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

Overclocking 7700k. Frustration...

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

14MTH30n3

New Member
Joined
Apr 10, 2017
I am overclocking 7700k delidded. I passed 2 hours test with RealBench with 5Gh @ 1.4v. Temperatures peaking at 68C.

However, with Prime95 I always fail. The computer just reboots. I looked at the blue screen error and it is DPC_WATCHDOG_VIOLATION. I have all latest drivers, so that is not it. Just cannot figure this out.

i7-7700k (delidded), OC to 5Gh @ 1.4v
ASUS Rog Maximus IX Hero
Corsair H100i v2 Extreme Cooling
EVGA GTX GeForce 1080FTW
G.Skill Ripjaws 4 series 288pin DDR4 SDRAM 2400 (PC4 19200) 4x8 32Gb
Samsung 960 EVO Series 1Tb
EVGA Supernova 850 G2
 
Prime95 is harder to pass than Realbench when overclocked. There seem to be two camps. One that thinks Prime95 type usage will never be seen in real world usage so they don't use it to stress test their OCs. They use easier to pass tests like Realbench, AIDA, IBT, XTU, x264. The other camp wants their CPU to be 100% stable no matter what, if a stock CPU can pass it than an OCed CPU should be able to.
 
Explained in a bit more detail : it all revolves around which type of instructions you use for your day to day, Prime95 tests them all (FFT/AVX/AVX2/FMA3) while the other only test a few (the most widely used, FFT/AVX). Testing for all translates in more voltage/heat to pass the test. If you only game/web browse/listen to music etc, Realbench, Aida64, XTU is more then enough, if you do x265 video editing or more serious work like maths/sciences/fold etc you want Prime95 to make sure you're 100% stable.
 
Ive seen prime fail @ 25 hours.. I like IBT tho. If it doesn't pass that, then its not stable lol :D

I like it because its not the easiest thing to to pass, and it lets you know if that cooler you bought is really worth its beans.
 
IBT to my knowledge only uses FFT/AVX and is sorely outdated (2012), but to me it was a hit and miss, pass-relog-fail-relog-pass-relog-fail, it never seemed to give me assurance of being stable. Realbench is good and it also tests the GPU but lacks the capacity to tell you what went wrong when it fails. First and only time I used XTU (couple years ago) it was considered a joke and worked like one. OCCT has good rep. Never used x264 but all the other stress tests do the same thing (test for AVX) so meh. Aida64 is a curious case, I don't think it tests for AVX, only FFT, therefore lower voltages/temps and also seems to require more time to make sure you have a decently stable overclock (6h+).

Prime95 so far has been the only one that allowed me to get rock solid overclocks for the instruction types I specifically want.
 
OCCT has an option to use the Linpack library that IBT uses with the addition of AVX1 if you check that option.
 
Last edited:
I am overclocking 7700k delidded. I passed 2 hours test with RealBench with 5Gh @ 1.4v. Temperatures peaking at 68C.

However, with Prime95 I always fail. The computer just reboots. I looked at the blue screen error and it is DPC_WATCHDOG_VIOLATION. I have all latest drivers, so that is not it. Just cannot figure this out.

i7-7700k (delidded), OC to 5Gh @ 1.4v
ASUS Rog Maximus IX Hero
Corsair H100i v2 Extreme Cooling
EVGA GTX GeForce 1080FTW
G.Skill Ripjaws 4 series 288pin DDR4 SDRAM 2400 (PC4 19200) 4x8 32Gb
Samsung 960 EVO Series 1Tb
EVGA Supernova 850 G2

Have you tried more Vcore 1.44v

As of 2/22/17, the top 59% of tested 7700Ks were able to hit 5.0GHz or greater. https://siliconlottery.com/collections/lga-1151/products/7700k50g
 
I continue to run stress test using RealBench. XMP enabled now. 4 hours passed, still 5Gh @ 1.40v @ 70C max temp

Question - does it make sense to push it past 1.4v to hit 5.1? Is temperature the only thing I need to be concerned about?
 
I continue to run stress test using RealBench. XMP enabled now. 4 hours passed, still 5Gh @ 1.40v @ 70C max temp

Question - does it make sense to push it past 1.4v to hit 5.1? Is temperature the only thing I need to be concerned about?

Short answer is no, for the added gain in performance it's not worth the stress to the CPU for long term 24/7 usage.
Temperature is only one half of the equation. Higher voltages will damage the inside of the CPU over time almost like erosion ( electron migration). The higher the voltage the faster it "erodes". Keeping the voltages within reason will extend the life of your CPU. You're still within recommended ranges so I would just leave it that way and get with your putering.
 
All depends on what you expect but as Johan said, it's not worth to push CPU to the limits for 24/7 use. In most applications there will be barely any performance gain while CPU may degrade much faster and no one will tell you how fast. It's too new processor and no one made any tests to check that. Funny is that even most specs made by Intel were not tested for 24/7 use. They are usually telling not to pass some values but it's not always based on any longer tests.

Personally I try to keep my processors at about 1.2V 24/7 while for quick tests I set up to 1.8V+ ( but on sub 0 temps ). Even at low voltages you may see some degradation after longer use. It's maybe not KL but recently I've noticed that my 6800K is losing stability. It was running at 4.2GHz 1.22V for about 4 months and in last days I had to set 1.25V because of random blue screens.
 
Intel does not test like one would think to achieve the specifications on their processors that the engineers make, they use estimates and simulations or empirical data. Intel knows what they are doing when making the processor and specifications.
Intel's specifications are specific for stock operation.

Note 1 below.

1.52Vcore notes.JPG
1.52 Vcore Max..JPG
 
Back