Intel i7 6700k - temps and voltages

Sep 1, 2018
13
0
10
Here is my build info (sorry for formatting, using Notepad lol):

CPU: i7 6700k Skylake, purchased 2016 https://ark.intel.com/products/88195/Intel-Core-i7-6700K-Processor-8M-Cache-up-to-4_20-GHz
Cooler: Stock Shuttle ICE cooler (Shuttle SZ170R8 v1)
Core speed: 4.0 ghz
Core voltage at load: avg 1.25; 1.359 max w/ Novabench, max 1.421 w/ Intel ETU benchmark
Load test software: Intel ETU (latest version), Novabench 4.0.5
Temperature software: CPUID HWMonitor 1.32.0
Load & idle Core temperatures: idle 31c, load 79c during bench, occasional 83c running No Man's Sky cranked, 84c prime95 v26.6 Small FFTs for 10+ minutes
Memory: Kingston HyperX FURY Black 8GB 2133MHz DDR4 Non-ECC CL14 DIMM Desktop Memory (HX421C14FB/8)
Motherboard: Shuttle FZ170
Graphics Card: ASUS GeForce GTX 1070 8GB Turbo Edition 4K & VR Ready Dual HDMI 2.0 DP 1.4 Auto-Extreme Graphics Card (Turbo-GTX1070-8G)
Ambient temperature: 72-74f? (A/C set at 71f; thermometor on order)

I built this PC in August 2016 and haven't really used it since (30 hours of Skyrim SE, 50 hours of No Man's Sky, mostly off since then, or running a Plex server for a movie here and there). I recently did some thermal tests and the temps were getting near 80c, and I was hoping for 75c so I could absentmindedly play games and not worry about heat. I re-seated the CPU cooler with Noctua NT-H1 thermal paste (using the pea dot method), and that seemed to get me from idle 32c to 28c, but then after stress testing some more, it went back up to 31c.

The real worries I have are the 83c during No Man's Sky (though I think that title just hits the CPU a bit harder than most) and the 1.4v during Intel ETU (I didn't think it would jump past 1.3v!). Am I just being paranoid? I've never really had to troubleshoot heat issues in Shuttle cases before, no matter what I threw at my last one, it never hit 80c. Maybe this is more normal for this CPU, I'm not sure. I can't really find Intel docs that specify that I can ram the CPU to 85c forever without a care, lol.

Anything I can do to make sure I'm good to go?
 
Solution
It seems to me you could have gotten an H or B series motherboard. The voltage options appear to be very limited for that board with a K processor.

orifl1221

Prominent
May 3, 2018
23
0
540
first you should limit your CPU to 1.35V so you dont damage your long term lifespan, second 80c isnt that bad but could be improved by cleaning dust from your pc.
 
Sep 1, 2018
13
0
10


I've been trying to find out how to limit to 1.35v in the bios, but the amibios aptio v2.17.1255 bios only seems to have an overclock feature for watts or something. I'll keep researching - I don't feel comfortable adjusting those values, and documentation is hard to find. I'd rather not keep enabling that in the Intel ETU on launch though.

As for dust, it's nearly dust free - minimal use and the Shuttle ICE cooler only has one radiator to clean, which I did when re-seating the cooler. Airflow seems optimal.

Thank you for confirming the voltage being a concern!
 

orifl1221

Prominent
May 3, 2018
23
0
540
were the temps always like this or did they recently start to increase recently? also, make sure cabling is out of the way for easy air flow inside case, and if you built your pc yourself and not by someone professional try re-applying a good thermal compound, since it might be a problem
 

orifl1221

Prominent
May 3, 2018
23
0
540
I didnt notice before that you are using a stock cpu. switching up to an after-market cooler will help a lot. I recomment something like the Noctua NH-U14S
but it depends on your budget
 
Sep 1, 2018
13
0
10




Agreed! I found a way to limit the VCORE (details in another reply), which seems to be the real issue.

The choice was based on the Shuttle case itself, the motherboard is included. The only limitation there is the seemingly locked BIOS options.
 
Sep 1, 2018
13
0
10




Thank you for replying!

The temps were a tad warmer than I was happy with when I actually used the PC early on, which is part of the reason I stopped using it until I had time to troubleshoot properly. I didn't record them then, though, but I don't think it was any different.

As for the cabling, the case is tiny (13" x 8.5" x 7.7" - here's a spec sheet: dimension drawing), so cables are very challenging to get out of the way. Also, with only one small 80mm fan in the front by the drives, and one 92mm fan in the back, airflow doesn't seem like it's able to be improved easily. The stock ICE cooler had done the job in my last build, and I don't know what else might fit in that case with the GPU and PSU all taking up some space.

I did just re-apply thermal paste and it helped initially with the idle temps.

What I ended up doing earlier this evening was running ThrottleStop to limit the adaptive VCore by -100mv. Since I just can't seem to find documentation on the Aptio AMIBIOS software at all, this utility seems to be able to modify the values without have to be run at startup or in the background.

Undergoing all the same tests, the voltage never hit 1.3v, and I get the same performance as before, while never getting to 69c! I think the BIOS or the CPU might be programmed to over-volt even when turbo and OC are turned off at the BIOS level.

I guess my new question, which may need a new thread in another area, is that is this safe long-term? I had never heard of ThrottleStop before. It seems like the simplest solution though.

Thank you to everyone who's looked and replied! If anyone has any further thoughts, I'm always happy to hear improvements!
 

larrycumming

Prominent
Aug 15, 2018
422
0
410
I was curious enough to look up the manual for this. Nice looking case, BTW.

Have you tried enabling "Overclocking Feature" in BIOS under Frequency & Voltage?

You should be able to set values for VCORE VCCSA VCCIO.

