Processor overheating with Corsair Hydro Series H60

I have a H60 AIO and my processor is overheating.

Full System specs:

Motherboard: GIGABYTE GA-Z270X-UD3

Processor: Intel i7-7700k

Memory: 2x CMK16GX4M2A2133C13

Video Card: Gigabyte GeForce GTX 1080 G1 Gaming 8G

Case: Cooler Master: HAF 912

Power Supply: EVGA 650W 80 Plus Gold - 100-GD-0650-V

Nothing is overclocked, everything is factory default - the only thing I've done is go into the bios and set fans to maximum speed instead of SmartFan. My maximum vCore according to HWMonitor is 1260v when in use. Idle is 0.636v.

My Pump RPM is on a constant 4354-4440. The fan attached to it goes between 1254-2027 RPM.

While System is idle I can get temperatures on the processor around 37-50 C. However, if I game (WoW, Assasins Creed Odyssey) or if I try any sort of stress test (Prime95, Intel Processor Diagnostic Tool) my temps instantly reach 90 degrees and within less than a minute my processor is hitting HOT 100 degrees C warnings.

I've already removed the old thermal paste from the contact and applied a new thin layer, I've also tried to keep the radiator outside the Case as a way of testing airflow/intake. I've also cleaned the radiator out with running water, let it dry and there's no dust left in it. Nothing helped.

My Pump is connected to Sys_Fan1 but the RPM is constant.

I don't know what else to do.

When you load a game the CPU temps instantly hit 90C? Pump speed checks out and the voltage is reasonable. Best guess is the cooler has a bit of a flow restriction going on. Do you feel any temperature difference between the hoses (one hot, one cold)? Coolant differential on a H60 is going to be 1C at most, so that should be imperceptible. Any kind of clear difference in temperature suggests heat is not being released. You can also check the radiator end. The exhaust air should get warm. If it feels cool like normal room air, that also suggests heat is not making it to the radiator for dissipation.

I would also start a support ticket with Corsair. That will be a necessary step if you wish to replace it under the warranty terms.

When you load a game the CPU temps instantly hit 90C? Pump speed checks out and the voltage is reasonable. Best guess is the cooler has a bit of a flow restriction going on. Do you feel any temperature difference between the hoses (one hot, one cold)? Coolant differential on a H60 is going to be 1C at most, so that should be imperceptible. Any kind of clear difference in temperature suggests heat is not being released. You can also check the radiator end. The exhaust air should get warm. If it feels cool like normal room air, that also suggests heat is not making it to the radiator for dissipation.

 

I would also start a support ticket with Corsair. That will be a necessary step if you wish to replace it under the warranty terms.

Yeah, it's an instant 90 and it hits 100 under a minute.

The air coming out of the radiator is cool room temperature, though it does warm a bit after prolonged use (IE: more than 20-30 minutes), before that it stays cool even with the processor burning up.

I wanna say I can sorta feel a difference in temperature (after 5 mins or so of load) between the hoses but it might be just confirmation bias, it's not terribly different as in one is cold the other is burning up. But one feels somewhat warmer to the touch than the other.

When I close the game or stop the load the drop is fast, it drops to 60-70 immediately and back to 40-60 in a minute or less.

This doesn't really seem like a blockage or flow restriction. The +50C jump to 90C when you power on the stress test is the Vcore jump to maximum. This has nothing to do with the cooler flow, fans, or pump speed. It is determined by voltage, CPU design, and the conductivity of the metal. Of those three, only the voltage is under your control. If this is a substantial change overnight, it might suggest the bracket has come loose or the cold plate and CPU are not making good contact. You idle temps seem a bit warm, but it is so hard to tell on the Lake processors because they never sit still and constantly bounce of max vcore at idle.

The second, slower rise in temperature you see after initiating load is the coolant temp rise. You can't read that on the H75, but that is the coolant heating up. It should be relatively slow up and down. Obviously that is a much smaller piece of the temperature increase compared to the voltage based element, but that is how cooling works. +10C in 1 minute seems a little fast to me, but then I have not used a single pane 120mm radiator in a while.

1) Verify your actual load Vcore is 1.26v. On full auto settings, most Z170-370 motherboards will spit out an absurd amount of voltage. What is the peak CPU frequency is registers under load? Even if you don't want to overclock, a manual setting of frequency and voltage is usually preferable. Any BIOS update puts you back at square one for these settings.

2) Was the cooler fine one day and bad the next? Or a slow degradation in performance over time? The other to watch out for at this time of year (N. Hemisphere) is the advent of Summer and warm room temps. I have a +10C differential in this room between Winder and Summer. That is passed on to every component. So a peak CPU temp of 80C would become 90C under otherwise identical conditions. You may need to weigh any environmental changes into the analysis.

