Jump to content
Corsair Community

Water temp hardly changes?


wzrd

Recommended Posts

VFpfFqN.jpg

 

Hi

I'm new to AIO and I don't fully know how the heat of CPU is transmitted while its stressed. The temperatures of idle and stressed scenarios are like below.

 

Idle

CPU: 50C

AIO: 20.6C (X)

 

Stressed

CPU:80C

AIO:21.7 (Y)

(Picture)

 

Water temp difference(X-Y)= 1.1C

 

The water temp hardly changes. Is it normal behavior of AIO? My impression is its not working properly but PC doesn't shut down so far, so it is at least cooled by AIO?

 

[EDIT]

Clarify the numbers.

Edited by wzrd
Link to comment
Share on other sites

Yes, it’s normal. As you are putting heat into the cooling system, the fans and radiator are moving it out. Coolant temperature change is the measure of conducted heat in vs dissipated heat out. When you put in more than you can remove, the temp goes up. Move out more heat than is added, coolant goes down. Room temp or more specifically case ambient temp is the minimum possible coolant temperature. Realistically, +2-3C over ambient is the bottom rung.

 

If you want to test the premise, set the fans to zero and run the stress test again. This time you can watch the coolant temp tick upwards. Regardless coolant temp changes involve a sustained wattage over time. It never jumps about like the CPU dancing on the pins.

Link to comment
Share on other sites

Yes, it’s normal. As you are putting heat into the cooling system, the fans and radiator are moving it out. Coolant temperature change is the measure of conducted heat in vs dissipated heat out. When you put in more than you can remove, the temp goes up. Move out more heat than is added, coolant goes down. Room temp or more specifically case ambient temp is the minimum possible coolant temperature. Realistically, +2-3C over ambient is the bottom rung.

 

If you want to test the premise, set the fans to zero and run the stress test again. This time you can watch the coolant temp tick upwards. Regardless coolant temp changes involve a sustained wattage over time. It never jumps about like the CPU dancing on the pins.

Thanks for the response.

So is it right that you are basically saying I didn't put enough heat into the AIO to change the temp of water? So the AIO temps didn't change even after the benchmark? The combination of the CPU/AIO is 3900X(TDP:105W) and H115i pro by the way.

 

 

Additionally, I noticed that the fan speed never increased above 1200 rpm while benchmark and CPU down clocks to around 2k Hz sooner than the bench starts. The temp were 95C. And the water temp is still 25C :sigh!:

Edited by wzrd
Link to comment
Share on other sites

Two separate issues. The cooler is simple enough. A 280mm can easily handle any CPU. Add in a 300W gpu and you’ll see plenty of coolant change.

 

The CPU temp is always going to be the limiting factor. Otherwise we could all smother our systems in radiators and run whatever clock speeds we wanted. The first part of the cooling process is conductive. Voltage at the pins creates the heat, through the CPU -> TIM -> Cold Plate, then into the coolant stream. You can’t change that process or the rate of it. Voltage + CPU design will always be the limiting factor.

 

Now that said, a 70C+ coolant to CPU temp difference is out of bounds unless you are deliberately running maximum possible Vcore. 1.17v in the picture above is not that high. I also think 50C idle cpu temp is high, if the voltage is really stepping down. I am a little concerned you might not have ideal physical contact between the cpu and the block. You may have to take it off and inspect the TIM to see if it has spread. Normal coolant rise for a 10 min stress test should be about +6C. It you don’t go +3C in 3 minutes, something may be wrong.

Link to comment
Share on other sites

Two separate issues. The cooler is simple enough. A 280mm can easily handle any CPU. Add in a 300W gpu and you’ll see plenty of coolant change.

 

The CPU temp is always going to be the limiting factor. Otherwise we could all smother our systems in radiators and run whatever clock speeds we wanted. The first part of the cooling process is conductive. Voltage at the pins creates the heat, through the CPU -> TIM -> Cold Plate, then into the coolant stream. You can’t change that process or the rate of it. Voltage + CPU design will always be the limiting factor.

