Abnormal Overclock i7 6700K ???I Think???

LilTwist

Reputable
Mar 28, 2014
772
0
5,160
I will start off with the specs first

CPU- i7 6700K Skylake
GPU- Zotac GTX 1070 Amp Edition!
RAM- 2x8 Crucial Ballistix Sport 2400 CL16
MOBO- Gigabyte GA-Z170X-UD3
CPU Cooler- Noctua NH-D15S w/ 2 fans in push pull
PSU- EVGA Supernova 750W B2
Storage- 1Tb Samsung 850 EVO SSD
OS- Windows 10 64bit

Alright, So I am a noob at this overclocking thing and, to tell you the truth, I think I might have lost the silicon lottery on this chip but hopefully I will soon stand corrected. Right now I am only able to get a stable OC of 4.5Ghz at (((WAIT FOR IT)))... 1.4v...?.?.? I was initially trying to go for a 4.6Ghz OC but found out real quick using RealTemp GT with Prime95 and Large FFT's that no more than 5 seconds into it I would get a BSOD. When I do it at 4.5Ghz it runs all the way through both small and large FFT's and surprisingly the max temps only reached 70c at 1.4v. I am just wondering if I am doing something wrong because I see a lot of people going around saying that they have a stable 4.6Ghz, some around 1.236v and others around 1.35v, but mine wont become stable at 4.6Ghz at or bellow 1.4v and I am too scared to push it past that point. Is there something else I need to mess around with in the BIOS to get a stable 4.6Ghz at or lower than 1.4v or am I screwed?

 
Solution
Gigabyte doesn't state these modes. It presumes you know what you doing.
The cpu vcore is where you decide what mode you are in. If you choose your own voltage, then you using manual mode. If you leave it at auto, it is auto/adaptive mode since gigabyte tied them together. If you put normal, then it is in offset mode where is base voltage is 1.2v. You adjust the dynamic vCore to add or subtract from the base 1.2v.
14446820283pfrzHrFEC_3_15_l.gif


GIGABYTE as I said before does use submenus properly, but it also tends to make unnecessary use of submenus in its BIOS layout. The advanced voltage settings menu for example leads into separate voltage menus for the CPU core, chipset, and DRAM...

iamacow

Admirable
Well you want to make sure you have the ring cache/ratio to go up with your CPU multi. So if your CPU Mutli is 44x get, try to make the ring with 44x or close but not above. Also bump the voltage on that ring ratio to get it stable, usually above 43x I have to apply 1.2v for it to become stable. 4.5 isn't the lotto for a Skylake chip, it is however for a Haswell. I've seen Skylake upwards of 4.8ghz @ 1.35v.
 
Are you using prime95 26.6 or higher? Using v26.6 is recommended as version above 26.6 is running AVX with the stress test which require more voltage leading to BSODs and higher vcore.
Chances are you could run your 6700k with 4.5Ghz with 1.3V "stable" with prime 26.6.
You should also run Small FFT for stress testing.
Are you using auto/adaptive, offset or manual mode?
 

LilTwist

Reputable
Mar 28, 2014
772
0
5,160


I don't quite know what you mean by auto/adaptive, offset or manual mode. Like I said "I am a noob at this." Also forgot to mention that all other benchmarking and stress test applications that I run finish without a problem its only the latest version of prime95 ran through realtemp on large FFT's that fails, but when running stock, large FFT's pass without an issue. Even if the latest version of prime95 is faulty, what is the reason it would BSOD? If anything, it adds a different situation my CPU may run into in the future that would make a botched OC fail and BSOD anyways. That's just how I see it, but again, like I said "I am a noob at this" so please guide me through this, I need this knowledge if I want to further my CET college computer courses or to help other people with OCing.

Benchmarking and stress test applications that I use are
Prime95 ran through RealTemp, Large and Small FFT's
Cinebench
Intel Burn Test
Performance Test
wPrime
 
If you are new to overclocking, I would suggest reading up guides on how to overclock and learn what each setting you change does. Don't just go to guides where it tells you to plug in some numbers in one setting without explaining why.
When referring to auto/adaptive, offset or manual mode, it's asking about how the voltage will be applied to your cpu. In short, Auto is essentially letting the cpu/ motherboard handle what voltage is to be given. Offset allow you to add/decrease voltage based on auto. And adaptive mode is applying a defined voltage at turbo speed so it allow you to run "stock" voltage at non-turbo speeds.

As I said, prime version above 26.6 will most likely cause a BSOD due to AVX testing. Whilst you are "stable" at stock speed, you will be running at a higher voltage which means higher temps.
It completing depends how "stable" you want your system to be. If you want to be "AVX" stable, then you do want to use the latest prime95 but you most likely have difficulty getting a decent high overclock of 4.6Ghz. If you use any application that have AVX, then you do want to do a stability test with AVX testing.
Not all benchmarks have AVX support for testing.

 

LilTwist

Reputable
Mar 28, 2014
772
0
5,160
I just ran prime95 version 26.6 Small FFT's at 4.5Ghz at 1.3v and got a BSOD at 3 seconds. Keep in mind I am running Gigabyte UEFI version F6 and I searched through it and couldn't seem to find anything that resembled power options that would allow for auto, adaptive, offset or manual mode.
 
