Corsair H115i Temperature Issues

Hey guys,

I just got a Corsair H115i AIO Liquid Cooler and was wondering what sort of idle/max temps other people are seeing.

IMO, the temps are too high on this thing. I've tried changing the fan config and applied a different thermal paste (CM Ice Fusion) several times, taking care to tighten and seat the block properly.

I'm running a 4790K on an Asus Maximus VII Formula inside a Cosmos II, with plenty of other fans installed as well to throw all the hot air out of the box.

The ambient temps in my room right now are about 27C since I don't have an AC installed in here at the moment and the summers in Pakistan are quite warm. The fans for the cooler are currently acting as exhausts since they cannot be mounted in the bottom part of this case.

My idle temps are around 40-45C and jump up to about 60-65C on gaming load. Prime 95 on small FFTs takes the temps to about 80C. Intel XUT stays at around 70-75C.

The water temps start at about 36C on idle and settle at about 39C. The highest on load I've seen it go is 45C.

Would appreciate any help and information you guys can provide on how comparable AIO set ups are running and what I can do to improve these temps because they seem quite horrible to me.

Unfortunately, hot room = warmer CPU temps for anything except exotic cooling methods. Based on the screen shots, your internal case temperature is pushing 40C, so this is the temperature your CPU starts at plus voltage. So, if you knock 15C off your top end to match the starting temperature of most published results, do you still feel they are that bad? There is a reason all these chronic benchmark posters do their heavy work in the Winter with the windows open.

Water temperature is slow to change both up and down, so once you start running some loads, you need to let it cool down for quite a while to maintain precise results. However, I would recommend you give the stress tests a break until you get some AC or the weather turns. The results will never be good and you adding even more heat to the room. To my eyes, the cooler is working properly. Water temperature is changing, fans reacting. If there was a contact issue between the cold plate and CPU, Prime 95 would be untenable. As it is, your delta over water temperature looks good. One thing you can do right now is take manual control of your fans. The base curve is set with a 20-23C room in mind. Your 40C case temperature represents +20C rise in water temperature and so the system is ready to blast the fans at 100%. Now matter how long you let them run, you can't drop below the case ambient temp. Cut the fans down to something you can stand. A fixed setting of 700 rpm will be just fine for normal use and most games.

The one thing I did notice is your case temperature delta over room temperature. Most people are between 4-7C over the room temperature when at idle. You are at 10C, if the 30C measurement is accurate. Now, I am just looking at a snap shot and if you have been running hard for a while, this may be perfectly normal. However, reducing the starting water temperature is another way to reduce overall CPU temperatures. People that have high idle water marks over the room temperature usually have one or more of the following: 1) Case with poor ventilation, either through design or fan placement and speed; 2) Big watt GPUs such as a 980 Ti SLI. I don't think one 1080 qualifies, but if you have been running Heaven or some GPU bench, it would make a difference. 3) A restrictive physical environment surrounding the case -- sandwiched between two walls and a desk, in a cabinet, covered in cats, etc.

Now all of this could easily be from a hard day's work at the bench. Take another check of water temperature (H115i Temp) tomorrow morning after first boot. Both the system and room should be cooler at that point. Take another glance 15 minutes later after the system has had a chance to settle into a normal operating temperature.

Thanks for the detailed response. I'm not into running benches. The screenshot was just to show what is happening when load is placed on the PC. The most load I've placed on this after installing the H115i is a couple of rounds of Overwatch and some No Man's Sky.

I was actually getting better performance from my Cooler Master V10. With that, the idle temps were 35-38C and maximum of 60C on load. No way that outperforms a water cooling solution.

With a cold boot, the starting water temp is about 35C and the CPU temp is about 40C. I don't think I have poor ventilation. I have 2 intake fans (230mm + 140mm) and 4 exhaust fans (2 x 140mm + 2 x 120mm) Not counting the ones on the radiator.

Here a few odd things I am noticing:

- Opening even a browser like Vivaldi with multiple tabs already open pushes temps to the 50s

- Running the system with both doors open only lowers temps by about 2C.

- The fans don't seem to make any difference to the temps, not even the water temp! I have a custom curve set up that turns fans to full when CPU hits 60C. But I've ran the system with fans on full speed for an hour and the temps were still 42-44C on the CPU and 38C for the water.

- Ran the system for about 10 mins with the fans on the radiator stopped and the temps remained the same. Just water temperature rose slightly higher than normal, which is expected.

Now all of this may point to a bad contact, but I've already seated and re-seated this thing like 5 times. (yes, that many)

So, I am looking at any other issues that may be causing the high temps until I can get my hands on some Arctic Silver in a few days and try again because I think the old IceFusion I had lying around may have gone bad. (I've had that for about 2 years now)

The thing that irks and concerns me the most is the high temp with a cold boot.

