• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

6700k has odd overclocking behavior, need help!

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

kaiff75

Registered
Joined
Oct 16, 2016
hey! just got my 6700k a few weeks ago and ive been messing around with overclocking. im having a bit of trouble and coming from the asus bios to the new gigabyte bios is a bit confusing. basically i have only found 2 clocks that are stable. i can either set the multiplier to 42 @ 1.27v for a full time 4.2GHz stable or set the multiplier to 45 with voltage on auto for a 4.5GHz @ 1.32-1.38V stable. However if i try to manually set the exact same voltages as auto does with a multiplier of 45 it BSOD consistently. i have also tried lowering the voltage to see if that helps but it doesnt and i really dont wanna try and run 1.4v 24/7. Anyone who has a gigabyte board know what setting is for the voltage offset? why is it working on auto and not manual? did i just get a crappy chip? or is it possible my xmp profile for my ram is causing issues? ive also tried switching off c states, turbo boost and HT with no difference. cooling shouldnt be an issue with my h110i as it peaks at 75c on 100% load but usually hovers around 65c during prime95 and runs a little cooler during realbench. Gigabyte seems a little slow on the bios updates as im currently on the factory f2 and the new f3a is still in beta testing....if there are other bios updates im not seeing let me know, but these are the only 2 on the gigabyte site for my mobo. Dunno if that would help or not. any help would be greatly appreciated! thanks for reading!

i7 6700k (currently the 4.5GHz oc),gigabyte z170x designare, 32GB(4x8) ddr4 gskil ripjaws v (xmp 3200MHz), corsair h110i aio, corsair hx850i psu, asus 1070 strix, samsung 850 evo raid 0, windows 10 home 64bit

- - - Updated - - -

forgot to add, any other clock speed from 4.3-4.4 and 4.6 and up is not possible with or without settings being on auto or manual. doesnt make any sense.
 
No i havnt yet, i see you have basically the same setup and were able to get it to work. are you using xmp or manual settings for your ram?
 
No i havnt yet, i see you have basically the same setup and were able to get it to work. are you using xmp or manual settings for your ram?

Xmp, but I would overclock you cpu with stock ram values first.
 
so i just gave it a try, stock 2133 ram with cpu 45 multiplier at 1.33v and instantly had a bsod, never even made it to the home screen. maybe i need to start with the 2133 ram and work my way up from the bottom for the cpu as voltage requirements might have changed. if this ram is really the cause i just wasted extra money on oc ram..... just weird that it only works at 4.2 or 4.5GHz so far, i cant figure that part.
 
Your board has a dual bios, right? Flash to that beta and see if it helps.

I believe in our review of that board, we didn't have any trouble with it on that bios.

Voltage is good to 1.4V+ so for giggles jack it up there and see if it doesn't bsod. Also what is the bsod error?
 
the error code is always CLOCK_WATCHDOG_TIMEOUT. ill have to give the bios update a try, i was just hesitant to try a beta version. i am a bit worried about going to 1.4+ though . i have an h110i with noctua nf a14 ippc 3000 fans(loud as hell but they push air like crazy) and even at 4.5GHz @ 1.32-1.38 i get pretty close to touching 80c at short spikes which makes me a bit uncomfortable but usually hovers around 65c. im wondering if my cooler isnt seated correctly with those fluctuations.
 
ive tried lowering the ram clock but so far havnt achieved anything stable, i could still tinker some more though. i do have a second 6700k as my original shipment was stolen, i purchased a new one then the shipper decided they would send me a replacement for the original stolen one. so i could try the other chip but then selling it would have decreased value. havnt decided whether its worth the try.
 
Ram clock? Why? How is yours, Neon, at 4.0 (what is that???!)??? I would leave the memory on the stock 2133 Mhz and work on the CPU clock... one thing at a time!!!!!!!! Right now, I wouldn't touch the memory except to make sure its at 2133 Mhz.

You need to keep temps below 90C while stress testing. The CPU will throttle and protect itself at 100C so you have NOTHING to worry about, particularly only hitting 80C. This is just a test to see if the CPU boots with more voltage... it doesn't mean we are going to leave it there. ;)
 
I'd start baby steps to figure out the issue. If you got it to work at 4.2Ghz... up the voltage til 4.3Ghz is stable. Continue to 4.4Ghz and 4.5Ghz.
Make sure your Cache Speed / Ring Memory / Uncore (which ever it is) is set at or below your maximum multiplier. Keep it at 40 or 41 multiplier which I believe is the default speed.

Keep ram as stated at default speeds (2133) just to eliminate that issue. If anything maybe pull 2 sticks of ram just to eliminate even more potential issues.
 
So I think im starting to get somewhere finally but still need some tweaking. So far ive found that changing ram to stock speeds or manually overclocking it made no difference at all. ATM i have it at 4.6GHz @1.4v which is a little higher than id like. it passed 30 minutes of realbench and several consecutive runs of cinebench no issues but its not happy when i try prime95. i did notice however that in both cpu-z and hwmonitor my reported vcore was 1.28v under 100% load and 1.35v idle. is this the vdroop ive been reading about? if thats the case i have plenty of voltage headroom left if thermals allow but id like to make sure it isnt a fluke and my cpu is actually recieving the full 1.4v i entered in bios (i hope not, id rather stay at 1.38 or lower)
 
Try to find what speed is stable at 1.35v 1st, that will give you the starting place to overclock and a temperature baseline. Ring cache defaults at 4.1ghz so it SHOULDN'T be an issue and try setting LLC to 4-5 to stop vdroop :)
 
thats part of the problem, hwmonitor and cpu-z are reporting 1.356v at idle and i entered 1.4v in the bios. going into bios and setting it back to 1.35v at this point would only give me 4.3-4.4 GHz. will changing the load line to 4 negate the .05v vdroop? and in that case i would basically be changing the llc and restarting my oc process? sorry, alot of this is all new to me, i came from amd and it was just a simple multiplier/voltage/test process for that.
 
Sweet! Thank you! LLC was the ticket! I need to do more testing, but so far @ 1.35v I achieved 4.7GHz and passed several runs of cinebench and 30 min of realbench. Temps did peak at 85c though which makes me a bit uncomfortable although I know its theoretically fine. I do have an h110i AIO so does that even sound right? Gigabyte didn't have number values for LLC and instead used "standard, high, or turbo." I needed turbo for it to be stable. 4.8 required over 1.38v which I'm not comfortable with yet and would likely have poor thermals anyways. Now I suppose I could mess with offsets to help temps. Is it possible to only allow negative offset to lower voltage during load? Gigabyte doesn't have an "offset" option in bios and chose to name it something different and idk which it is. Anyone familiar with Gigabyte bios? EDIT: I am idling in the mid 30s so that seems fairly normal.
 
Load temps are a tad high but overall a good OC :) try to keep it below 80c, even if you have to bring it down a notch, I run 4.6ghz @1.3v daily and it rarely hits 65c with fans on silent ;)
 
85C is fine... many say stay under 90c while stress testing. The chip throttles itself at 100C... and games etc won't come close to P95 temps...
 
OK thats good. Is it possible to set an offset for negative voltage only though? I noticed the VID request was only 1.26v under full load and all that extra voltage is unnecessary heat.
 
You can do that, or manually, sure. I don't like offset (simply a personal preference) so I would use manual. Then again, stock speed, it's not making squat for heat anyway.
 
Back