 

Now that said, a 70C+ coolant to CPU temp difference is out of bounds unless you are deliberately running maximum possible Vcore. 1.17v in the picture above is not that high. I also think 50C idle cpu temp is high, if the voltage is really stepping down. I am a little concerned you might not have ideal physical contact between the cpu and the block. You may have to take it off and inspect the TIM to see if it has spread. Normal coolant rise for a 10 min stress test should be about +6C. It you don’t go +3C in 3 minutes, something may be wrong.

The idle CPU temp even jumps over 60C frequently. And by just watching youtube it goes to 74C.I found this info in tomshardware

 

https://forums.tomshardware.com/threads/ryzen-9-3900x-with-h115i-platinum.3521780/?view=date

Karadjgne said:

Dunno how you have it set, I believe default is cpu temp, my aio uses Cam and it's switchable to liquid/coolant temp.

 

With your ambient, open case, cpu should have idle temps barely over room temp, so Corsair might be right that there was an issue. But while an inconvenience, at least you got some relief that they are perfectly willing to do right by you and replace it. It's a loss for them in $ no matter how you spin it.

 

True. They said with 23-25C room temp your cpu should be low 30C idle and even light web browsing it should hover around max high 30s of low 40. When i said no, mine is 53C using youtube they said okay thats it we will send you a replacement.

 

I will put the new one with the thermal grizly and post all findings!

 

My room temp is lower than his 23-25C (mine is around 13-15C) but overall temp are higher than his H115i Platinum which turned out to be faulty. I checked thermal paste and mounting screw. I don't know what is causing underwhelming performance.

Link to comment
Share on other sites

This is all on the CPU end of things. Either you don't have good physical contact between CPU and block or your voltage and BIOS settings are out control. I don't think 1.17v shown above is too much Vcore, but you make sure that's where it stops. You can't really run synthetic stress tests on most motherboards right out of the box. You have to put limits on certain elements to keep it from piling on Vcore. I can't help you with the AMD settings since I don't have one.

 

What are you using the measure the CPU temps? There is a pretty short list of programs that properly read Ryzen 3000 series CPUs. Link is isn't one of them. Must be something else.

 

Assuming the CPU temp data is accurate, you are still looking at contact issue. +70C over coolant is very high. It's difficult to tell you what to do beyond the instructions. It's a physical fit thing.

Link to comment
Share on other sites

What are you using the measure the CPU temps? There is a pretty short list of programs that properly read Ryzen 3000 series CPUs. Link is isn't one of them. Must be something else.

I'm looking at Core Temp and Corsair Link but the data I posted here are all Corsair Link status. In terms of Core Temp measurement, it reports data more frequently and shows higher temp at highest (+ 5C).

 

Assuming the CPU temp data is accurate, you are still looking at contact issue. +70C over coolant is very high. It's difficult to tell you what to do beyond the instructions. It's a physical fit thing.

Yes, it's strange. I checked if it is mounted properly and water block thing contacts CPU surface.

Link to comment
Share on other sites

*some data addition

 

5 mins cpuz stress test.

  • Room Temp:13
  • CPU: 88C
  • AIO Mode: Performance

 

So theoretically, winter will be eventually over and the room temp should be +10C and 280AIO won't be able to handle the default CPU without severe thermal throttling.

 

It shouldn't be normal. It could be contact issue as c-attack said.

Link to comment
Share on other sites

Make sure you double check the CPU temps with Ryzen master or something else. The Link was program was replaced by iCUE almost 2 years ago now. It certainly was not updated for Ryzen 3000 and frankly Link was never good at reporting motherboard data anyway. Other than Corsair specific temps (coolant temp, Corsair RAM temps, Corsair PSU temp), I would verify any other data from Link for your motherboard.
Link to comment
Share on other sites

  • 2 weeks later...

UPDATES:

I'm not 100% sure what was cause, but I can add more information.

 

1) I ordered a sheet of thermal pad. I put this pad instead of grease and tightened the screw as same way as before. The difference of temperature was in the range of error. 2) I unmounted the aio and tightened the screw a little, where I can move the water block if I apply force from each corner. The Temperature went down delta 9C.

 

Additionally, before I employ this method, I found the the thermal paste only dissolved specific part. I'll add more finding if I get. *I took out radiator from case so no force from hoses are applying to water block while checking.

 

UPDATE2:

I can see the diminish of the abruptness of full load temperature spike.

Edited by wzrd
Link to comment
Share on other sites

×
×
  • Create New...