Hi xstrike9999, I had a similar issue with my H115i after my build. I couldn't get my cpu temps down to a level of sanity. I too re-seated my cooler several times. After many google searches I eventually solved my problem.

First; Make sure your CPU is set to manual not auto voltage, Not sure what your stock cpu volts are but 1.2v seems an average starting point. On auto my temps were awful at load and at stock ghz. After a manual voltage I found a great improvement.

Second; I dropped an average of 7-10 degrees at load when I fitted the supplied washers on the back plate. I think some Asus mobo PCboards are slightly thinner. I was quite happy after these two changes.

Third; AT YOUR OWN RISK. I delidded my cpu and used liquid pro. This also made a difference.

I'm now running 4.8ghz at temps that are great in synthetic benchmarks.

Hey lebel,

Thanks for the tips.

I'm attaching a pic of my voltages at load. I think I have the voltage set to manual but will have to check that. Although I do know that I have all the Intel power management features like speed step etc enabled.

On idle, voltages stay below 1.22v - 0.97v to be exact. On load they jump to 1.22v and sometimes up to 1.27v as you can see in the HWInfo pic.

I can definitely take a crack at using the washers once I get the Artic Silver and if there is still no difference after using that.

Could you explain how you fitted the washers on the backplate? The Maximus VII is supposed to be one of the "tough" ASUS boards so I don't think board thickness is the issue here, but I am definitely willing to try anything at this point.

Hopefully, the Arctic Silver will get delivered tomorrow and I'll report back with the results.

I'm attaching a pic of my voltages at load. I think I have the voltage set to manual but will have to check that. Although I do know that I have all the Intel power management features like speed step etc enabled. If you reset your bios to basics (F5) then add your manual voltages, then this should leave everything else at auto. Just a tip, I always clear CMOS before I change/alter any settings

On idle, voltages stay below 1.22v - 0.97v to be exact. On load they jump to 1.22v and sometimes up to 1.27v as you can see in the HWInfo pic. I wouldn't expect to see those load temps around 80+ that your getting at those volts. I also noticed your SSD at a strange temp as well, not sure what's going on there. Does look like the heat isn't getting away quick enough from the cpu to the cooler.

I can definitely take a crack at using the washers once I get the Artic Silver and if there is still no difference after using that. I would recommend Artic MX-4, I tried Artic silver and found a couple of degrees less performance. Also MX-4 isn't conductive so no worry's on overspill. Also I used a garden pea size and not a rice grain as some reviews have said. I found the rice grain approach gave mixed result's due to the unevenness of clamped cooler to cpu. A little more is better/safer than a little less.

Could you explain how you fitted the washers on the backplate? The Maximus VII is supposed to be one of the "tough" ASUS boards so I don't think board thickness is the issue here, but I am definitely willing to try anything at this point. The washers I used came with the cooler, I put them straight onto the back plate around the thread/bolt holes before pushing the backplate through the back of the mobo. Putting the washers on will have no issue unless there is any conductive areas at the back-near where the washers would now sit.

Hopefully, the Arctic Silver will get delivered tomorrow and I'll report back with the results. Good luck.

Hopefully others in here could give you more tech insight as I'm new at all this after a long spell away. My last build involved an Athlon 3200 and a Nvidia 6800 ultra.

I was actually getting better performance from my Cooler Master V10. With that, the idle temps were 35-38C and maximum of 60C on load. No way that outperforms a water cooling solution.

I suspect your room temperature and case temperature were lower for that. It's 40C in your case right now. Your CPU is 40C with zero voltage. Since the air tower can't remove every watt, you would probably idle in mid 40's under these conditions.

With a cold boot, the starting water temp is about 35C and the CPU temp is about 40C. I don't think I have poor ventilation. I have 2 intake fans (230mm + 140mm) and 4 exhaust fans (2 x 140mm + 2 x 120mm) Not counting the ones on the radiator.

No, you have plenty of fans and the balance seems OK. There were other reasons you could have a very high case temperature compared to room temp. +5C idle over water temperature is fairly normal. However, use your individual core temperatures for discussion purposes. The "CPU Temp" and CPU Package Temp can be strange numbers and are absolutely CPU model specific. Right now my "CPU temp" is 4C above my hottest core. Certainly not an average. Package temp is +14C over the hottest core. Under a load, my cores will all rise above the "CPU temp" even at their lowest levels. HW-E is weird.

Here a few odd things I am noticing:

 

- Opening even a browser like Vivaldi with multiple tabs already open pushes temps to the 50s

Steady temps? Or just a momentary spike? Windows 10 is rather prickly with its resource allocation. I run Aida64 as a monitoring tool and it has rather nice monitoring line graphs. The highest peaks in my day are when I open Chrome or launch a game, not the game play itself. Now if launching a browser causes sustained 50C temps, that is more interesting. Does it also happen with Windows Explorer/Edge?