In addition to limiting VCORE, you should make sure your VCCSA and VCCIO aren't overvolting. Some of these boards have a tendancy to overvolt them if left on AUTO.




 
Sep 1, 2018
13
0
10


Hey Larry, thank you for helping me look into this!

I did look at the Overclocking feature, but after enabling it, the data entry fields don't seem to allow negatives or many characters, it was more of a single-digit entry. It seemed like I was supposed to enter a 1, 2 or a 3, etc, but I could not confirm. I even contacted Shuttle support to see if they had answers, and they have no more insight than I did.

Once you enable the OC feature, you get a core, power level 1 and power level 2 option - with the same text entry fields. I was looking to put in -100mv or something there, but it just doesn't seem to allow that. It just wants to over, not under.

Thank you for the compliments on the case - it's super easy to set up and with the right 92mm fan in the back, it's quiet as a mouse. This odd OC option is the only problem I've ever had, so maybe Shuttle has spoiled me a bit.

I can snap a photo of the OC options if you would like to see them - they don't really match any of the online Aptio documentation I ran across. Those versions all seemed to have useful options.

For future researchers, Aptio v2.17.1255 is not a lot of fun to underclock! And I don't know that I'd trust it for overclocking, either!
 

larrycumming

Prominent
Aug 15, 2018
422
0
410
It seems to me you could have gotten an H or B series motherboard. The voltage options appear to be very limited for that board with a K processor.



 
Solution

DanKem06

Reputable
Feb 24, 2014
192
0
4,710
30c at idle is normal. But something isnt right here man. You ran a stress test for how long? And got 30s but turn on no man sky and it jumps to 80?

How close is the GPU to the CPU? Can you shield them and see if that helps? Just for like an hour place something between them. If that actually helps it's your case design.

Sounds like you already did new paste and cooler. Dot method is fine, I used to do that but realised I wasnt using enough and missing the edges. Now I use a cardboard cutout and kinda paint it on. I also wipe any excess compound away.


 
Sep 1, 2018
13
0
10


Possibly there are options when purchasing the Shuttle cases, but I didn't see any or recognize them at the time. I'll take a look at the H & B and make sure I'm better equipped on my next purchase.

Thanks again!!
 
Sep 1, 2018
13
0
10



Hey DanKem06,

I'm afraid I don't understand what you're referring to by "you ran a stress test for how long? And got 30s but turn on no man sky and it jumps to 80?" For the idle, new thermal paste lowered that temp but it went back up a few degrees, but the "30s" doesn't make sense otherwise. Can you be more specific there? Stress tests were getting up to 83c and 84c (counting No Man's Sky as a heat test for the CPU). Now they're at 68c.

The case is small, but if I see more temp issues, I can give the shielding a shot. After ThrottleStop, I haven't seen any issues. As for the thermal paste dot method, I used to use a business card to paint it on, but then I saw this video and it convinced me to stop doing that: https://www.youtube.com/watch?v=-hNgFNH7zhQ

Thank you!
 

DanKem06

Reputable
Feb 24, 2014
192
0
4,710
I may have my numbers mixed up but prime95 aka "stress test". You are stressing the CPU. Sorry for the terminalogy.

Forget what I said above.

Without really getting into your machine it is possible No Man's Sky is stressing your GPU and in turn it gets hot, heat then moves to your CPU. Think of a laptop, smaller spaces do not dissipate heat as well as a large better ventilated space. That's why I suggest shielding the GPU and maybe they to vent the heat away from the processor.
 
Sep 1, 2018
13
0
10


Understood, and I'm sure heat transfer like that does happen. Thank you for your suggestions!
 

larrycumming

Prominent
Aug 15, 2018
422
0
410


The typical GPU heats up to about 70C max after a few minutes of 100% utilization. If you don't overclock them they might run in the 60Cs in stressful situations. It also depends on whether the GPU has 1, 2 or 3 fans on its heatsink. Overclocked or factory OC'd graphics cards will run hotter because of increased power usage but I don't think you are doing that with your card.

There is no direct contact between the CPU and GPU so it shouldn't transfer the heat and even if it did, it wont make your CPU overheat to 80C+.

I think you just need a better cpu cooler. There should be plenty that support your case.


 
Sep 1, 2018
13
0
10



Sorry for the very late reply and to bring up an older thread, but I wanted to close it out on a good note. I didn't know if it was OK to post my own answer, and that's the only other option I had.

This took a while to settle and I've tried out nearly all recommendations - Shuttle Support couldn't answer the over/underclocking with the BIOS above, I can barely fit my hands in the case let alone fit a bigger cooler, and blocking off air between the GPU and CPU didn't have an effect (the GPU is pretty much hogging the entire case anyway). And the difference in workloads (stress tests vs. No Man's Sky) heat values never budged through any of those tests.

So, the only resolution I've found is using ThrottleStop to undervolt (mentioned in a previous reply above, and I documented the settings used separately here) because the "K" series Intel CPUs are just unlocked and will crank up the voltage.

I have thrown everything I could at this machine now, with reasonable settings (usually Ultra but I do tone some down), and the GPU is no higher than 69c (with custom fan curves to ramp the GPU fan based on the heat level) and the CPU sits in the 50's with Prime95's Small FFTs test maxing out all the cores for a bit. It hasn't gone past 68c no matter what I've tried to do to it. I only run HWMonitor once in a while just to check now.

So, to sum up, "K" series Intel i7's are truly unlocked in some cases, and you can trust ThrottleStop to cap them safely (again, I only used this to undervolt by -100mV, I left everything else alone). Beware of BIOS settings, and if you apply them, make sure to test. And test again.

Thanks again to everyone who replied and assisted! I'm back to blissfully spot-checking my temps when I'm curious!