H100i 4790K temps!!!!

Hello,

I recently bought the H100i V2 AIO to replace my hyper 212 Evo so i can have more overclocking headroom.

I installed it last week and with no overclocks, I see temps rising to 82-84 after 5 minutes of Prime 95 blend test!!. This seems ridicolous as my hyper 212 could do 4.4ghz overclock and stay under 70c. I have reseated the backplate into all 4 possible directions and each one in the same and gives me those high temps.

I am using a h440 case with 3 front intakes and 1 back exhaust. Radiator is mounted at the top using a push exhaust configuration. I have also tried a pull but it didn't work. I have plenty of air flow and my case is dust free and clean.

CPU im using is 4790K. I have the fan header connected to CPU_header in mobo and I have disabled smart fan control in the bios.

Here is a picture of 5 minutes of Prime 95

https://gyazo.com/c3f0d1f59dcf32480e4493754acd28cb

At this point, I'm thinking this is a defective cooler but I'm not sure so give me your feedback. All the 4 times I remounted it, I used arctic silver 5 and used different amount of thermal paste each time. two of them being the line and rice method.

Currently I have the blackplate like the first picture here : http://www.hardwarecanucks.com/forum/hardware-canucks-reviews/69218-corsair-hydro-h100i-gtx-h80i-gt-review-4.html

Keep in mind this is STOCK settings so according to other people, I should be under 60c on full load.

Thanks

Your coolant temperature (h100i v2 Temp) looks a bit high, but this is relative to case ambient temperature and you need a basis value to put this into context. I would normally expect a +6-9C rise in coolant temp in about 10 minutes of any stress test.

1) What is the H100i v2 temp when you are at idle and not after any recent stress tests?

2) After you ran Prime95, did that 47C coolant temp starting dropping as soon as the load was cut? Or did it linger in the 40's?

3) Your drive temps seem a bit warm suggesting the case is as well. However, Link throws up some weird numbers on occasion. What does HWMonitor show for the drives and motherboard temp? How does this compare to the room temperature?

You will never run under 60C on a 4790K at stock settings and voltage with any cooler while running Prime95 without delidding. That is the physical limitation for nearly all CPUs.

The idle temps of the cpu on a cold boot hover from 32-40c. They jump around way too much so I can not tell you a exact average. Probably around 37ish though.

My ambient temp when running the test was about 22-23c.

HW Monitor shows 43-44c for the HDD and 37-38c for the SSD. I actually had my 3TB seagate die last week so I'm not sure if it was because of high temperatures or if this is all related.

After I turn Prime 95 off, the cpu temperature drops to 40 then degrades to 36-37 and the h100i water temprature stays in the 40s and doesn't drop too much. maybe a few degrees at most.

Also, since you said 60c or under isn't possible with any aio under prime 95, what temps should I be expecting at stock. Because I was planning to push this CPU to 4.6ghz but still haven't.

Also for some weird reason, sometimes when I'm watching temperatures, it would hit 70c for half a minute then go back to 60c and stay for 10 mins or so then it would go back up to 75 and repeat the process. This doesn't happen all the time but seems like really big fluctuation in the temps so is this normal?

Thank you for your reply! Appreciate it

After I turn Prime 95 off, the cpu temperature drops to 40 then degrades to 36-37 and the h100i water temprature stays in the 40s and doesn't drop too much. maybe a few degrees at most.

This part is concerning, unless the temperature inside the case is also in the low 40's. Then it is absolutely normal. As such, a lot of this hinges on that measurement. In order to take other case heat out of the equation, the next time you don't need your PC, put it to sleep or shutdown for at least 30 minutes, then boot up. Take a H100i v2 temp reading straight away. In 10 minutes at idle, take another. I'll continue to assume normal room temps 20-23C unless you tell me otherwise.

Also, since you said 60c or under isn't possible with any aio under prime 95, what temps should I be expecting at stock. Because I was planning to push this CPU to 4.6ghz but still haven't.

No, overclocking the CPU is not an issue. My statement was contingent upon a few keys things in yours. 1) Prime95 and 2) stock voltages. Prime 95 is already pretty brutal on Haswell and later, but with the voltage on Auto it will suck you dry throwing voltage at the test. If you set a fixed voltage and run AIDA, OCCT, or some other non-AVX instruction test, you can get under 60C. The problem is Prime + Auto voltage and since those two are both unnecessary, that is an easy fix. You will want to set a fixed or specific adaptive voltage, but for now please leave it as it is so the data doesn't change. Prime + Auto is not responsible for the high coolant temps and that needs to get sorted first. However, in the meantime no reason to run Prime. If you need to create a 100% load, use Intel XTU. Nice and mild with a smooth load pattern. Not a great stability test. Useful for diagnosing cooler problems because it won't put you into thermal shutdown in 5 seconds.