Gigabyte doesn't state these modes. It presumes you know what you doing.
The cpu vcore is where you decide what mode you are in. If you choose your own voltage, then you using manual mode. If you leave it at auto, it is auto/adaptive mode since gigabyte tied them together. If you put normal, then it is in offset mode where is base voltage is 1.2v. You adjust the dynamic vCore to add or subtract from the base 1.2v.
14446820283pfrzHrFEC_3_15_l.gif


GIGABYTE as I said before does use submenus properly, but it also tends to make unnecessary use of submenus in its BIOS layout. The advanced voltage settings menu for example leads into separate voltage menus for the CPU core, chipset, and DRAM. The problem is that each of these categories only has a handful of settings. These could easily have been lumped together to save a few keystrokes in the user experience. The power menu being roughly lumped in with the voltages makes sense to me as the power menu controls power phase behavior. So again there is a logical progression which is normally followed in the design, but that design tends to incorporate more separation of settings than is generally needed.

Despite the huge myriad of choices in the UEFI BIOS, there are times where you don’t have as much granular control as one might like. For example, I dislike how GIGABYTE implements voltage control for the CPU in the BIOS and that is putting it mildly. There isn’t a clearly defined specific adaptive or fixed / override voltage mode as such. These features are supported, but how these features are used isn’t exactly obvious, or it won’t be obvious for many people. You can’t hit enter and see all the values and choices present either. In "auto" mode the voltage is essentially "adaptive" and this is the default configuration. It will scale the voltage automatically in this mode based on CPU load. The choices made by the system with regard to how much voltage it supplies in a given application isn’t always what you want. The software that governs the voltage regulator may output too much or too little voltage. The IVR in Haswell CPUs is known to be overly aggressive and that’s why you never wanted to use adaptive voltage for stress testing programs like Prime95 which put an unrealistic load on the CPU. If you set the voltage to "normal" then it puts the voltage control into offset mode. The voltage is appears to be fixed at 1.2v as a base setting. You can then add voltage via the dynamic vCore (DVID) setting that will light up below it. Any voltage offset you add or subtract seems to be based off the 1.2v setting. However the name implies the voltage mode to be a dynamic and offset combination. In truth, it is a dynamic value with a max of 1.2v combined with whatever offset voltage is added or subtracted from that 1.2v value.

The methods GIGABYTE has chosen for control do work ultimately, but it isn’t the most user friendly way they could have gone about it. Frankly, virtually everyone in the industry handles CPU vCore settings in a better way.
source:http://www.hardocp.com/article/2015/10/14/gigabyte_z170x_gaming_g1_motherboard_review/3

 
Solution

LilTwist

Reputable
Mar 28, 2014
772
0
5,160


That's weird because mine says 1.300V
It seems I obtained a stable OC of 4.5GHz by setting my CPU Vcore to normal and setting my Dynamic Vcore to +0.050V, turned off Turbo, turned off onboard GFX. For now, I am able to run Prime95 Large FFT's without BSODing. The image I posted bellow shows RealTemp with a load at 24.0%??? Why is that when MSI Afterburner and CPUID HWMonitor shows a 100% load??? Also CPUID HWMonitor shows the max voltage it read was 1.363V when obviously I set voltages in the BIOS 1.300V Base and a dynamic of +0.050V making it a max of 1.35V??? Very confusing. But while watching the screen and keeping an eye on the CPU voltage I do realise while stress testing the CPU voltages rarely go above 1.3-1.4V. Am I stable or do you recommend for me to do some more/different testing?

I can't figure out how to post a picture on here so here is a link
http://imageshack.com/a/img923/19/ZpCmQg.png
 
No point stress testing with turbo off. Ignore what realtemp says about cpu load. You only interested with the core temperature with realtemp. Load line calibration will take into account so you won't see exactly 1.35v.
I don't really consider prime95 as a stability, plus you only ran it for a few mins. You should run stress tests for at least 24 hours or more if you you want to be sure you are stable 24/7. Real bench or aida64 is my preferred stability test.
 
From your screenshot, it looks like you just running the sensor test from realtemp. You should run the actual prime95 small fft test within the actual application. Also Vid is different to vcore. Vid is what the cpu request the voltage it wants. You want vcore readings as that is what is the actual voltage given.
 

LilTwist

Reputable
Mar 28, 2014
772
0
5,160
Thank you so much Suztera. Thanks to you I am now able to get a hopefully stable overclock of 4.5GHz at 1.35V. Watching HW Monitor I kind of feel safe bumping up the multiplier to 46 for a 4.6GHz OC seeing that when stressing I don't really see VID going much above 1.293V and CPU Vcore going above 1.332V. I will run Prime95 for a few hours and then I will do it again on Large FFT's just to make sure it's stable on that also. I will also download Real Bench and use that for a few hours to see if it's stable on that as well. Thank you so much Suztera.
 

LilTwist

Reputable
Mar 28, 2014
772
0
5,160
OK so I decided that I will update as I go and I am happy to say that I was able to let small FFT's run for a few hours without any problems. I will now download RealBench and run that for a short while. If all goes well I will up my multiplier to 46 at 1.3V +0.050V and do the same testing and hope it is stable also.
 

LilTwist

Reputable
Mar 28, 2014
772
0
5,160
So I started the stress test with RealBench and I wasn't able to do it with 16GB of ram, RealBench kept crashing, but I was able to start it at 8GB of ram. Does anyone know why RealBench crashes at 16GB?