AIO liquid temperature consistently higher than GPU. But how?

[yoadknux]

Hey guys,

Wanna discuss something I've noticed on my GPU. The X41 is an "interactive" pump in a way that you can monitor it's pump speed and the liquid temperature using the CAM software. I have it mounted on an Aorus 1080ti. Long story short, the liquid temperature of the AIO is ALWAYS higher than that of the GPU, always +3c or so. I don't think it should be possible.

The X41 is cooled by 2x Noctua NF-A14 fans (500-1500 RPM), the PCB is cooled with a 92mm fan, and the GPU is more or less babysitted by 2x 200mm fans that blow air on it, one from the side and one from the front.

Just some numbers for reference, in idle: GPU 27c, X41 liquid 30c. FireStrike loop: GPU 53c, liquid 55c. Plague Tale 200% scale (~4k) 2 hour load - GPU 55c, X41 Liquid 58c. The GPU temperatures are overall great, especially considering we're at summer, the card is heavily overclocked, and the fans are more optimized towards noise (they reach 100% only at 60c GPU, which never occurs)
Why do those numbers seem OFF to me? because it seems that the GPU cools the COOLER... Some possible explanations:

a) CAM is lying, or the temperature of the liquid is not detected properly.
b) Maybe something related to the pump sitting on a GPU, so it sits horizontally (i.e. parallel to the ground). Those AIOs are mostly designed for CPUs, where the pump sits on the vertical position.
c) The heat dissipation of the backplate is so good, that combined with the airflow from the 2 large fans, it shares the load with the liquid cooler.

This is just a scientific/engineering discussion, I'm not trying to solve a problem because there isn't one. I can buy stuff and do experiments. What would you guys suggest? I was thinking maybe a vertical GPU mount, but I'm not sure it would fit my case (could clash with the side fan). Anyone have any experience on this?

Thanks!

 

[freakdiablo]

My guess? Either

A) The AIO is designed so flow hits the GPU first, then the pump, then the sensor, so the pump adds heat to the liquid on top of what the GPU puts out. The liquid is then cooled by the radiator before returning to the GPU.

B) The two temp sensors not being calibrated quite the same so a +/- 3c "drift" between the two is normal, especially at the temps you're seeing.

C) Combination of the two.

Are you seeing any possible heat related issues? Crashes, artifacts, etc.? Because if not, I'd simply say taking the average of the two should be close enough to true temp, in which case you just have good temps on an OC'd 1080ti.

 

[EarthDog]

Erroneous reading. It's just that simple.

An AIO pump isnt dumping enough heat in the system to raise temps in a loop. We also know from watercooling 101 that a properly radded and flowing loop the temp.differences between any point in a loop varies about 1-2C. Gpu/cpu temps are typically 10-30C more than water temp (delta) depending on several factors.

 

[yoadknux]

My guess? Either

A) The AIO is designed so flow hits the GPU first, then the pump, then the sensor, so the pump adds heat to the liquid on top of what the GPU puts out. The liquid is then cooled by the radiator before returning to the GPU.

B) The two temp sensors not being calibrated quite the same so a +/- 3c "drift" between the two is normal, especially at the temps you're seeing.

C) Combination of the two.

Are you seeing any possible heat related issues? Crashes, artifacts, etc.? Because if not, I'd simply say taking the average of the two should be close enough to true temp, in which case you just have good temps on an OC'd 1080ti.

You could be right. I may end up plugging this cooler to my CPU and see if it has similar behavior. But that's a lot of work

Erroneous reading. It's just that simple.

An AIO pump isnt dumping enough heat in the system to raise temps in a loop. We also know from watercooling 101 that a properly radded and flowing loop the temp.differences between any point in a loop varies about 1-2C. Gpu/cpu temps are typically 10-30C more than water temp (delta) depending on several factors.

Yes, if the loop is flowing properly, there won't be a temperature gradient. What if it's not flowing properly? I know the pump RPM is fine, but maybe the orientation of the pump does something? Hmm... Maybe I'll lay the case on the side and see what happens...