Jump to content
Corsair Community

does this cooling seem correct? new to liquid builds


jonatron5
 Share

Recommended Posts

was going to overclock my Chip today when I ran into some strange issues I've never seen before(because I probably never looked), this is my first liquid cooled build and I'm suspicious that something may be incorrect here:

 

I have an I9-9900k with a corsair brand H100I platniumSE with everything powered through ICUE.

 

first my temp findings:

 

According to core temp, idle on my desktop my cores are ~30-35c

gaming in modernwarfare i usually hit 65-70c

 

I installed prime 95 and ran the small fft torture test, and the second I clicked the button my Core temps spiked to 90c

 

Icue monitors my coolant temperature which on idle is sitting at 30 c

 

but what really concerned me was that, after sitting with Prime 95 running for a full minute I did not register any increase in coolant temp.

 

I am suspicious that my liquid cooler is not cooling as it should be. I know the coolant is a liquid and liquids take a while to heat up. but below is an image I took after 1 minute of prime95 stress testing:

 

e8ea0f6f8c03ff786df1e736324324c7

 

 

 

as you can see the cpu is cooking at nearly 100c and my coolant isnt even elevated from baseline

 

EDIT apparently my url to the img got eaten, here is a new one

 

https://gyazo.com/e8ea0f6f8c03ff786df1e736324324c7

Link to comment
Share on other sites

It is pretty strange because your voltage is very high, but the clocks are running at their default setting, 4.7 GHz on all cores.

It is overvolted rather than overclocked apparently. Core temp reads 223W which is pretty high, so the cooler temps are not surprising, but at that kind of voltage you should be running at 5 or 5.1Ghz.

 

It looks like a classic case of MCE enabled in the BIOS. Can you tell what model/manufacturer your mobo is?

From the looks of it, the power limits are disabled in BIOS and the motherboard is pumping stupid voltages into the CPU making it overheat.

Link to comment
Share on other sites

With your fans and and pump going at near maximum rate, you are removing added heat into the coolant stream nearly as fast as you add it. The cooler's watts dissipated is almost never the limiting factor unless you have a large TDP CPU combined with a small 120mm panel cooler. The limiting factor for all of us is voltage. The heat must go through the CPU and be conducted out. There is always a point where the initial heat to the CPU from voltage level X is too high and the temperature is critical. You are using small FFT Prime to test the cooling system. What you are really doing is testing the physical properties of your CPU. It wouldn't matter if I gave you a 120mm radiator, a 560mm, or a wall sized array. All of them would read 90C+ for the first minute. Difference in cooler size/performance come over time and you can't run Prime small FFT long enough to demonstrate that. Frankly, I don't see any reason to run small FFT at all. Prime is useful for quick binning the physical properties of CPUs or sometimes very targeted, select range testing. For thermal performance of the system, you need to use something that doesn't push the CPU to the limit 3 seconds in. Try something simple like the stress test in CPU-Z (Bench tab). It is a linear, even load so it makes it very easy to see when coolant goes up and the direct +1/additive relationship between CPU temp and coolant temp.

 

That difference between coolant temp and CPU under 100% load is normal, but also a direct factor of the voltage level and again, the unique physical properties of that specific piece of silicon. Most CPUs in full Intel/AMD spec settings with no motherboard overrides will be about +30-35C. Everything else is overclocking and the range varies from 35-50C. At +60C you are on the edge of possibility. Since coolant temp is also the lowest possible CPU temperature, you would need to keep coolant at a max of 20C in order to keep that test at 80C. That's not realistically possible without external chillers or something along those lines. However, what you probably need to do right now is check the maximum actual reported Vcore when under load. I suspect your BIOS was pouring it on under small FFT in an effort to keep things stable. Most motherboards cannot run small FFT right out of the box without tinkering with the settings. If I were guessing, I would say you hit at least 1.40v+. You don't need that and you don't need to test this with Prime. This is another reason to try an alternate fixed load test like CPU-Z or some type of Linpack test.

 

What motherboard is this? The necessary changes vary by brand.

Link to comment
Share on other sites

 Share

×
×
  • Create New...