Also for some weird reason, sometimes when I'm watching temperatures, it would hit 70c for half a minute then go back to 60c and stay for 10 mins or so then it would go back up to 75 and repeat the process. This doesn't happen all the time but seems like really big fluctuation in the temps so is this normal?

There are possible natural reasons for this to occur, but I am concerned it connects with the high coolant temperature. Could be the pump is cutting in and out. That would answer two questions. Keep an eye on pump speed when this happens. Look for any strange fluctuations. It should be relatively steady +-50 rpm at around 2000 rpm for Quiet or 2950 (ish) for Performance pump settings.

No need to reapply TIM. That does not appear to be an issue. Test and voltage management can be tweaked, but the high coolant temp is the prime concern. Things to watch out for:

1) You cold boot or wake from sleep and the H100i v2 temp just keeps going up and up. A 240mm in the roof should hold about 6 or 7C over room temp, so my expectation is around 30C for coolant at idle.

2) A 980 Ti is space heater that also does rendering. Ultimately, it may be what's heating the case up during gaming, etc., but heat management can be dealt with and predicted. For now, run CPU only tests and don't try and take readings after gaming. I would need a lot more hands to count the number of times a 980 Ti sent someone running to this forum thinking their cooler was going to explode.

Thank you so much for clarifying. I will let my PC overnight and then check coolant temps tomorrow on a fresh cold boot. It's getting hot where I live so please assume and ambient temprature of 25c maximum.

I will keep you updated. I'll be on the look for any wierd temps with the coolant.

Okay so I've let my PC off for about 12 hours.

I just booted it up and the coolant temprature upon bootup is 27.4c and my cpu idle temp is about 31c! Much better than yesterday.

Maybe it took the thermal paste a bit of time to cure.

After 5 minutes, the coolant temp has increased to 30ish, give or take 0.5c.

Since tempratures look much better, I will run a stress test and see if the results are any better.

I'll update you

This part is concerning, unless the temperature inside the case is also in the low 40's. Then it is absolutely normal. As such, a lot of this hinges on that measurement. In order to take other case heat out of the equation, the next time you don't need your PC, put it to sleep or shutdown for at least 30 minutes, then boot up. Take a H100i v2 temp reading straight away. In 10 minutes at idle, take another. I'll continue to assume normal room temps 20-23C unless you tell me otherwise.

 

 

 

 

 

 

No, overclocking the CPU is not an issue. My statement was contingent upon a few keys things in yours. 1) Prime95 and 2) stock voltages. Prime 95 is already pretty brutal on Haswell and later, but with the voltage on Auto it will suck you dry throwing voltage at the test. If you set a fixed voltage and run AIDA, OCCT, or some other non-AVX instruction test, you can get under 60C. The problem is Prime + Auto voltage and since those two are both unnecessary, that is an easy fix. You will want to set a fixed or specific adaptive voltage, but for now please leave it as it is so the data doesn't change. Prime + Auto is not responsible for the high coolant temps and that needs to get sorted first. However, in the meantime no reason to run Prime. If you need to create a 100% load, use Intel XTU. Nice and mild with a smooth load pattern. Not a great stability test. Useful for diagnosing cooler problems because it won't put you into thermal shutdown in 5 seconds.

 

 

 

 

 

 

There are possible natural reasons for this to occur, but I am concerned it connects with the high coolant temperature. Could be the pump is cutting in and out. That would answer two questions. Keep an eye on pump speed when this happens. Look for any strange fluctuations. It should be relatively steady +-50 rpm at around 2000 rpm for Quiet or 2950 (ish) for Performance pump settings.

 

 

 

No need to reapply TIM. That does not appear to be an issue. Test and voltage management can be tweaked, but the high coolant temp is the prime concern. Things to watch out for:

 

1) You cold boot or wake from sleep and the H100i v2 temp just keeps going up and up. A 240mm in the roof should hold about 6 or 7C over room temp, so my expectation is around 30C for coolant at idle.

 

2) A 980 Ti is space heater that also does rendering. Ultimately, it may be what's heating the case up during gaming, etc., but heat management can be dealt with and predicted. For now, run CPU only tests and don't try and take readings after gaming. I would need a lot more hands to count the number of times a 980 Ti sent someone running to this forum thinking their cooler was going to explode.

After 30 minutes of Prime 95. Seems much much better! The max temps were only hit once for maybe 5 seconds. It's probably averaging 66c.

Everything is on stock settings. This seems reasonable as its hovering around 60-70. Nothing higher or lower. I will try a small 4.4 oc and let you know how it goes.

