Jump to content
Corsair Community

Can a Corsair Hydro H115i Pro properly cool a 9900k ??


MrZyanide

Recommended Posts

Hi everyone, I'm asking this because I'm having issues with my current setup.

 

My 9900k is overclocked to 5GHz all cores and when gaming it's temperature is in the 40s. However, if I run the CPU-Z stress test , which is not that intensive, the temperature immediately spikes to 64 and stays there. If I run Prime95 (smallest) , temperatures immediately go straight to the 80s. Coolant temperature slowly climbs up from mid 20s to 30s. If I stop any of the aforementioned tests then temperatures go back to the 20s . The VCore is 1.26V, room temp is in the low 20s.

 

These are the results after a thermal paste change, stock solution was doing worse. Also I've reseated the pump twice and I've already checked and reseated the back plate to make sure everything is ok. All these to no avail.

 

Is this the expected behavior? If not, should I look into doing something else?

 

Thanks for your help guys!

Link to comment
Share on other sites

That is normal behavior for any CPU and any cooler. The CPU temperature is determined by voltage less the amount of heat physically conducted away by the CPU materials, cold plate, and TIM. That heat is then transferred into the cooler where it can be dispersed elsewhere. The coolant temperature is the measure of how much heat is in the coolant system. Your coolant temperature represents the minimum possible CPU temperature with zero voltage. So as coolant temp rises, so does your CPU temp by the same amount regardless of load state. However, since coolant temp shifts typically range from 4-10C, the overwhelming majority of your CPU temp concerns are always going to be voltage based.

 

You have a maximum CPU temp to coolant differential of around 50C during the small FFT test. Most CPUs will have a 100% load CPU to coolant differential of 35-50C and this is very dependent on voltage and physical construction. You are running an 8 core CPU and near maximum overclock and using one of the most extreme actions to test it. If you are only 80C, you are doing OK. What exactly were you expecting?

Link to comment
Share on other sites

Thanks for the quick reply. I want to understand the following, as heat transfers from the cpu to the cooler and temperature increases, should it do so in a somewhat smooth curve or in a sharp step? I'm asking this because I'm witnessing those sharp steps as temperature climbs from 20 to 60 or 80,depending on the test, in less than a second.
Link to comment
Share on other sites

The CPU temp is going to be jumpy and all over the place during normal use. This was always true to some extent, but the last several Lake CPUs are really active. The way the power management works now, it essentially cuts power on and off. When on, full voltage big temp jump. Off, temps drop.

 

Now one big caveat to this is load type. Prime 95, CPU-Z linpack like loads are static. You should get a big 40-50C jump the moment you initiate it, then the max CPU temp holds steady. Coolant temp creeps up +1C every 60s or so until it caps out in the +4-8C range. As that happens, you will be able to see the end CPU temp also go up +1C every 60s or so to match the coolant rise. The load temp is even, the coolant temp slowly rises. Good illustration of how it works.

 

However, tests like OCCT, AIDA, XTU and others are primarily designed as stability tests and very dynamic, despite what you see as 100%. XTU is somewhat like a sine wave, but the others are really scraggly across a 10-15C range. That is normal for those specific tests and they were designed to mimic real use, highly variable loads.

 

That big jump when you first start the test I usually refer to as the "voltage on" temp jump. As mentioned, it has a particular value for a specific CPU model at 1.xx volts. So you can compare to other 9900K owners in accordance with voltage. And as you can see, not all 100% loads are created equal. The difference between CPU-Z and Prime 95 small FFT is expected. My CPU-Z will run about +30C over coolant temp with a delidded 8700K at 5.1/1.33v - 1 second after the test begins. However, Small FFT will be at +53C in an instant. *(Just re-tested at 5.1/1.33v and my coolant/CPU differential is +53C at those settings). The good news is most of us have no need for Prime 95 and it exceeds any CPU load type we every use. The fact that your game temps are good and your idle temps match the coolant means you got the contact right and frankly, only 80 something on small FFT is a really good temp for 5.0 on 9900K.

 

The cooler can only deal with the heat in the unit, or the coolant temp change. So if you are running a maximum of +8C (for example, 25C start, 33C max), then the most you could reduce CPU temps at any fan speed or with any cooler is 8C. Now, no cooler has a zero coolant temp delta without using some science tricks, but that is the general idea. For comparison, a H150i 360mm might get you +6C or a 2C savings. Not really worth it in terms of cooling. What is appealing is the larger your surface area, the less fan speed you need to get that cooling. If I give you a 2m long radiator 20x120mm fans, you will still hit 80C one second after initiating Prime 95 small FFT. That is all CPU, voltage, and material. The cooler plays no part. The difference would be you never rise from that point. The 2m radiator would be able to move out all of the heat dumped in and at low fan speeds. The coolant temp is the difference between one cooler or another.

Edited by c-attack
Corrected my small FFT temp data
Link to comment
Share on other sites

×
×
  • Create New...