This doesn't really seem like a blockage or flow restriction. The +50C jump to 90C when you power on the stress test is the Vcore jump to maximum. This has nothing to do with the cooler flow, fans, or pump speed. It is determined by voltage, CPU design, and the conductivity of the metal. Of those three, only the voltage is under your control. If this is a substantial change overnight, it might suggest the bracket has come loose or the cold plate and CPU are not making good contact. You idle temps seem a bit warm, but it is so hard to tell on the Lake processors because they never sit still and constantly bounce of max vcore at idle.

 

The second, slower rise in temperature you see after initiating load is the coolant temp rise. You can't read that on the H75, but that is the coolant heating up. It should be relatively slow up and down. Obviously that is a much smaller piece of the temperature increase compared to the voltage based element, but that is how cooling works. +10C in 1 minute seems a little fast to me, but then I have not used a single pane 120mm radiator in a while.

 

1) Verify your actual load Vcore is 1.26v. On full auto settings, most Z170-370 motherboards will spit out an absurd amount of voltage. What is the peak CPU frequency is registers under load? Even if you don't want to overclock, a manual setting of frequency and voltage is usually preferable. Any BIOS update puts you back at square one for these settings.

 

2) Was the cooler fine one day and bad the next? Or a slow degradation in performance over time? The other to watch out for at this time of year (N. Hemisphere) is the advent of Summer and warm room temps. I have a +10C differential in this room between Winder and Summer. That is passed on to every component. So a peak CPU temp of 80C would become 90C under otherwise identical conditions. You may need to weigh any environmental changes into the analysis.

The thing is, I am not sure when It all started. I only measured temperature after I took my computer to a computer service company to have it cleaned from dust. (They did a terrible job, didn't even change the thermal paste on the CPU and didn't clean the cooler radiator) - so that's why I made the check. Before that, it has been somewhere between 1 and 1/2 years since I've done and temp checking. My bad for that I guess, but I can't pinpoint when it started.

So I changed the thermal paste and cleaned the radiator myself. As for the mount being loose, doesn't seem so. Maybe it is too tight? I usually tighten the screws with a screwdriver instead of by hand - should I do it only by hand?

I'm in the southern hemisphere and I am starting to see cooler temperatures, so room temp is probably not the issue.

HWMonitor says my max frequency is 4502MHz on all cores, which is consistent with 7700k stock, vCore voltage says 1.248v. I will check the bios settings and try to set it to manual instead of auto and report back.

Edit: So I set the voltage to a fixed 1.200v on the CPU and the temp hit 100C when booting up the computer (Chrome auto opened so that might be the cause). However, when "idle" (voltage at 1.200 but low load and low frequency) the temperature is not going below 57C. It's averaging 57-65. Any increase in frequency causes it to warm up to the 90 and 100.

Everything about this suggests a contact issue between the cold plate and cpu. Unfortunately, there is no obvious thing we can tell you to do. DO NOT over-tighten or try to clamp with maximum force. That is never the problem. More common is the thread from the stand-off does not go all the way into the backplate hole, the backplate is turned the wrong way, or the threads were unevenly tightened leading to non-uniform pressure. You’re going to need to take the block off and check the condition of the TIM. If it never fully spread or “melted” across the block, that is a dead give away. You’ll want clean it and start over regardless.

Hand-tightening should be enough. If you want a 1/4 turn with a screw driver, that’s ok but not until after you have hand tightened all of them down. Move in a cross pattern as often as you can, tightening one, then the next on the opposite side, each a little at a time.

Everything about this suggests a contact issue between the cold plate and cpu. Unfortunately, there is no obvious thing we can tell you to do. DO NOT over-tighten or try to clamp with maximum force. That is never the problem. More common is the thread from the stand-off does not go all the way into the backplate hole, the backplate is turned the wrong way, or the threads were unevenly tightened leading to non-uniform pressure. You’re going to need to take the block off and check the condition of the TIM. If it never fully spread or “melted” across the block, that is a dead give away. You’ll want clean it and start over regardless.

 

Hand-tightening should be enough. If you want a 1/4 turn with a screw driver, that’s ok but not until after you have hand tightened all of them down. Move in a cross pattern as often as you can, tightening one, then the next on the opposite side, each a little at a time.

Apparently, there was too little thermal paste (I'm always conscious of putting too much).

I re-did it all (removed, applied again) and now the highest temp I got while playing was 81ºC.

Hopefully, it stays that way and I'm glad it wasn't anything more serious - thank you very much for your time!