The water coolant temps also look much better under full load.

https://gyazo.com/e4d7329fb99e8d3ef4ab4b4395066ff4

....

Looks like still problems.

4.4 overclock and 1.23v

https://gyazo.com/686305d0f10e0bbd0ee7cd119eb2e116

Hit 90c within 6 minutes at full speed fan and pump

Okay so I've let my PC off for about 12 hours.

I just booted it up and the coolant temprature upon bootup is 27.4c and my cpu idle temp is about 31c! Much better than yesterday.

Maybe it took the thermal paste a bit of time to cure.

After 5 minutes, the coolant temp has increased to 30ish, give or take 0.5c.

OK, that is more or less on the mark and +-1-2C is completely irrelevant here. The bad cooler would keep climbing right to 40C and higher at idle.

Since tempratures look much better, I will run a stress test and see if the results are any better.

Your Prime core temps are exceeding the coolant temperature by 40-50C. A +40C delta over coolant temp might be common when running into the last ticks of usable voltage, but that is not where you are. Normally, this would suggest you have contact problem (not TIM) with the backplate and CPU. It can be rather fiddly. However, there are two things bothering me. First, I am surprised to see the coolant temp go +10C in 5 minutes with the fan at max. Inevitably, it will go +10C, but that is a little quick. Second, it is extremely tedious to separate the mechanics of Prime, which version you are running, is AVX instructions enabled, how is the test configured, and how will all of that react with the Auto voltage table. This is why I suggested you run Intel XTU to take those elements out of the equation. As a general principle, it is always good to run a different stress test when another brings back an unexpected result. If XTU hits 80C+, we know there is a problem and not just voltage and test settings.

OK, that is more or less on the mark and +-1-2C is completely irrelevant here. The bad cooler would keep climbing right to 40C and higher at idle.

 

 

 

 

Your Prime core temps are exceeding the coolant temperature by 40-50C. A +40C delta over coolant temp might be common when running into the last ticks of usable voltage, but that is not where you are. Normally, this would suggest you have contact problem (not TIM) with the backplate and CPU. It can be rather fiddly. However, there are two things bothering me. First, I am surprised to see the coolant temp go +10C in 5 minutes with the fan at max. Inevitably, it will go +10C, but that is a little quick. Second, it is extremely tedious to separate the mechanics of Prime, which version you are running, is AVX instructions enabled, how is the test configured, and how will all of that react with the Auto voltage table. This is why I suggested you run Intel XTU to take those elements out of the equation. As a general principle, it is always good to run a different stress test when another brings back an unexpected result. If XTU hits 80C+, we know there is a problem and not just voltage and test settings.

I do think it's a contact issue too. I had rotated the backplate in both directions that are possible. I have tried lightly screwing the cooler and firmly screwing it.

Im running prime 95 V. 28.10 (latest one). I am running blend test. No settings in prime 95 were changed so it is all default.

I will overclock it and try Intel XTU and update you with the results. I will also try Aida 64.

OK, that is more or less on the mark and +-1-2C is completely irrelevant here. The bad cooler would keep climbing right to 40C and higher at idle.

 

 

 

 

Your Prime core temps are exceeding the coolant temperature by 40-50C. A +40C delta over coolant temp might be common when running into the last ticks of usable voltage, but that is not where you are. Normally, this would suggest you have contact problem (not TIM) with the backplate and CPU. It can be rather fiddly. However, there are two things bothering me. First, I am surprised to see the coolant temp go +10C in 5 minutes with the fan at max. Inevitably, it will go +10C, but that is a little quick. Second, it is extremely tedious to separate the mechanics of Prime, which version you are running, is AVX instructions enabled, how is the test configured, and how will all of that react with the Auto voltage table. This is why I suggested you run Intel XTU to take those elements out of the equation. As a general principle, it is always good to run a different stress test when another brings back an unexpected result. If XTU hits 80C+, we know there is a problem and not just voltage and test settings.

https://gyazo.com/09371b81c4055de6313114494a7aed46

Much better using Intel Stress Test.

This is 4.4ghz on 1.22V. That was a full 20 minute run and it probably averaged around 75c throughout the 20 minute run.

I still do think those temps are a little high for a 4.4ghz OC. Also, are those water temps right? 45c seems on the limit.

And, I thought about the radiator placement being the problem? I currently have it as exhaust at the top. Will there be any difference if I move it to front intake?

Thanks!

Whoops, I should have told you to hit the wrench on the line graph and click cores 1-4 to show. No matter. The delta over coolant is better at around +30C at that is more in line with what I would expect at that voltage. The actual load voltage is still sliding around a little between 1.24-1.26, but that is OK. It is expected and can be addressed through BIOS settings.

