Corsair H110i runs hot

Hi,

Yesterday I bought Corsair H110i. It is installed on 7700K @4.8ghz 1.295V

My idle temps are around 35-38, sometimes they remain around 40 on idle, after heavy load. RPM is around 2250-2380, so I'm thinking there is something wrong here.

This is the setup.

I've read that people with 5.0ghz OC's have their 7700K sit on 26 on idle and around 55-60 on load.

Mine goes to 80's on load which is basically what my previous air cooler gave me.

Can it be an installation problem or a product defect ?

We can't see the coolant temperature in that shot. Can you post the home screen or otherwise describe what happens with the coolant temperature (H110i Temp)? This is the difference between a cooler issue and all the other stumbling blocks.

What is the coolant temperature ?

Post a screen shot of your home page. The coolant temperature is the measure of how much heat is in the system. Also known as liquid temp, water temp, or in Link the temp reading in the H110i box on your home screen.

Coolant temp should hover about 4-7C above your room temp at idle. When you cold boot or wake, it will start off exactly at room temp, then warm slightly to match the internal case temp and small amount of idle CPU heat. At load, you might go +6C or so for a CPU only 100% max. The trickier bit is gaming where GPU waste heat warms the entire case and radiator along with it. You might see +10C, although most of it is not CPU related.

We are looking to compare your coolant temperature to the room/case temp at idle and when at load to see if it behaves in a normal fashion. If you wake the PC and the coolant goes right to 40C, you know there is a problem. If it appears normal, but then stays elevated after a long gaming session, you most likely are looking at case heat management issues. Lots of variables, but hot on cold boot is never good.

Post a screen shot of your home page. The coolant temperature is the measure of how much heat is in the system. Also known as liquid temp, water temp, or in Link the temp reading in the H110i box on your home screen.

 

Coolant temp should hover about 4-7C above your room temp at idle. When you cold boot or wake, it will start off exactly at room temp, then warm slightly to match the internal case temp and small amount of idle CPU heat. At load, you might go +6C or so for a CPU only 100% max. The trickier bit is gaming where GPU waste heat warms the entire case and radiator along with it. You might see +10C, although most of it is not CPU related.

 

We are looking to compare your coolant temperature to the room/case temp at idle and when at load to see if it behaves in a normal fashion. If you wake the PC and the coolant goes right to 40C, you know there is a problem. If it appears normal, but then stays elevated after a long gaming session, you most likely are looking at case heat management issues. Lots of variables, but hot on cold boot is never good.

This is HWmonitor + Corsair link when the PC is idle, approximately 5 minutes after booting the PC. Ambient temperature is around 20-22 C°

And these are the temps right after I used the stress test on cpuz for a minute.

And finally this is right after a Heaven Benchmark with slight OC on the GPU ASUS DUAL-rtx 2060

Seems like the problem occurs when GPU kicks in. Now I hear the fans going loud and the radiator area seems hotter.

You don't have the cooler connected to USB? It's not listed in Link.

And I don't see anything wrong with those CPU temps.

You don't have the cooler connected to USB? It's not listed in Link.

And I don't see anything wrong with those CPU temps.

I guess not. I didn't do the installation myself.

As for the temps, those temps are similar to when I had a Thermaltake mid range air cooler, with quarter of h110i's price.

I was expecting to get better cooling so I could amp my OC.

Without that USB cable, you have no control over the cooler, its fan speeds, the pump, the lighting, and no access to the relevant coolant information to see how its performing. Water cooling is going to give you better thermal capacity than air cooling. You can go harder for longer without suffering additional heat climb. However, that instant Vcore max temp spike is always going to be the same whether you have $20 blower or $1000 external cooling system. The source of the heat is the CPU pins and that heat must travel through the CPU before the cooling system conducts it away and disposes of it elsewhere. The rate of conductivity through the block is pretty much the same regardless of type. Even with the $1000 cooling system, you can't set your Vcore to 1.60v and you will only last a second or two longer than a cheaper cooling method.

However, presumably you didn't put the system together to run CPU-Z or any other stress test. The better comparative measure would be avg and peak CPU temp over time doing what you normally do. A 280mm cooler probably has a solid 10C advantage over a good box air tower. You already have a fairly decent overclock to 4.8 GHz and probably can go higher, but understand if you choose to run max stress tests you likely will go into the 80s regardless of cooler. That is a product of the voltage. The only 7700K owners who dodge that issue at 1.30v+ are those that have delidded.

Without that USB cable, you have no control over the cooler, its fan speeds, the pump, the lighting, and no access to the relevant coolant information to see how its performing. Water cooling is going to give you better thermal capacity than air cooling. You can go harder for longer without suffering additional heat climb. However, that instant Vcore max temp spike is always going to be the same whether you have $20 blower or $1000 external cooling system. The source of the heat is the CPU pins and that heat must travel through the CPU before the cooling system conducts it away and disposes of it elsewhere. The rate of conductivity through the block is pretty much the same regardless of type. Even with the $1000 cooling system, you can't set your Vcore to 1.60v and you will only last a second or two longer than a cheaper cooling method.

