110C-115C with constant 350W is nothing to be worried about.

Well it's your chip... I guess you know better than Intel engineers.....right.

 

It's not me getting unstable crashes out of anything that requires merely 100W.

Running 350w is not even half the potentials of 13900K.

And Intel engineers don't do overclock with consumer level cooling. They run 13900K and they know it needs 1.5V to be fully stable at 6GHz lol.

 

Swapping the 3.1 out for the 4.2 may help. The 3.1 was dogging down my loop as well. It was ok with 1 rad but anything more and the pressure just wasn't enough. Just my 0.02

I said ddc 3.2 but honestly I no longer remember what the exact number is, however I remember that the tests did not give big differences with the new revisions, in case I change it, since I mounted the heatsink case for that pump and the cylindrical tank, they will be compatible ?
P.s it's a DDC 3.1

 

One year is not long...

 

But considering others posting “degradation” when say crashing at way less wattage loading shaders in a game might simply need more voltage in the first place.

 

Again 100C-110C 350W is perfect fine

It is NOT only about the CPU. That kind of heat over time could kill the pump in an AIO. Short bursts up to 100C is fine yes, but not for extended periods of time. You do what you feel is okay for you.

 

Oh really? So the heat kills the pump while the water inside the pump usually stay below 50C lol.

 

You're trying to cool 280W package with a 280 AIO. Given that it doesn't shoot straight to 100c during Cinebench on all cores, I'd say its about what you can expect from a 280 AIO.

What you can do to improve temps is mount the AIO at the front as intake, in case you have it put up top as exhaust. This will drop temps by about 5C and keep you within Tjmax on all core load.

I see you're using about 1.23v which is a little much for your stock all core frequency. This should be adjusted dynamically to give you higher vcore for light load so it can boost to 5.8Ghz, but lower vcore when all core load is applied. You could be stable with about 1.20v under all core load, which would reduce your package draw to ~250W, dropping temps even more.

The most concerning part tho, is the System Agent voltage at 1.33v. You are dangerously close to 1.35v which is known to case CPU degradation, I would not run it beyond 1.3v and you really don't need that much anyway. I'm at 1.25v with highly tuned RAM and its perfectly stable. You should drop it to the same value and test stability.

 

If he was using let say, arctic freezer II 360/280 temps could be much much better.

 

Sure i get that. Delid/relid - direct die is much better for temps. I guess its the nature of the beast with 13900k. A fast pumping 360 aio is needed to run reasonable temps. Alot of aios dont pump fast. Look for ones with high pump speed.

High pump speed can also cause worse temperatures via turbulence, you need to test a range of flow rates combined with your airflow.

 

A fast pumping 360 aio is needed to run reasonable temps. Alot of aios dont pump fast. Look for ones with high pump speed.

Too bad most of these AIOs need bloatware to control them. It's one of the main reasons why I bought the Arctic LF2 360, as I can control it through the BIOS fan settings.

 

Better pumps, like D5s, are rated for 60C. That means that, at 60C, they will probably survive at least as long as their warranty period. That does not mean that temperatures close to that but still below it will not decrease the pump's real useful life. Higher heat generally decreases the life of components across the board. A pump run at 50C may last 5 years, maybe 10, but whatever lifespan that specific unit had it almost certainly would have been longer if it were run at only 40C.

How much that matters to you is a subjective matter, but by extension, telling other people they should run their pump at a higher water temperature is making an assumption about how much risk they are willing to take on in terms of the reliability of the pump. The same applies to a CPU or any other component that is known to degrade with heat. Maybe someone buys a new high end CPU every year and doesn't care if they've taken the lifespan from 10 years to 2 by regularly running it at 100C, but maybe they only build once every 7 or 8 years and don't want to risk having to blow a bunch of money in 2 or 3 years.

 

This really opened my eyes to the fact that there are many concerns about whether only 350W can stress a cooling component, such as a pump designed for cooling.

Even a pump from a 120mm EVGA 1080Ti still works like new after entire five years with 300+W.

If you want better cooling you should suggest to chip makers that they consider creating a larger die for a more extensive cooling surface, instead of being overly concerned about pumps.

 

I lost the thread a bit, now I don't know if my costum cooling system is or how much better than yours, it's not a competition, but I can tell you that by following the guide (that incredible guide written by @RobertoSampaio
) I am managed to reduce the total consumption of the CPU which does not exceed 250 watts and from the screenshot you can see the temperatures, if you want some advice try this tuning too, at least reduce the temperatures and if you don't succeed then it will be a possible and hopefully not a problem tied to your AIO

 

Glad you got your temps sorted, but I see your RAM is only at 5600MT/s in gear 2. That's not great. I think performance on Intel scales pretty well up to the mid 6000s, so it may be worth looking into either overclocking that RAM or replacing it (if that's in your budget).

 

Glad you got your temps sorted, but I see your RAM is only at 5600MT/s in gear 2. That's not great. I think performance on Intel scales pretty well up to the mid 6000s, so it may be worth looking into either overclocking that RAM or replacing it (if that's in your budget).
[/CITAZIONE]

I only just started to get involved in overclocking, I had pushed the RAM up to 8000 MHz but then I abandoned it for a while...mount hinyx A-die

 

Oh good... had me worried. Think of all the sweet, sweet MHz you could have!

 

Funny I have been running daily long time rendering with 350W 100C all core 5.7GHz for almost a year now.

Nothing happened. It doesn't have a single crash, no degradation whatsoever.

It seems you don't know the product you bought.

With Jedec spec ram I suppose.

 

Ram has nothing to do with how much longer and hotter IMC can take. After a few generations of IMC the frequency of the same ram goes up regardless.

 

The interconnects in these newer chips are getting smaller and smaller and therefore the amount of current they can withstand is the limiting factor.

Simply put, this chips will not last long running tons of current through them.