h100i rgb platinum se - cpu load temps 50c over coolant

I am running my 3600x stock with a Corsair 240 AIO (h100i plat se).

my load temps are ~85c, a full 50c over my coolant temp. idle is ~47c, +18c over coolant

is this reasonable? or do I have bad contact? Should I repaste it (using stock corsair TIM that was on coldplate)

Check your voltage first. Most people in that range are going be on auto voltage with motherboard “auto overclocking” unknowingly enabled. Ryzen 3000 owners also tend to see elevated Vcore levels at idle and the value does not relax as expected.

Check your voltage first. Most people in that range are going be on auto voltage with motherboard “auto overclocking” unknowingly enabled. Ryzen 3000 owners also tend to see elevated Vcore levels at idle and the value does not relax as expected.

 

My CPU is in "auto", the default behavior, and is running around 1.4v when doing stuff.

The alternative is "manual", setting a voltage manually and running an all-core steady clock rate. Obviously, that is not ideal, and trying to force a lower voltage leaves a lot of performance on the table:

 

[ame]

[/ame]

 

It's unfortunate that Ryzen behaves this way which leads to higher than necessary voltage (and higher temps) in many cases, but that's not really my question here...

 

Is this cooler really not capable of keeping a package drawing 100w cooler than 50c delta? (35 coolant 85 cpu)??

 

BTW i reseated the cooler just to make sure. The paste spread looked good. It was fresh enough and enough of it that I left it. Same temps as before.

Edited September 25, 2020 by sorokyl

You’re misunderstanding the data. CPU wattage (TDP) is the coolant rise in degrees it’s watts in vs watts out. Most Ryzen lower core count CPUs will not see a large coolant rise because they are relatively low in terms of watts.

 

The CPU temperature is not a function of watts. It’s the physical properties of the material and voltage. A 1 core CPU with a TDP of 20W will still get very hot at high voltage and it won’t matter what cooler is bolted on. The heat must pass through the cpu and be conducted elsewhere and there is no getting around the voltage problem. It is the limitation for all of us.

 

I can’t say the exact Vcore needed for your desired clocks, but hand tuning is about the only place you can make gains. Most auto settings will be deliberately padded to make sure the worst cpu ever made still boots up. It is unlikely you are the owner of that prize and cutting off the excess is the place most of us make gains, overclock or stock settings.

The CPU temperature is not a function of watts. It’s the physical properties of the material and voltage. A 1 core CPU with a TDP of 20W will still get very hot at high voltage and it won’t matter what cooler is bolted on.

A little basic physics for you:

P=IV ... Power is a function of voltage and current. (power is measured in watts)

Heat = Energy. Energy is power over time (1 watt = 1 joule per second)

Temperature and heat... well I don't have to explain that.

So yes, the CPU temperature is certainly a function of wattage. The wattage tells you how much heat you're going to need to disspiate. 1V 100A is the same as 10V 10A in terms of how much heat needs to be disspiated.

 

The heat must pass through the cpu and be conducted elsewhere and there is no getting around the voltage problem. It is the limitation for all of us.

 

To be honest, the main issue most of "us" (corsair users) have with voltage is the ****ty iCue software that polls the CPU in a way that prevents it from going into low power states, causing 1.4V and high temps at idle, with no way to disable this undesired polling without completely closing the software.

 

My main concern / issue was is my cooler performing as well as it can. I Guess the answer is yes, and 50d delta under full load is the best I can hope for with this cooler.

A little basic physics for you:

 

You're not in a position to lecture me on physics and I am not going to spend a lot of time trying to unravel someone who doesn't understand the basics principles while tossing sophomoric equations about. Heat is a form of energy, but P does equal T. That should have been in your first year introductory class. If power is the sole determining factor for CPU temperature, than it should be impossible for me to keep my 10900K at 66C pulling 240W with 42C differential on the same cooler. By your assessment of the relationship, all TR and 3950X owners with substantially higher TDP levels than you should have even greater coolant to CPU temp differentials. Of course that's not the case and they are not running 1.0v on their Vcore either. So again, you misunderstand the data and the relationships.

 

 

 

My main concern / issue was is my cooler performing as well as it can. I Guess the answer is yes, and 50d delta under full load is the best I can hope for with this cooler.

 

That is as low as you are going to go at that voltage regardless of the cooler attached at the end. The fact that the coolant to CPU temp differential changes substantially between idle and full load, despite the voltage already being high should tell you there is more going on than what you can see. The idle behavior on Ryzen 3000's is somewhat problematic, but that is going to be an issue with power consumed at idle. It will not add extra voltage or watts to the maximum load state you are assessing now. That is also the problem. If your coolant to CPU differential is +50C and your maximum acceptable CPU temperature is 85C, then it means you must keep the coolant below 35C at all times. That is right on the edge of possibility for most people in the warmer months.

 

You can set the fans to a fixed speed, quit CUE and the Corsair services, and test your yourself. You won't be able to see coolant temperature, but then it is isn't going to change rapidly anyway. The coolant to CPU differential can be assessed instantly with any fixed load test like Prime, CPU-Z bench, OCCT Linpack, or anything else that is a fixed load. That differential should hold constant for these fixed load tests with incremental +1C changes to CPU temp as the coolant temperature increases +1 over time and there is your function of power/watts. Highly dynamic stability tests like XTU, AIDA, and normal OCCT are quite variable and make the relationship difficult to see.

Edited September 26, 2020 by c-attack

I don't know if your board has PBO enabled by default but you may want to test with and without it enabled. Sometimes the big rise in temps it induces isn't worth the slight uplift in performance.

 

Ryzen master has some presets you may want to play with too.. but it's worth checking the bios settings in more detail because you may have surprises.. the only R5 3600 build i did on a gygabyte mobo was giving me over 1.5V vcore in some instances..

 

Even without that, Ryzen does run hot, but the temps you see are quite high, so there's more to it than the H100i perfs.

It's very focused heat, completely offset on one corner of the IHS since the 3600x only has two CCX (which is why der8auer designed an offset bracket for Ryzen to drop temps even further ).. so your voltage will have a lot more influence on the temps than the AIO size or fan speed.

 

And yeah.. badly tuned overclock tests on 10900k.. 310W, but 85°C.. there's more to it than watts. the biggest culprit is usually Vcore, and all settings touching vcore (PBO, MCE and other motherboard boosting functions, Loadline.. etc)