What you can do if the temperature of your water cooling loop is too hot
3.10.2019
Translation: machine translated
After all your efforts to water cool your PC, you're disillusioned: the temperatures aren't much better than with air cooling. What could be the reason for this and what can you do?
If there's one thing you don't want to read after assembling a new PC, it's the following user comment:
The GPU seems way too warm for a water cooler. With my Heatkiller block, my overclocked 1080Ti never goes over 10°C above water temperature even under maximum load. Either there is something wrong with the flow (see flow-indicator) or the cooler block is not properly seated.
User Anonymous refers to my build for the Hydro X Series from Corsair.
Do I actually have to take the whole loop apart again and start from scratch?
Don't panic
If you are also struggling with high temperatures, stay cool. As long as they don't rise above 80° Celsius, you have nothing to worry about: Your components won't suffer. But 80° is very high and something is definitely wrong. In my case, it was up to 66° Celsius for the GPU.
Before you remove your GPU again in such a case, remove the heat sink, slap new thermal paste on it and reinstall the whole thing, you should check your settings.
As I mentioned in the review, I left everything at the default settings for the tests in the bios. At MSI, this is called Smart Fan Control. If this option is activated for the pump, it only turns faster at certain temperatures. However, leaving this setting as it is is not a good idea. The pump should always run at maximum speed. With water cooling, the best way to regulate the temperature is via the fan speed.
Pump: maximum
I want to know what influence the pump speed has on the cooling performance. Firstly, I deactivate Smart Fan Control for the pump and set it to maximum. For the Thermaltake fans, I leave Smart Fan Control activated with the default settings and set the Corsair fans to Balanced. Compared to the review, I only change the pump setting. This time, I first play "Rise of the Tomb Raider" for 20 minutes and then run Heavy Load and Fur Mark for half an hour.
By the way, I play "Rise of the Tomb Raider" in 1440p with the best possible settings so that the graphics card is utilised as much as possible. Core Temp records the temperatures of the CPU and GPU-Z those of the graphics card. The CPU reached a maximum temperature of 42° during the 20 minutes of gaming, the GPU a maximum of 48° Celsius. The coolant is around 32° warm.
For comparison: The CPU was on average 41° Celsius warm during the review, but the GPU was 66°. The temperature of the CPU has not changed significantly, but that of the GPU is 18° Celsius lower with the pump.
Next, I run Heavy Load with the "Utilise processor" process and the GPU stress test from Fur Mark at 1440p. During the 30 minutes, the CPU reaches a maximum temperature of 50° Celsius and the GPU a maximum of 49°. The temperature of the coolant is around 34°.
In this case, the difference is not so great. In the review, the CPU reached a maximum temperature of 51° and the GPU 57° Celsius. But at least it's eight degrees less. The fact that the GPU didn't get quite as warm during the first test was due to the fact that I only ran Fur Mark at 1080p at the time. This meant that the GPU wasn't utilised as much as it is now at 1440p. The eight degrees less is therefore a good value, as the card also had more to work with.
Fan and pump: maximum
I really like the new results. User Anonymous wrote in his comment that the temperatures of his GPU are rarely more than 10° Celsius above the water temperature. For me, it's still 15° and 16° respectively. Now I want to know what influence the fans have.
To do this, I set the Thermaltake fans to maximum in the bios and the Corsair fans to performance in iCue. First up is another 20 minutes of "Rise of the Tomb Raider". Here the CPU reaches a maximum temperature of 39° and the GPU 45°. The temperature of the coolant is again around 32°. When tested with Fur Mark and Heavy Load, the GPU reaches 46° and the CPU 47°. The coolant rises to around 34°.
With the fans set to maximum, it's three degrees less for the GPU and two to three degrees less for the CPU. The difference with the fans on maximum is no longer as striking as with the pump on maximum. Nevertheless, this brings me very close to the results of user Anonymous: the temperature difference between the coolant and the GPU is only 12° to 13°.
What else could it be?
In all honesty, I'm happy with this result. I have enough headroom - in the figurative sense - to overclock the CPU and GPU. The fact that I didn't quite reach Anonymous' temperatures may have the following reasons:
- There is still air in my loop
- The fans at the bottom cannot draw enough air through the radiator because the case is not high enough above the floor
- The fans at the bottom would probably be more efficient in a push configuration
Although I have vented my loop to the best of my knowledge and belief, it is still possible that air bubbles have formed somewhere. These can affect the cooling performance.
I have already mentioned in the review that the placement of the 280 radiator at the bottom of the case is not optimal. The feet of the case are not high enough for the fans to draw in enough air. The fact that I then opted for pull instead of push for the fans is just as unfavourable. Some time ago, I compared push with pull and push/pull. I found out that push is better than pull in most cases. For my build, however, I deliberately favoured form over function. I really wanted to illuminate the reservoir indirectly from below with the RGB of the fans.
In the test, I achieved three degrees cooler temperatures with a push configuration than with a pull. If the airflow of the lower fans were even better, it would certainly be another two or three degrees lower. I would then be below ten degrees Celsius difference to the coolant temperature and could compete with Anonymous' results.
With brains to better results
If the temperatures of your custom water cooling loop don't meet your expectations, you don't have to remove everything straight away. With just a few settings, such as in my case setting the pump to maximum and regulating the fans, you can achieve much better temperatures. Only if you're still not satisfied should you consider taking your loop apart.
Why is user Anonymous still achieving better results than me? It could be because of his setup. Where are his radiators placed? Which ones and how many does he have? Does he use push, pull or push/pull? So many factors influence a water cooling loop that a comparison only makes sense with exactly the same configuration.
Kevin Hofer
Senior Editor
kevin.hofer@digitecgalaxus.chFrom big data to big brother, Cyborgs to Sci-Fi. All aspects of technology and society fascinate me.
