H100i GTX and overheating 5820k

Right now i'm almost pulling my hair out trying to figure out what is causing my 5820k & H100i GTX combo to overheat.

For a long time (years) I ran my 5820k at 4.5Ghz at 1.375v, a fairly aggressive overclock with a decent amount of voltage, yet despite that the H100i GTX (using 4 high-CFM fans in push-pull) was able to keep temps under control during benchmarks and gaming. During the last year or so things have begun to get worse. Now when I run a benchmark that uses all 6 cores, temps spike up to 90C almost instantly and I will generally get a "WHEA UNCORRECTABLE ERROR" blue screen at that point. This also happens often (though not instantly) in games such as Farcry 5 that are well multi-threaded. Most recently, and as ambient temps start to heat up, I am even getting it sometimes when I first boot the computer. I had quite a few programs such as discord and temp monitor apps starting automatically, enough to peg the CPU at 100% temporarily. I had to disable some of those from starting automatically just so I can boot without getting a BSOD.

Right now i'm trying to figure out if my H100i GTX is still cooling properly, or if it needs to be replaced.

Diagnostic steps I've taken and observations I've made at this point:

I've backed my OC down all the way from 4500Mhz (36x125) to 4250Mhz (34x125), and the voltage down from 1.375v to 1.250v, yet it still overheats as described.

I've re-mounted my block a half-dozen times or more, using thermal-paste (Noctua NT-H1) configurations ranging from "grain of rice" to pre-smearing a thin layer across the entire heat-spreader first. In each case it made almost no difference.

When I feel the water tubes/hoses on my H100i GTX, one is usually warm while the other is usually cold.

I have my 4 high-CFM fans set to go full-speed when CPU temp (as reported by BIOS) hits 50C so airflow through the radiator should not be an issue. All fans are directly controlled by the BIOS.

I have the pump set to performance mode 24/7, and corsair link always shows ~2900-3090 rpm which seems correct. Pump receives full power at all times (not attached to a variable fan header).

With an ambient air temp of 75F/24C, I am getting water temps usually 32-34C at idle and 40-43C after an hour of gaming. When I say "Hour of gaming", I mean World of Warcraft, since that is really the only game I can still play for that long at this point. It only uses about 1.5 cores so it doesn't overheat my CPU and give me blue screens; it only gives me CPU load temps of about 70C (still too hot for WoW IMO). I seem to recall a much smaller variation in water temps before, mainly between about 36C-40C, even with loads much greater than World of Warcraft. Not sure exactly what that indicates.

I know the H100i GTX is an older model at this point and mine already has a few years on it. If I have to buy a new cooler, so be it. I'm just trying to avoid spending money on a new cooler only to find out the hard way that it wasn't the problem and the H100i GTX was working fine all along...

Sorry for the long post, and Thank You to anyone who is still reading at this point! :biggrin:

One hot/warm tube and one cool is usually a clear warning. The most the radiator can shed in one pass is 1-2C, right on the edge of perception. They should be indistinguishable in temperature feel.

There seems to be a lot going on, but let’s troubleshoot the cooler first since it seems like there may be an issue. Run Intel XTU or the stress test or in CPU-Z for 5-10 min. Both of these are mild. Note the starting coolant temp. Expected rise at your current clocks is about 6C. If you see the coolant temp rapidly shoot up, stop and report back. Also keep an eye on the cool down. A working system should shed half the heat in 1-2 minutes, then slowly trickle down over several minutes more. If you stop the load and the coolant just hangs there, that also suggests some kind of flow problem.

I ran the CPU-Z stress test. Using anything in the 4-12 thread range would cause the CPU temp to jump to over 80C in less than 15 seconds and the next lower setting is 2 threads so that is what I used in order to be able to run the test for at least 5 minutes. This still brought my CPU temps up to the mid-70's. Still, that represents only 1/3rd of my cores being used, so these numbers should be taken in that context.

Ambient air temp: 75F/24C

Starting water temp: 33C

CPU-Z 2-thread Stress-test:

Water temp after 1 minute: 37.0C

Water temp after 5 minutes: 40.1C

After test had been stopped:

Water temp after 1 minute: 37.8C

Water temp after 2 minutes: 36.1C

Water temp after 5 minutes: 34.5C

Also, based on what you said I went ahead and got my IR temperature gun out and used it to measure the temperature of the hoses, figured this would provide more useful info than "this one feels warm/cold".

Idle temps:

Water Temp: 33C (as reported via corsair link)

"Cold Hose": 28C (measured via IR temp gun)

"Warm Hose": 31.5C (measured via IR temp gun)

CPU-Z 2-thread Stress-test after 5 minutes:

Water Temp: 40C (as reported via corsair link)

"Cold Hose": 28C (measured via IR temp gun)

"Warm Hose": 37C (measured via IR temp gun)

Wow this is actually the first time i've ever used the IR temp gun for this purpose, i'm guessing that the "cold hose" temp being exactly the same (28C) idle vs load is not a good thing :(:

The CPU temp and conductive heat is a separate thing, but I am afraid the hoses are pretty damning. I ran a 5820 and 5930K @4.5Ghz for quite a while and am pretty familiar with what you should get back. 2 threads heating up the coolant +7C in 5 minutes is not good. I am a little surprised it cooled off normally (or almost), but the tube stuff is strong evidence and unfortunately this specific model is prone to this condition.

If you contact Corsair through the Support Ticket System above you can set up the RMA. You can ask for an Advance RMA where they ship you the new one on credit card hold, then you return the defective one within 30 days or whatever. You will get the H100i v2 that was the re-named version of the GTX. However, if you've been thinking you like a little more cooling with a little less fan speed, this would be the time to move up a size. Even if you like the 240mm and where you are, there is something to be said to moving to a newer cooler like the H100i Pro or something else. We could be having this same conversation one year from now.