 

Is there any way of testing whether it is plugged in or not ? Or if it is connected, what might be causing the detection failiure ?

However, presumably you didn't put the system together to run CPU-Z or any other stress test. The better comparative measure would be avg and peak CPU temp over time doing what you normally do. A 280mm cooler probably has a solid 10C advantage over a good box air tower. You already have a fairly decent overclock to 4.8 GHz and probably can go higher, but understand if you choose to run max stress tests you likely will go into the 80s regardless of cooler. That is a product of the voltage. The only 7700K owners who dodge that issue at 1.30v+ are those that have delidded.

I won't be able to delid mine, I don't have the equipment nor have I done anything like that before so I wouldn't risk breaking it down.

I was just hoping to OC it up to 5ghz without going over 75 C° with this cooler.

Is there any way of testing whether it is plugged in or not ? Or if it is connected, what might be causing the detection failiure ?

 

 

 

I won't be able to delid mine, I don't have the equipment nor have I done anything like that before so I wouldn't risk breaking it down.

 

I was just hoping to OC it up to 5ghz without going over 75 C° with this cooler.

That was kind of my point. No one can run 5.0 Ghz on 7700K without breaking 75C on full bore stress test - regardless of cooler. However, that does not meant you can't run 5.0 GHz and not break 75C while gaming or other sub-maximal loads. Nobody accidentally starts a stress test while doing something else. Use your real use data to make judgments about temperature.

The only way to get by the voltage barrier is to make heat conduct better through the CPU (delid) or to reduce the starting CPU temperature to the point where a +50-60C voltage spike no longer puts you in the red. That's what extreme overclockers do with liquid nitrogen. Obviously that is not a serious avenue of consideration, but those are the limiting factors.

The USB cable would be hard to miss. It should be a right angle mini-USB coming out of the pump block, which likely then runs behind the motherboard down the bottom where the internal USB connectors are most often found. My guess is there's no cable. Most boards only come with one internal USB 2.0 these days and they may have used it for something else. That's OK. This is not a terribly difficult obstacle. ANY usb mini to internal 2.0 connector should work. They are cheap. However, first find out if you have an internal connection available. If not, you might need a splitter (also cheap) or could run a standard mini- to USB A to back of your I/O panel like any other USB (probably less desirable). In the short term, I don't think there is anything wrong with your cooler, but you certainly do want control over it.

That was kind of my point. No one can run 5.0 Ghz on 7700K without breaking 75C on full bore stress test - regardless of cooler. However, that does not meant you can't run 5.0 GHz and not break 75C while gaming or other sub-maximal loads. Nobody accidentally starts a stress test while doing something else. Use your real use data to make judgments about temperature.

 

The only way to get by the voltage barrier is to make heat conduct better through the CPU (delid) or to reduce the starting CPU temperature to the point where a +50-60C voltage spike no longer puts you in the red. That's what extreme overclockers do with liquid nitrogen. Obviously that is not a serious avenue of consideration, but those are the limiting factors.

 

 

The USB cable would be hard to miss. It should be a right angle mini-USB coming out of the pump block, which likely then runs behind the motherboard down the bottom where the internal USB connectors are most often found. My guess is there's no cable. Most boards only come with one internal USB 2.0 these days and they may have used it for something else. That's OK. This is not a terribly difficult obstacle. ANY usb mini to internal 2.0 connector should work. They are cheap. However, first find out if you have an internal connection available. If not, you might need a splitter (also cheap) or could run a standard mini- to USB A to back of your I/O panel like any other USB (probably less desirable). In the short term, I don't think there is anything wrong with your cooler, but you certainly do want control over it.

Alright thanks for the all input. Tomorrow I'll open the case and see for myself in the day light, if there is a cable missing like you said, then I'll take my case to tech support in the weekend and see what they'r going to say about it.

One more question, I tried OC'ing it to 5ghz and my PC starting BDOS'ing. I searched the web and read that such BSOD's might be caused by unstable OC'ing and voltage defficiency. Should I try that 5ghz OC with higher voltages ? Is it safe ?

Also I have corsair vengeance led 3000mhz(cl15). Right now I'm running them on 2666mhz(for almost 2 years now) with 13 13 15 timings which seem to work well. Do I need to increase the ram speed for a better CPU overclock ? Does it matter alot ?

I also read that kabylake cpus run hot with ram over 2400mhz. Would that be an issue ?

You will need to take some manual control over the Vcore voltage when overclocking to 5.0. Your 1.265 at 4.8 seemed OK to me, but this is both a CPU individual trait and probably best answered by 7700K owners. My 8700K is very similar in most things, but slightly different in Vcore tolerance. I would start at 1.31v and see how that goes.

