7700k running hot with Kraken x62

Robert_446

Prominent
Jun 6, 2017
5
0
510
I've been looking at several other forums to find an answer to my hot CPU problem, but nothing has helped. I have been running the Prime95 torture test and getting CPU temperatures at around 90-92 C. I've seen others were able to get as low as 70-75 C. My build is a 7700k with the Kraken x62 aio cooler. I can provide other details about my build as needed. I'm running both CAM and CPUID HWMonitor to check CPU temps/load. I'm in the "turbo" mode set by my motherboard which gets my voltage to ~1.35 V, and 4.6 GHz. I've attached the pump plug to pump aio header on my motherboard, and my fans to the cpu_fan and opt_fan headers, instead of attaching to the pump setup. I have setup my pump to run at 100% as a test and my fans are set to performance mode in the bios. Pictures of both monitoring software are below, please advise.

http://imgur.com/a/A34yO
 
Solution
Bluescreens are most commonly associated with lack of vcore voltage to retain stability or lack of ram voltage when the ram is pushed hard. If, as it seems that you said, you are using the motherboard software OC 'GO' type stuff, that's probably your entire issue, with both heat and instability. Hardware OC via bios is far more reliable once set as you'll only change a few things like vcore, soc, llc etc and disable the eco settings like c-states etc. By using the 'GO' button, you will see a good OC level, but it's generically set, to cover a broad range of cpus, yours may require slightly different settings to be perfect. On top of that, software OC usually also gives a bump to such things as BCLK, which is the buss clock. For Intel...

Lutfij

Titan
Moderator
When you speak of others are they on the exact identical setup as you with the AIO oriented and positioned in the same manner? On top of that what are your ambient temps? If the cooler was installed by you, you may need to reseat the cooling block and see if temps change. Mind you, you're going to need a good thermal compound like Arctic Cooling's MX-4.

Please list your full system's specs as well as your OS.

IMHO those voltages are way too high for a 4.6GHz overclock. What are your temps when the CPU is on stock clocks?
 
Prime95 isn't the most reliable. I do hear the that the newest test runs unreasonable tests. How about you try a realworld test? Like running benchmarks in games, or ASUS Realbench? Or Aida64? You just can't run one test and think your AIO doesn't work. I'm surprised your water is under 30c.
 

Robert_446

Prominent
Jun 6, 2017
5
0
510


Here is my build:
https://pcpartpicker.com/user/huquad/saved/VKH4D3

I've been looking at others with the same cpu and cooler. I understand that there will be some variations between different external conditions, but 90 C still seems really hot. My ambient temps are about 21 C and my idle cpu temps run about 33 C.

I am using the Arctic Silver 5. I understand there is a break-in time, but again I don't think thats the problem.

I've tried reseating the pump three times. The most recent time I redid the entire installation and tightened the standoffs/thumbscrews super tight with pliers and screwdriver. Thermal paste looked like it had a good smear when I reseated the pump the last two times.

Finally, I haven't tried stock clock testing, but I'll go ahead and give it a shot.
 

Karadjgne

Titan
Ambassador
Prime95 version 26.6 was the last reliable version made. Any version newer uses extreme amount of instruction sets like AVX which are not used in 99% of normal applications. These instruction sets will drive an intel cpu of Haswell or newer generation to extreme fits of heat usage, so temps of 90's are not uncommon, but are in no way reliable as you'd almost never, if ever, have any real world usage for such.
http://www.mersenneforum.org/showthread.php?t=15504
With an i7-3770K at 4.9GHz, 8 workers on p95 26.6 small fft after ½hr stress test I get 74°C on a kraken x61.

Oh, and AS5 realistically is junk imho, better to stick with a good paste such as MX4, Noctua NT-H1 or Gelid Extreme 3. The stock paste from nzxt would have been preferable over the AS5.

Go ahead and setup the fans to the pump as per directions or the CAM software won't work correctly, if the fans are set to the cpu_fan headers, they'll be operated by the bios, which is not ideal for the krackens and the pump/fan won't actually spin up to full rated speeds necessary at higher temps.
 

Robert_446

Prominent
Jun 6, 2017
5
0
510


I initially tried running Overwatch and noticed that it was getting around 84 C. That's what started me down this rabbithole. I'll try the other tests you recommended at stock speeds to see the temps from that. Thanks
 

Robert_446

Prominent
Jun 6, 2017
5
0
510


Just ordered the noctua thermal paste as I've heard much better things about it. Also just ran a Asus realbench test under what I thought were stock speeds (CPUID still shows it running at max 4.5GHz). It failed and reported an instability about 8 minutes in. But temps were all below 70 C as shown: http://imgur.com/a/MZl6O
I'll go ahead and flip the plugs around and try the Asus real bench with XMP settings and report back.
 

Karadjgne

Titan
Ambassador
Those are much better looking temps. Asus RealBench is a great program for stability testing as it works the entire pc, all at the same time, but in different areas. P95 is good only for max temp simulation, as it doesn't test other things like a good amount of ram combined with medium cpu and high gpu usage, which is what most games tend to use. All you need now is to figure out the instability, which I'm guessing would be linked to the software OC you applied, instead of using a hardware OC, manually through the bios.
 

Robert_446

Prominent
Jun 6, 2017
5
0
510


I changed the headers so that the pump plugs into the CPU_fan and the fans plug into the pump headers. This should leave CAM in control to my knowledge. Just ran through another test with XMP settings from bios. It blue screened me, so I wasn't able to take picture of the temps. I also noticed that CAM was lagging and could have failed, resulting in the blue screen. Speeds were at 4.6GHz, and temps seemed to be at highest 80 C. I'll reapply the new noctua paste when I get it and go from there as to determine if delidding is for me. Would you mind helping me with the instability issue/blue screen? I have no idea what the source could be.
 

Karadjgne

Titan
Ambassador
Bluescreens are most commonly associated with lack of vcore voltage to retain stability or lack of ram voltage when the ram is pushed hard. If, as it seems that you said, you are using the motherboard software OC 'GO' type stuff, that's probably your entire issue, with both heat and instability. Hardware OC via bios is far more reliable once set as you'll only change a few things like vcore, soc, llc etc and disable the eco settings like c-states etc. By using the 'GO' button, you will see a good OC level, but it's generically set, to cover a broad range of cpus, yours may require slightly different settings to be perfect. On top of that, software OC usually also gives a bump to such things as BCLK, which is the buss clock. For Intel cpu's this is best left at 100.0 but software likes it at 103.7 or 105.2 etc because that affects the cpu multiplier, meaning the multiplier can be at 42, but the OC speed at 105.7 BCLK registers at 4.435GHz. It's a cheat. BCLK has its nose in many things, even transfer speeds for ram and the hdd, so by messing with that, it can mess with everything. Software OC is a gimmick and should be avoided, no matter what any mobo manufacturer claims.
 
Solution