- Running the system with both doors open only lowers temps by about 2C.

OK, I agree and don't think your exhaust flow is the critical issue. I think chances are the difference in case temp versus room temp was at it's apex after a long hot day. Also room temperatures are far from even and it almost certainly warmer in the area around the case than on the wall across the room, which would that delta smaller and closer to expected. Either way, we are talking about a couple of degrees difference and not the reduction you are looking for. This is something that can be examined at a later date if you are looking to squeeze every out every drop you can.

- The fans don't seem to make any difference to the temps, not even the water temp! I have a custom curve set up that turns fans to full when CPU hits 60C. But I've ran the system with fans on full speed for an hour and the temps were still 42-44C on the CPU and 38C for the water.

 

- Ran the system for about 10 mins with the fans on the radiator stopped and the temps remained the same. Just water temperature rose slightly higher than normal, which is expected.

The fans remove heat from the water. They can't do anything about the voltage passing through the CPU or the rate it is transferred through the lid, TIM, and cold plate, and finally into the water. That transfer rate is more or less fixed and beyond your control. It's down the properties of the materials in that chain. High end cold plates that run $150-200 for the block alone, sometimes advertise a 1-3C improvement over others. This is not an area where you can make great strides and there certainly isn't much difference between various AIO and quality air towers in that material. This is a long way of saying fan speed and pump speed have little effect on momentary voltage spikes that often represent your "peak values". If you take an air tower, a water system, and a cheap stock cooler and bolt them down on the same CPU with the same TIM, and (we'll pretend) identical plate material, they would all produce the same CPU temps for the first few milliseconds of identical voltage. It's what happens after that makes the difference. The cooling system must remove the heat as fast as it's applied or the temperatures will increase. The little stock cooler is overwhelmed in a few seconds. The air tower can do several seconds longer before needing to increase it's fan speed to keep up. The water system can absorb quite a bit of heat into the water system before CPU temps are affected. You must raise the water temperature by 1C to raise the CPU temp by 1C. With the fans off you can see this play out. Even with moderate fan speed, a big radiator like a 280mm is going shed quite a bit of that heat. That is the big advantage of a water system. You have a large capacity to absorb heat, unlike the other types.

Your water temp increase by 1.2C between the idle and load screen shots (first post). The most you can reduce your CPU temps is by the same 1.2C.

You can achieve your best CPU temperatures by keeping the water temperature as low as possible. This is your baseline and also represent the lowest possible CPU temperature. You were already at your lowest temperature. More fan speed won't help. To that end, I would suggest you eventually return to a fan speed based on water temperature (H115i Temp) instead of CPU temp. The fans don't work that way, it will make them run more often and change speeds far more frequently than needed, Windows 10 makes a mess of CPU activity and that does not seem likely to change anytime soon, and other users have reported some strange fan behavior when linked to CPU temp on the H115i. It is designed to run from the water temp and the original H110 GTX did not have this capability. It is possible not all the kinks are worked out.

Now all of this may point to a bad contact, but I've already seated and re-seated this thing like 5 times. (yes, that many)

I don't think you have a contact problem. When you do, CPU temps are jumpy and erratic all loads, including idle. More importantly, your Prime95 run would have not be in the 70's. 90-100-shutdown. Sometimes one or more cores are massively warmer than the others. Not the normal 5-8C difference in peak values, but 20-25C. Your spread looks pretty normal. But most telling is your CPU core temp over water temperature delta. You are at +30-38C peak over water. People with contact issues hit 90C with a 25C water temp. Higher than expected water temperature is never a contact issue.

Just for comparison, lets look at your Intel XTU temps since I can run that one too. XTU has a very clear wave pattern in its loading the temps will fluctuate +-7-8C as it cycles. In your screen shot, you are in the upper 50's as I would expect. However, that is the benchmark run which will produce a higher peak value than the actual stress test. You can run the stress test portion to get a better feel. 5 minutes is plenty to assess thermal transfer in the cooling system.

The thing that irks and concerns me the most is the high temp with a cold boot.

But what was the room temperature at cold boot? You stated 35C water temp. I would expect a room temperature in the neighborhood of 28-30C.

@lebel22

I only had the option of AS5 here in Pakistan. It was either that or the Ceramique 2.

I'll definitely try the washers as soon as I receive the new TIM and see if that improves contact and temps.

@C-attack

I removed the V10 just two days ago when I received the H115i. The ambient temps these days range between 23-27C in my room so those have not changed.

For example, the ambient was 24C today and as you say, the water temp should be around 30C at those temps but it starts at 35-36C with a cold boot and settles at about 39C.