As for the memory, leave it where it is. 2666 is a good stable timing. All boards can run it. You get better performance than 2133. Once you cross over to 2800 and up, you will need more DRAM voltage and things get more difficult. You may be able to run 5.0/3200, but leave the memory for last. It could tip the balance the wrong way and for gaming or other non-professional apps, there is not much to gain when going from 2666 to 3200 - besides pride.

You will need to take some manual control over the Vcore voltage when overclocking to 5.0. Your 1.265 at 4.8 seemed OK to me, but this is both a CPU individual trait and probably best answered by 7700K owners. My 8700K is very similar in most things, but slightly different in Vcore tolerance. I would start at 1.31v and see how that goes.

 

As for the memory, leave it where it is. 2666 is a good stable timing. All boards can run it. You get better performance than 2133. Once you cross over to 2800 and up, you will need more DRAM voltage and things get more difficult. You may be able to run 5.0/3200, but leave the memory for last. It could tip the balance the wrong way and for gaming or other non-professional apps, there is not much to gain when going from 2666 to 3200 - besides pride.

On MSI forums(my MOBO is msi z270 gaming m7) someone suggested me to get the ram speed down to 2400 and then OC the CPU. Maybe I'll start with that.

What do you think the upper limit for the voltage is. You told me to start with 1.31, when should I stop ?

I also opened the case but I couldn't figure out if the cables were connected properly, I think I'll take the case over to the tech support in the weekend.

Have a look at the 2nd and 3rd pictures going down (really all of them) to see the right angle USB connector jutting out of the side of the pump block. If it's there, it is hard to miss.

You could lower you RAM frequency down to JEDEC 2133 default, but at this point the BIOS is fully mature and all Z270 boards should be able to run 2666 until the very last tick of stability. You don't want to live on that edge anyway, so if you find CPU settings that will run at 5.0 + 2400/2133 but not 2666, that's not really stable enough for 24/7 worry free use.

Take a look at this blog post. I know it's Asus and you have a MSI board, but the initial page details are universal, although a bit technical in places. At any rate, it should give you an idea of expectations.

In terms of actually doing this, if everything else in on auto, you may need to tweak some other MSI specific settings related to Load Line Calibration , VRM power management, etc. These become relevant at 5.0 on most boards. There are likely some MSI specific guides out there.

Set a manual Vcore of 1.31. Go back out and run a stress test (not Prime 95) and see how things go. Lots of opinions on how long you should go. I tend to be quicker and prefer real world use once initial stability is demonstrated. If it passes the stress test for 30-60 min, you can try it out on normally. If it fails the stress test in seconds, you need to go up to 1.33 or check other settings. If it were to pass at 1.31 and then subsequently everything is good on your end for a week or two, you could try and reduce the Vcore 1/100th at time. You won't be able to come down much unless you have a real winner of a CPU. You can also see from that guide that stability can be influenced by temperature as well.

I would not get into this until you get the USB cable sorted and have control over the fans and an eye on coolant temp. Running stress tests blind is no fun.

Have a look at the 2nd and 3rd pictures going down (really all of them) to see the right angle USB connector jutting out of the side of the pump block. If it's there, it is hard to miss.

 

You could lower you RAM frequency down to JEDEC 2133 default, but at this point the BIOS is fully mature and all Z270 boards should be able to run 2666 until the very last tick of stability. You don't want to live on that edge anyway, so if you find CPU settings that will run at 5.0 + 2400/2133 but not 2666, that's not really stable enough for 24/7 worry free use.

 

Take a look at this blog post. I know it's Asus and you have a MSI board, but the initial page details are universal, although a bit technical in places. At any rate, it should give you an idea of expectations.

 

In terms of actually doing this, if everything else in on auto, you may need to tweak some other MSI specific settings related to Load Line Calibration , VRM power management, etc. These become relevant at 5.0 on most boards. There are likely some MSI specific guides out there.

 

Set a manual Vcore of 1.31. Go back out and run a stress test (not Prime 95) and see how things go. Lots of opinions on how long you should go. I tend to be quicker and prefer real world use once initial stability is demonstrated. If it passes the stress test for 30-60 min, you can try it out on normally. If it fails the stress test in seconds, you need to go up to 1.33 or check other settings. If it were to pass at 1.31 and then subsequently everything is good on your end for a week or two, you could try and reduce the Vcore 1/100th at time. You won't be able to come down much unless you have a real winner of a CPU. You can also see from that guide that stability can be influenced by temperature as well.

 

I would not get into this until you get the USB cable sorted and have control over the fans and an eye on coolant temp. Running stress tests blind is no fun.

Yeah I can see that thing but the problem is I don't know where the other end is going :D

I have an MSI Mag Bunker case and on the rightside there are holes that you put the cables through and they come from another end and into the mobo. So the cable dissapears in that hole.

Anyway I'll take it to the guy who built it and ask him about it. Once he can get it to work, then I'll go into overclocking.

Btw in the bios, there are different options for fan settings. The pump / cpu show pwm, dc and auto and pwn is selected.

Does it have anything to do with detection ?