New Desktop High CPU Temps On Water At Idle

[ovanrakjarac]

Hi guys.

Well, the title says it all. This is my first build, and I have a problem. I have an Intel Core i7-6700K that is cooled by a closed-loop AiO water cooler, namely the Arctic Liquid Freezer 360. In case you didn't guess already, this unit has a 360mm radiator. I have it mounted at the top of my case as an exhaust and in a push-pull configuration with 6 fans because it came and is supposed to be used with all of them. I used the thermal paste that came with it, and its the Arctic MX-4. The application method used was the grain of rice and then let the pump spread it evenly. As for my case, it is the Thermaltake Versa H35, a mid-tower case. As far as other cooling, I have a GTX 980 Ti with its own AiO water cooler i.e. a single 120mm radiator and 1 fan, that I have mounted at the front of my case somewhere at the middle as an intake with the rad next to the case side and the fan behind the rad. Beneath that I have another 120mm fan also set as intake, but it is in front of a 3-slot HDD cage, very close to it. I have mounted a second 120mm fan at the other side of the HDD cage to pull the air through, so... Finishing out the cooling is another 120mm fan at the back of the case placed as exhaust.

The problem is that when I go into the UEFI, the system shows that the CPU is already at 50oC (122oF) just from being in the UEFI. My ambient temp is around 22oC (72oF), the operating frequency shown in the UEFI is 4.00GHz, I don't know the voltage, however. I also don't know if the BIOS is automatically overclocking my CPU. XMP profile is not engaged. The CPU fan is registered as running at 5000rpm+. All 6 fans from the AiO cooler join together by a special splitter that plugs into the CPU aux fan header for PWM control, with a separate branch connecting to a SATA power cable directly from the PSU to get power. The system has only one 8-pin for CPU power.

Also this system doesn't even have an OS yet, so no background processes to consider.

My question is, with this setup, is 50oC (122oF) a normal temp, and EVEN IF SO, what can I do to lower it. During summer I get ambient temps in excess of 40oC (104oF), and I have no AC, just window draft, so I really need to get those temps down.

And on a more philosophical note, what are my chances of ever getting a decent overclock from this thing sometime down the road.

My system is:
CPU: Intel Core i7-6700K
CPU Cooler: Arctic Liquid Freezer 360
MB: Asrock Z170 Fatal1ty Professional Gaming i7
RAM: DDR4 G.Skill Ripjaws V 32GB (2x16GB) 3200MHz CL 16
GPU: NVidia GTX 980 TI Gigabyte Waterforce x 1
SSD: Patriot Pyro 240GB (maybe 256GB) (for OS)
HDD 1: Toshiba X300 4TB (Storage)
HDD 2: Samsung 500GB
DVD: LG DVD Burner
PSU: EVGA SuperNOVA G2 850W
Case: Thermaltake Versa H35

Thanks.

 

[VirtualLarry]

A two-year lurker? Nice!

Anyways, to answer your question, the CPU is under load (and with NO power-saving in effect) in the UEFI. So, while max "stress load" may be higher, you can sometimes treat the UEFI temps as a form of "load temp".

Chances are, with power-saving enabled (what I mean by that, is "inside Windows"), you will be at a cooler temp at idle, say 23-30C, but shouldn't be a lot higher at load.

That's not to say that possibly the AIO isn't connected right. How is the pump powered? I am using a MasterLiquid 120 on this box, and I have the pump plugged into a CHA_FAN header, with the single fan on the Rad connected to CPU_FAN, so that the fan can regulate with temperature.

You mention SATA power, does that power the pump directly, or does it need a PWM signal from a fan header port on the mobo?

My AIO does not use SATA port for anything.

Btw, is this 50C temp, seen IMMEDIATELY after connecting everything, and powering up? And not an hour later, while the AIO was running?

Because, that could indicate, that either the pump isn't running, OR the cold plate / pump isn't attached to the CPU properly, which would seem more likely, especially if this is the first time that you've ever installed this AIO, or any AIO in general.

 

[rchunter]

My kraken x62 just bit the dust the other day. The pump went out when I was on vacation and the 7700k heat throttled itself off. It's being RMA'd now..

 

[VirtualLarry]

To clarify, assuming the the cold plate is connected properly, even if the pump and the fans aren't running, the liquid already in the system should be enough to absorb a significant amount of heat right away, so you shouldn't really see 50C temps in the EUFI right away, unless your mounting is incorrect.

Credit to one of our other experienced members for pointing that out last time this subject came up.

 

[ovanrakjarac]

Hi VirtualLarry, thanks for your reply.
Regarding my status on this forum, I will follow my grandads advice: Stand up tall and say it loud "I'M A LURKER AND I'M PROUD!!!!"

