Jump to content
Corsair Community

H115i RGB Pro XT - high CPU temps during stress tests


Recommended Posts

I'm a new owner of the above AIO from Corsair and have never used any form of water cooling in my previous 25 years of PC building. My system is as shown in my specs. I have the H115i mounted vertically in the Meshify C configured as intake, with the coolant lines entering at the bottom of the radiator. I have 2x140mm exhaust fans up top, a 120 exhaust fan in the rear, and another 120 intake in the bottom of the case.

My concern is that I have some issue with the H115i, be it my installation or the hardware itself.

What I've tried so far:

I've started with the thermal paste as supplied on the cold plate (got to thermal throttling, ~100C), and have since removed it, cleaned it, and re-installed with some Noctua material. Initially saw temps maxing in the upper 60s C under P95, then a step up after 5 minutes or so with temps hitting upper 80s. At this point I suspected a air bubble had migrated to the pump that may have caused this step up in CPU temp, so I shutdown and spent a few minutes rotating the case to provide impetus for the suspected bubble to migrate out of the pump and up into the radiator. I then left the case sitting vertically with the radiator at the top for several hours.

Based on feedback from others with similar builds, I should be expecting mid 60s C during stress testing. I seen 70C if using the CPU-Z test, but more of a concern, I see 90C with either OCCT or Prime 95. I'm using the monitor results from HWINFO64, OCCT, and iCUE. All three provide the same results.

I've observed the CPU voltages to be ~1.3V during any of the tests. Coolant temperature never exceeds 28C, front fans typically max out at ~1700RPM. I'm using extreme cooling profile in iCUE.

I'm tempted to pull the coldplate and re-apply the thermal paste one more time, as I know too much paste isn't a good thing, as well as too little.

I can't move the radiator to the top of the case as the depth of the case is too shallow and would cause the radiator fans to interfere with the top components and connectors on the motherboard.

Do I have too high of expectations from this AIO? Are these results seen by other users?

Any other troubleshooting to do? Your inputs and suggestions are appreciated.

Link to comment
Share on other sites

The liquid, radiator, and fans part of the cooler are in the waste heat removal department.  They don't have a direct impact on instant CPU temperature other than coolant temp is the baseline or minimum possible CPU temp with zero voltage.  +1 to coolant temp is +1C to CPU temp and the same for cooling.


In the short term, your CPU temp is going to be a product of voltage, materials, and conductivity.  I don't see anything wrong with your load results.  In my custom loop with a 10900K at 5.2x10, my CPU core temps during the CPU-Z bench will range between 70-75C with a load voltage of 1.31v.  Actual BIOS set Vcore is 1.44v with LLC 4.  Running the more rigorous OCCT or Prime tests will bring higher numbers.  Furthermore, its a cold morning and my coolant is only at 21C, so the CPU-Z test is about +50C coolant to core temp differential.  That is about where most people will end up and it will effective be the limit for normal use.


15 hours ago, mikehd said:

Based on feedback from others with similar builds, I should be expecting mid 60s C during stress testing.

Most people are far too casual in the information they provide and do so without context.  At 1.31v load voltage, you will never see those temps.  So the first question is if you need that voltage.  That is what I run for 5.2x10 and I do not have a good piece of silicon.  I suspect you are on Auto voltage and that is why you are seeing extreme differences between the rather mild CPU-Z test and those with AVX2 instructions.  However, I think you can stop repasting the block and this is going to be about getting your settings dialed in for frequency vs voltage.

Edited by c-attack
  • Thanks 1
Link to comment
Share on other sites

@c-attack, thank you for the background and info. Yes, I am running on auto mode. However, one additional detail is that this CPU and motherboard were both purchased used from someone who used a H115i RGB Pro XT as well. He claims to not have overclocked it, and I've been running it as received with my own H115i and Fractal Meshify case. I have verified that it is in auto mode however. I'll check with him again to see which particular tests he used and the test methods as well.

Link to comment
Share on other sites

The motherboard is going to be the control mechanism for a lot of this.  It's not quite as bad as it was a few series back, but your standard mid to upper class board will overclock automatically and/or extend power limits.  So if Person A was using it on Brand A with a H115i, they might see different end CPU temps when moving to MB Brand B because of different voltage loading.  Unfortunately, I don't know the Gigabyte BIOS like I do Asus, so I can't pinpoint the exact settings you should be aware of.  Load Line Calibration is certainly one along with the Vcore.  The boards are so much more complex now with shifting values based on CPU integrity.  Maybe a Gigabyte OC guide for your MB series is the place to look.  


I do feel your ~40C coolant to CPU core temp load on CPU-Z is appropriate for 1.31v load volts.  As mentioned, I am the same but with a 7C lower baseline and triple 360s, plus a higher flow rate.  Perhaps the critical difference between your tests is the load voltage when under OCCT or Prime.  My OCCT is not too much higher.  Prime can be lower or higher depending on test.  If you set it on Smallest FFT, I am not surprised it went to 100C.  Most people can't run that without BIOS tweaking.  

  • Thanks 1
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...