Jump to content
Corsair Community

H115i with Ryzen 5900x weird temperatures


Recommended Posts

Hi all

I just replaced 3900x with 5900x and I have noticed CPU temp idle around 60, AiO is showing 44 coolant temp. After starting benchmark CPU is immediately increasing to 82-86 and coolant is around 46 ...

Normal or not really ?

Case cooled with 2xLL140 top, and 1xLL120 back

All with curve according to coolant temperature.

Link to comment
Share on other sites

The high idle temp could be from a number of BIOS settings or 5000 series behaviors that we are just learning about.

 

However, 44C coolant pre-load is not and wattage should be comparable to the 3900x. I would expect load coolant rise to be about the same. The problem is if you are starting at 44C then the CPU temp/coolant differential is going to be +35-50C at full Vcore. So the 85C is about right, there isn’t a contact problem, but that coolant is way too high.

 

Can you describe the case layout in regards to the radiator? Location, fan direction (intake vs exhaust)? Are there any special circumstances with dust filters, solid panels on the radiator side, or environmental location issues where heat can’t escape the area (wedged under a desk, etc.)?

Link to comment
Share on other sites

The high idle temp could be from a number of BIOS settings or 5000 series behaviors that we are just learning about.

 

However, 44C coolant pre-load is not and wattage should be comparable to the 3900x. I would expect load coolant rise to be about the same. The problem is if you are starting at 44C then the CPU temp/coolant differential is going to be +35-50C at full Vcore. So the 85C is about right, there isn’t a contact problem, but that coolant is way too high.

 

Can you describe the case layout in regards to the radiator? Location, fan direction (intake vs exhaust)? Are there any special circumstances with dust filters, solid panels on the radiator side, or environmental location issues where heat can’t escape the area (wedged under a desk, etc.)?

Nothing special about case, just attached photos. Rear is pushing out, top one are pushing out. Then I have two pushing in through radiator.

Nothing special in BIOS either. I did default first.

I tried default curves, and also some customs but tried to make them aggressive, ramping up every degree.

IMG_20201116_143309.thumb.jpg.c341794101ffc80eb83146532f35654c.jpg

IMG_20201116_143258.thumb.jpg.c20868b8ce53723c51983129947843ec.jpg

Edited by mucflyer
Link to comment
Share on other sites

Is this 40-something H115i temp after some long gaming? Or almost immediately on cold boot? GPU waste heat in a small case could explain elevated coolant temps that linger after gaming. The entire interior of the case would be that same temp and the MB, drive temps, etc. would all be similar. On the other hand, if you boot up on a cold morning and the H115i Temp starts at 22C and then builds up to 40C+ all on it's own, there is probably a flow problem issue with the cooler.

 

If you have a "H115i Pro" and not a "H115i". The fan speeds are mostly irrelevant for the larger change in coolant temp. I could turn my fans off and not make it to 44C. However, the pump speed is relevant on that model. The Pro series has an ultra quiet "1100 rpm" pump speed mode for desktop work. Depending on CPU type and power, you can easily tack on an extra 8C+ if running load in that state and if you were back at 36C coolant, we wouldn't be talking about whether the cooler is working or not.

 

*If this is load situations only, the other thing to try would be opening the front panel and see what change that has on things. Solid front doors block noise and everything else -- like airflow. Was this typical coolant temp range before on the 3900X?

Edited by c-attack
Corrected incorrect model attribution
Link to comment
Share on other sites

If that is a Platinum, then the low pump speed problem is out. You won't see differences between the ~1950 Quiet and 2300-2400 Balanced or likely even the Extreme setting.

 

Unfortunately this cannot be attributed to the CPU. While a CPU may be "hot" based on it's own characteristics and voltage, that can be assessed in the difference between coolant temp baseline and CPU end temp at 100% load. You were reporting a +40C differential, which is actually quite good for any modern CPU with no voltage tweaking. However, it also means that your peak temps can reach +40C over your baseline at any time and that 40 something coolant temp is the lowest possible CPU temp you could ever see with 0v.

 

Most people will see an idle coolant temperature of +4-7C above their room temperature with variation based on CPU type, power program, C-states, etc., and then case/radiator layout. So if your room temp is 21C, then you would expect something like 25-27C coolant at the desktop level. Regardless of room temp, the interior case temperature is going to be the minimum possible, so if you have a sealed box or throw the case into a shut cabinet, the inevitable follows. However, what this means to you is that +40C differential is being applied to a 40 something baseline, instead of 20 something. If you ran some stress test at 25C, I would expect to see ~65C peak temps back. If you run it at 45C coolant, you will get 85C back. Whatever is causing the coolant temperature to ride this high is costing you 20C in CPU temp at all times.

 

Since the coolant temp only seems to go up 4-5C on load (exactly as expected for that wattage CPU), that suggests the cooler is functioning properly. That is not conclusive, but 99% of the time someone in your situation is looking for a case or environmental problem. My guess is the front door is quite restrictive or somehow a bad combination with the fan blade geometry and position. If you have a dust filter on there as well, that will add to the problems.

 

There are a few things you can do to assess this are: 1) run with the front panel off and see how things change; 2) After the next time you power down for a while, immediately go to the CUE cooler tab when you power back up and look at the H115i Platinum temp. It should be right about room temp on boot. Does it immediately start ticking up +1C every 30 seconds until it gets to 40C? Is this the same with the door off. 3) How does the coolant temp compare to something like your main motherboard temp sensor? If central case temp is 45C, then the radiator is part of that zone too and you can't drop below that.