Now to answer a few of your assumptions, as far as the cold plate, that being the copper round thing at the bottom of the pump casing, I can't tell weather it's placed correctly or not, if I try to wiggle it by hand, it doesn't move, but when I first installed it I tried to tighten it, but after quite a number of turns (8 full turns or more) I still felt no resistance so I just stopped. I didn't want to bend the pins or deform the Mobo. I also don't know if I tightened all four corners evenly, but the notches on top of the screws all line up, so...
Also I may have used a little more thermal paste, it was a rice grain, but say a large one, maybe half the length of the IHS, so I don't know. As to the cooler wiring, my Mobo has TWO CPU fan headers, one main and one back-up, so, since the instructions don't specify, I used the main CPU fan header to connect the pump, I figured since it's the main CPU fan header it would have the pathways to handle the power needs of the pump. Now the fans, there are 6 of them for the cooler, and they ALL connect into one cable line through a special splitter witch at the other end splits into one 3-pin PWM and one SATA-male power connector, so I connected the 3-pin to the remaining CPU fan header for PWM control and the SATA directly to the PSU for power. The fans don't connect to and/or through the pump for control or power. There are no physical switches on the pump itself, and to me it makes no sense to have a water cooler that YOU have to activate by software, I mean by the time you boot into Windows your CPU is fried, so I don't know. And yes, the temperature was immediately 50C, it fluctuated between 48 and 50 for a few seconds but then settled at 50C. I do have to say that my case is pretty full and crowded and I have set the water cooler as exhaust, but all the stuff in the case I need, and I don't feel comfortable unscrewing the fans YET AGAIN (I've done it at least a dozen times so far) because I'm afraid of wearing out the holes they screw into.

 

[rchunter]

If your pump is bad it won't "settle" at 50c, temps will just keep climbing. Somethings wrong though because 50c is too hot for idle temp.

 

[XavierMace]

Anyways, to answer your question, the CPU is under load (and with NO power-saving in effect) in the UEFI. So, while max "stress load" may be higher, you can sometimes treat the UEFI temps as a form of "load temp".

No, not really. It may not be using power saving settings but it's still basically idling. 50c in BIOS is warm. If it's 50c immediately at boot using stock temps/voltages, odds are it's not mounted right. Even if the pump was dead, it would take a few minutes to heat up at idle.

Now to answer a few of your assumptions, as far as the cold plate, that being the copper round thing at the bottom of the pump casing, I can't tell weather it's placed correctly or not, if I try to wiggle it by hand, it doesn't move, but when I first installed it I tried to tighten it, but after quite a number of turns (8 full turns or more) I still felt no resistance so I just stopped. I didn't want to bend the pins or deform the Mobo. I also don't know if I tightened all four corners evenly, but the notches on top of the screws all line up, so...
Also I may have used a little more thermal paste, it was a rice grain, but say a large one, maybe half the length of the IHS, so I don't know. As to the cooler wiring, my Mobo has TWO CPU fan headers, one main and one back-up, so, since the instructions don't specify, I used the main CPU fan header to connect the pump, I figured since it's the main CPU fan header it would have the pathways to handle the power needs of the pump. Now the fans, there are 6 of them for the cooler, and they ALL connect into one cable line through a special splitter witch at the other end splits into one 3-pin PWM and one SATA-male power connector, so I connected the 3-pin to the remaining CPU fan header for PWM control and the SATA directly to the PSU for power. The fans don't connect to and/or through the pump for control or power. There are no physical switches on the pump itself, and to me it makes no sense to have a water cooler that YOU have to activate by software, I mean by the time you boot into Windows your CPU is fried, so I don't know. And yes, the temperature was immediately 50C, it fluctuated between 48 and 50 for a few seconds but then settled at 50C. I do have to say that my case is pretty full and crowded and I have set the water cooler as exhaust, but all the stuff in the case I need, and I don't feel comfortable unscrewing the fans YET AGAIN (I've done it at least a dozen times so far) because I'm afraid of wearing out the holes they screw into.

I'm suspecting it's not tightened down enough, that said the complete lack of documentation on your mobo's two CPU fan ports bothers me. So I'd plug your radiator fan chain into CPU_FAN1 to the top right of your socket and plug the pump into CHA_FAN3 or CHA_FAN2 below the socket and see what effect that has. That would seem to be how Arctic's instructions tell you to hook it up. Assuming that has no effect, unmount the pump and look at the spread of your paste. If the pump was mounted all the way, it should be spread nicely. If there's still a glob in the middle, it wasn't screwed down all the way.

Regarding your summer temps, 104f ambient is really warm. You're going to need a lot of airflow to keep it running cool. I don't know what the stock fans are like on this particular AIO but that's generally the weak point of AIO's. Replacing the stock fans with Noctua iPPC's will get you some gains on the cooling. It's just not cheap.

 

[bbhaag]

Like VirtualLarry mentioned your cpu is running at full speed when in the UEFI/BIOS with little to no power saving states enabled. 50*C is nothing to be worried about. Try loading up an OS then take some temp readings. Take some readings at idle and then some under load. Once you have done this let us know your results.

 

[XavierMace]

Like VirtualLarry mentioned your cpu is running at full speed when in the UEFI/BIOS with little to no power saving states enabled. 50*C is nothing to be worried about. Try loading up an OS then take some temp readings. Take some readings at idle and then some under load. Once you have done this let us know your results.