The spikes are temporary when opening browsers etc. I get temps of about 60-64C on a game like No Man's Sky.

I ran the XUT stress test as you said (image attached) and the peak temp is 74C. Overall the range for the test was 66-74C. The water temp as you can see is reaching 41C.

These kinds of temps are consistent with the overclocked load temps here: http://www.tweaktown.com/reviews/7320/corsair-hydro-h110i-gtx-high-performance-liquid-cpu-cooler-review/index7.html

But as you know, I have no OC on this CPU at the moment. The reviewers aren't just measuring the water temps, are they?

The room temperature right now is about 24C since it rained last night and a cold boot yielded the same water temps of 35C. As you said, at these ambient temps, the temperature should be 28-30C.

I pressed my ear to the pump in before one of the re-seatings and turned it on and it seems to be working fine. Granted I don't know what a bad pump sounds like, but I didn't hear any strained or weird noises.

Any other ideas why I might be seeing such high water temps? I could try lebel's suggestion of using the washers and see if it improves things. I am tempted to RMA this thing otherwise.

The washers will not help with water temperatures. They are for when it is difficult to get the cold plate and CPU to make full physical contact. When that is the problem, CPU temps will be warm at both idle and load, but the water temperature is unchanged and often cool.

I do think the XTU results are too high. I do think your idle water temperature is too warm. Coincidentally, it is about 7-10C over what I would expect and the CPU temps are also 7-10C high. I think this where we need to focus for now.

The Cosmos II is no micro-ATX. I would not expect temperature issues under normal conditions. Just to confirm the first post --- the H115i is mounted in the top with the fan as exhaust. Correct? That is how I would do it as well.

Your first two Link screen shots from Post 1 clearly have a 40C water temperature. I assumed this was from a hot room + retained heat + lots of use. However, in a 27C room I am not sure I could get my H110 up to 40C without deliberately chocking the fans down to some menial level for an extended duration. This is a big jump. The water temperature can't be less than the internal case temperature, but it can be more. If it's a lot more at idle or cold boot, that suggests an internal problem. Thus far, I have been looking at your Link shots and what I presume is your motherboard temp sensor (Temp #1). In first two screen shots, was the same as the water temps which were similar to non-C drive temps, all of which suggested the internal temp was also 40C. I am again seeing the same thing in the XTU screen shot. However, I cannot be certain Temp 1 is really your motherboard sensor. Do you have another way to confirm the motherboard temperature? An Asus utility or any other monitoring program. I am pretty sure your Corsair SSD isn't 128C, so Link is not infallible.

The quick rise in water temperature up to that 40C is concerning, but I need to be able to isolate it from the rest of the case. A cooler that rises to 40C at idle when it's 30C in the case has a flow problem and needs to be replaced. A cooler that rises to 40C and everything else in the case is 40C as well, is just at the mercy of something else. Is there any kind of dust filter restricting the H115i exhaust? Is there any kind of desk or shelving or other obstruction that would prevent the air from escaping the top? Solid top lid?

The Cosmos II doesn't support this radiator in the top part, so the radiator is mounted inside the case and the fans are in the top part in a pull config, i.e. throwing the heat from the radiator out of the case.

I have the case in a very open area. The only thing near it is my TV and Logitech Z5500 on the left of it and nothing but open space on the right side. The space from the back wall is about 15 inches.

The top lid is standard Cosmos II lid. It had a built-in dust filer and nothing else that could restrict air escaping from the top.

All the hard drives in there might be jacking up the temp. The SSD temp is definitely faulty. It has been that way since I got it. It's cool to the touch so the reading is definitely a bug.

You may be right that it is the temp inside the case that is causing this issue. I'm gonna see if I can get an infrared thermometer and just get the temp reading that way to see what the actual internal temp is.

With that many exhausts fans in this huge case and at this ambient temp, it shouldn't be that high. Something is definitely not right.

Having the fans above the radiator in the pull position is as complete non-issue. No problem there. Hopefully, the dust filter is removable. At the high fan speeds currently in use, I don't think it can cause this rise in temperature, but it would be a heavy restriction at the low fan speeds we are trying to get back to.

I don't see any clear problems. Even the HDD shelf, which appears to be below the main chamber, should not affect the internal temps much since it has it's own dual exhaust fans down there. It certainly is not capable of heating up the main chamber on a cold boot.

So, as I understand it, there is a 200mm front intake fan, the rear 140mm is also intake, two radiator fans as exhaust, and the two lower HDD bay fans as exhaust. It is possible prolonged gaming (or any long GPU load) might raise the water temp a little more than expected because all the waste heat is forced out the top through the radiator. However, that does not explain the quick rise on cold boot. This is starting to look like some sort of partial blockage or flow restriction, but I do not want to tell you take it back until we are sure.