The coolant temperature is still bothering me. What was the starting H100i v2 value? I am going to assume somewhere back at 30-31C, so that is a +15C rise in coolant temp on XTU at 113W draw in 20 min with the fans maxed out? I don't have another 240mm cooler to compare, but it seems to much to me. I couldn't make my H110 go up +15C with any stress test and that is with submaximal fan speeds. Your pump and fans are maxed out, so I know it is getting the full voltage. This leaves one of two possibilities:

1) The unit has some sort of defect or partial obstruction that is slowing the coolant flow when under stress. I would normally expect to see this in evidence at idle too, so I am not sold on this idea. Nevertheless, +15C coolant rise on 113W is a lot.

2) Something is cutting off the airflow or restricting it some way that affects the cooler's operation. In the past, there have been some issues with 440's in combination with those side inlets. In a CPU only load, I would not expect you to ever break 40C in your environment. With the Ti is a whole different issue. I certainly favor top mount exhaust, as you have it. Is there anything above the cooler that might contribute to this condition? Dust filter on top? Is the case squeezed under a desk or cabinet with no room above? How exactly does that top work? Can it come off? Same side inlets, it appears... If you put your hand on top of the case while gaming, stress testing, does it feel unusually warm?

I am a little concerned if this is case related. In combination with the Ti in a mixed moderate GPU/CPU load, you might see some very disadvantageous water temps.

http://imgur.com/a/MIrvw

that's cold start up after 20 hours of being left off.

then 10 mins idle

today is colder than it has been. but usually if its 25c outside I cant game without my cpu over heating.

this is after 2 hours of playing bf1.

http://imgur.com/a/pVzkg

I am not sure I would call a 70-71 peak on your hot core as overheating. Nevertheless, you coolant delta over the period was +9C. That is quite good on a moderate to high combo CPU/GPU load like BF1. The 1080 Ti tops out at 61C, which is also very good, certainly compared to its predecessor. Your temperatures look quite normal for a 4790K running above 1.30v.

The other thing to take a look at is whether you are steady running in the 60's or whether these are simple peaks that occurred for one instance in time and not again. I find it somewhat ironic the highest peak CPU temps I see all week come when I launch a game, not actually play it. BF1 is a moderate to high combo load game and these temps do not seem overly different from others report with similar voltages across the 4 core line. If you really care about the peak value, you'll need to tighten your BIOS settings a little.

I am not sure I would call a 70-71 peak on your hot core as overheating. Nevertheless, you coolant delta over the period was +9C. That is quite good on a moderate to high combo CPU/GPU load like BF1. The 1080 Ti tops out at 61C, which is also very good, certainly compared to its predecessor. Your temperatures look quite normal for a 4790K running above 1.30v.

 

The other thing to take a look at is whether you are steady running in the 60's or whether these are simple peaks that occurred for one instance in time and not again. I find it somewhat ironic the highest peak CPU temps I see all week come when I launch a game, not actually play it. BF1 is a moderate to high combo load game and these temps do not seem overly different from others report with similar voltages across the 4 core line. If you really care about the peak value, you'll need to tighten your BIOS settings a little.

oh no i agree those temps are not over heating but when its 25c out and last week i had 3 instances of my CPU overheating and shutting down while playing BF1. i hadnt noticed the LED on the cooler go red atall.

Alright, that is a little different. As your current peak is around 70C, you would need to be +25 to 35C hotter in order to reach your CPU thermal shutdown. You won't get that kind of environmental shift, so that points the finger back at the cooler. It seems to work just fine in some moments, so the most likely circumstance is the pump is shutting off for unknown reasons. Make sure CPU_FAN or wherever the H100i v2 is connected is receiving a full 12v, 100%/PWM signal or "disabled" on some boards. Obviously check to make sure the connection is secure. Watch out for desktop software fan controls (Gigabyte) that might override your 100% BIOS setting for CPU Fan. It could have nothing to with any of these things and might be an internal problem. Nevertheless, cross of the easy stuff first.

How do you know it was a "thermal shutdown". Did you get the CPU overheat error on reboot?

I've managed to solve the issue!

What I did was try my cooler in all 4 different directions including backplate and they all spiked up to 100c if I slightly oc the cpu.

I solved it easily!

What I did was, take the cooler off. Pack it back in the box and return it to memory express. Then buy a Nh-U14s and TADA! 57c max temp on 4.5ghz OC with XTU.

AND its like $30-40 cheaper.

Never am I buying a corsair product again.

big thank you to c-attack for helping me through the issue. Community is great, just not the product