It might not be a dangerous temp but it's absolutely something to worry about. At boot up, the CPU temp should be roughly ambient temp. According to OP's post, he's at 30c above ambient basically at boot.

 

[bbhaag]

I do not think it is anything to be that concerned about. I am also the proud owner of a 6700K and have been for the past year. I just took a couple pics. One of them is in the OS and the other is in the UEFI. As you can see the difference in temp is considerably different but nothing to be concerned about. That is just the way it is. These temps are using a lowly CRYORIG M9i Mini so the OPs temps on water are fine.

Sorry for the crappy pic. It is harder than you think to get a good UEFI pic.

 

[ovanrakjarac]

Hi everyone,

I'm back with an update about my CPU temp problem. I have tried all the solutions that you have mentioned, i.e. I had switched around the pump cable as well as the cable for all the cooler fans, and I have tried every combination with every fan header I have on my mobo, and I have confirmed that the pump IS operational and spinning at a high rpm. I'm guessing that's good. I have also once removed the pump from the CPU, cleaned both the copper plate and the CPU with nail polish remover and kitchen paper towels and q-tips, reapplied the thermal paste of the same type, I had some left over and it was still pasty, it hadn't hardened. I did notice when I removed the pump that the paste was almost non existent in the center of the CPU's IHS but had a lot of it at the corners of it. I don't know how it is now. This time around I tightened the pump screws with just my hand and I felt resistance as I tightened. Also I made sure to tighten evenly, and I think I succeeded because the notches on the screw heads all line up. As far as voltage is concerned, it is 1.264V, CPU frequency 4,000MHz, but temp is still 49C,50C just being in the UEFI, there was no change. No OS yet.

Now I have a few doubts.

First, could it be that the reading of the temps is wrong because I noticed that I have a pretty old version of the BIOS. Can something like that happen? But I wouldn't want to do a BIOS update until I figure out this temp problem, I wouldn't want the system shutting down on me in the middle of an update. Same with an OS.

Second, I bought my CPU, an Intel Core i7-6700K, from a friend who works for a large desktop retail chain that sells mostly to people, not businesses. He sold me one from his stock cheap. But all the CPUs he has are OEM versions, so could it be possible that Intel is doing some sort of binning, saving good CPUs for retail and putting crappy ones toward their OEM stock, and/or that this is just a case of REALLY BAD LUCK of the "silicone lottery"? BTW, I asked him for another one so I could swap, but he doesn't have them in stock anymore.

P.S. Sorry for the long wait, you know how it is, life and work happened.

 

[BonzaiDuck]

I remember a a C2D 8600 Wolfdale processor for which many users proved that the temperature sensors were "stuck" at around 50C under idle conditions. once you put the CPU under real load and temperatures exceeded 50C, the sensors behaved normally.

It's just hard to imagine that such a thing would occur for a 6700K. I suppose I'd want to see the Tj temperatures for all cores -- not just the package value, although the package value always reports the highest core temperature.

 

[Reinvented]

Hi everyone,

I'm back with an update about my CPU temp problem. I have tried all the solutions that you have mentioned, i.e. I had switched around the pump cable as well as the cable for all the cooler fans, and I have tried every combination with every fan header I have on my mobo, and I have confirmed that the pump IS operational and spinning at a high rpm. I'm guessing that's good. I have also once removed the pump from the CPU, cleaned both the copper plate and the CPU with nail polish remover and kitchen paper towels and q-tips, reapplied the thermal paste of the same type, I had some left over and it was still pasty, it hadn't hardened. I did notice when I removed the pump that the paste was almost non existent in the center of the CPU's IHS but had a lot of it at the corners of it. I don't know how it is now. This time around I tightened the pump screws with just my hand and I felt resistance as I tightened. Also I made sure to tighten evenly, and I think I succeeded because the notches on the screw heads all line up. As far as voltage is concerned, it is 1.264V, CPU frequency 4,000MHz, but temp is still 49C,50C just being in the UEFI, there was no change. No OS yet.

Now I have a few doubts.

First, could it be that the reading of the temps is wrong because I noticed that I have a pretty old version of the BIOS. Can something like that happen? But I wouldn't want to do a BIOS update until I figure out this temp problem, I wouldn't want the system shutting down on me in the middle of an update. Same with an OS.

Second, I bought my CPU, an Intel Core i7-6700K, from a friend who works for a large desktop retail chain that sells mostly to people, not businesses. He sold me one from his stock cheap. But all the CPUs he has are OEM versions, so could it be possible that Intel is doing some sort of binning, saving good CPUs for retail and putting crappy ones toward their OEM stock, and/or that this is just a case of REALLY BAD LUCK of the "silicone lottery"? BTW, I asked him for another one so I could swap, but he doesn't have them in stock anymore.

P.S. Sorry for the long wait, you know how it is, life and work happened.

If anything, it's the AIO and the mounting in combination of the fan speed. Even though there's multiple mounting kits, you still need to make sure you use the right standoffs with backplate.