Edited by c-attack
Link to comment
Share on other sites

Seems you are absolutely right. Because I have radiator fans behind panel, due to space, they don't have enough airflow. When I removed front panel, all temperatures dropped to normal level. I may need to rethink whole design. Unfortunatelly there is no space inside to fit RTX and fan... And radiator won't fit at the top... I may need to check if I can mount one fan inside, other outside of the radiator, not sure however if that won't affect efficiency even more.
Link to comment
Share on other sites

There's still the option to install slim fans on the inside, they are typically 15mm thick. That could work for the bottom fan position. Wouldnt be ideal, but probably wouldn't be worse.

 

But if you can squeeze the top fan on the inside, it would already drop temps quite a bit.

Link to comment
Share on other sites

I am sure you can't mount the radiator up top and this is a common issue to encounter. Noise suppression is energy suppression and what effects one effects the other. It can't be noise dampening and free air flowing.

 

A couple of options:

 

1) Pop off the front panel before gaming or extended load conditions. Not ideal, but solves the issue as much as possible.

 

2) Move the fans from sandwiched between the front panel and radiator to the interior side of the radiator. This limits the fan blade interaction to between the radiator and blade, not blade/radiator + panel. It is usually less noisy because of that, but also changes the angle of air pull in the panel chamber. This won't make it act like the panel isn't there, but it should knock off a few degrees and be less noisy. This would preclude using any kind of "half push-pull set-up", but I am not sure that is a winner anyway. Push-pull might net you back another 1-2C on a free standing radiator, not the 15C we are looking for here.

 

3) The ever obvious... start thinking about whether this is annoying enough to look at a new case.

 

4) As much as I am a proponent of water cooling, it's possible you would get better results with a large air tower compared to the current state. It would not be restricted by the front panel in the same way. It would take a heat penalty when the GPU is running hard. I would want to see some data from anyone running a 5000 anything on air in a gaming situation before committing to this. Large air towers can be a serious pain in the neck to install and while it would be better at idle, it is the gaming load temps that you want to bring down.

Link to comment
Share on other sites

  • 1 month later...

Hello,

I recently mounted a whole new build with a ryzen 5900x.

As a cooling I have a Capellix h100 240mm.

I have a bit of trouble keeping this CPU at bay.

In stress test in cinebenchr23 10 minutes in multicore temperature ok at 69/70 °.

 

The problem is when I go into gaming. especially farcry5 and battlefield5 brings me the cpu to 78-79 ° and the cooling temperature of about 39-40 °.

 

I wonder how this is possible. but I have read around temperatures of over 80 ° with this cpu.

 

The Capellix is ​​top mounted on a 4000x corsair with 3 front sp120 fans, 1 rear sp120 and 2 on the ml120 radiator.

I tried to play with the fan curves on icue, but it changes little, a few degrees but the noise is a lot. Now I have a silent profile on the radiator and balanced on the other case fans.

I thought if I went to mount the radiator in the front I would have remedied the temperatures but from what I see it changes little.

P.S. I tried to use both mx-4, noctua h2 and grizzly as thermal paste. maybe it changes + or- 1 °.

Link to comment
Share on other sites

  • 2 weeks later...

Similar problem here. New to corsair, my new build 5900X with H150i RGB Pro XT shows very low coolant temperatures compared to 5900X very high temperatures, see screenshots.

 

CPU Temp: idle 45° to 85° stress (sometimes even 90°)

150i Temp: idle 27° to 33° stress

 

Since cooler temp does not move at all, setting any curve makes little sense. And 5900X temperatures readings are not supported in iCUE.

 

Could this be a bug of the coolant temp reading? How can I set fan to increase when CPU gets hot?

 

 

PS: Also when setting new profiles in iCUE for the chassis fans, I have a strange behavior. Setting them to quiet / balanced they spin at full speed even in idle. Only custom curves help.

Idle.thumb.png.bcf4af3b8a2ed9cb7885058c01cd4826.png

Stress_Quiet.thumb.png.370738a989caabc818cda11b285ce79f.png

Stress_Extreme.thumb.png.fdb501b6e3a61c0fda58095347485cb4.png

Link to comment
Share on other sites

No, the fans are working correctly. They control the coolant temperature. They cannot prevent the CPU from heating up as voltage is applied from the socket pins. The actual CPU cooling is all conductive. The coolant system is waste heat transport. Think of it as a conveyor belt taking the heat out. As long the as the belt keeps moving, heat is being dumped out of the system. Stop the belt (coolant flow) or fans and things will back up quickly. Small changes in fan speed do not change your CPU temperature. The relationship is +1C coolant temp = +1C CPU temp. Same thing for cooling. So with a coolant rise of +6C, the most you could possibly reduce CPU temps by at any speed is theoretically 6C. You will never be able to get back to 0 rise and +2-3C is about as low as possible in any system.

 

When you create a new custom profile, make sure you set the sensor choice to H115i Temp. It likely defaults to CPU temperature and you have no data for that. Even if you did, the result would be the same. It's the wrong control variable for the job. Perfectly fine to use your own curve and I recommend most people do that. There are three shape tools in the upper right hand corner that correspond to the three presets. You can use them as a starting point, then drag the baseline to match your normal idle spot and soften the top end if needed. I can't see too many circumstances where you would need more than 1000 rpm for variable load use like most games and programs.

Link to comment
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
×
×
  • Create New...