OC: 6700K 4.6GHz@1.225V, Can't Pass XTU Benchmark

[grasspuff]

Clocked my 6700K to 4.6GHz@1.225V and it passed every single stress and benchmarking test that i throw at it EXCEPT for XTU.

 

IBT, Very High@30 loops

Realbench

Cinebench

AIDA64 Stress Test@30 minutes

Memtest

x264 Stability Test (from overclock.net)

 

Is this a cause for concern? Do i need to tweak it more to pass XTU? I read somewhere that XTU Benchmark Test is based on the Prime95 platform, so I'm not sure just how crucial passing the benchmark is 

[scottyseng]

2 hours ago, grasspuff said:

-snip-

It should be fine. Try Intel Burn Test at the maximum preset, it makes your CPU that much hotter.

[grasspuff]

3 hours ago, scottyseng said:

It should be fine. Try Intel Burn Test at the maximum preset, it makes your CPU that much hotter.

I came up with a compromise. Downclocked to 4.5 at 44x uncore and it passed XTU.

 

Now I'm trying to pass prime95 and I've been inching the voltage upwards. WIll update with results

[grasspuff]

6 hours ago, scottyseng said:

It should be fine. Try Intel Burn Test at the maximum preset, it makes your CPU that much hotter.

So finally got a a stable 30 min run on prime95 with 1.26 V. That's a whole 35 mV increase just to stabilize it in prime. So i then ran it again in IBT and temps reached 87C, compared to just 78C at 1.225V. I feel like this is a non-issue(?) as I probably won't be using the CPU anywhere close to how IBT stresses it, but still i need to ask if this OC setting is safe? should i downclock further to shave off a couple of degrees? or ignore stability in prime and just run it 1.225 V? or just run my CPU as it is currently hoping it will never even reach high 80s in temps?

[Juggern0ttt]

Are you on air cooling? 

[grasspuff]

11 minutes ago, Juggern0ttt said:

Are you on air cooling? 

yes, unfortunately. Using a noctua nh-u12s. I'm running it on push-pull though, but with different fans.

[scottyseng]

1 hour ago, grasspuff said:

-snip-

Well, I would say stick with the reduced voltage (1.225v) since it's enough to survive Intel Burn Test. The increase in voltage is too much for stability in just one benchmark. Your CPU should thermal throttle at 87C anyway (You'll see the clock speed drop up and down).

 

It never will reach the temps you get in IBT. Nothing ever gets the CPU as hot as IBT in my experience. My 2500K at 4.6GHz (1.4v) runs IBT at 78C. However, in rendering, it only gets to 60C.

[grasspuff]

46 minutes ago, scottyseng said:

Well, I would say stick with the reduced voltage (1.225v) since it's enough to survive Intel Burn Test. The increase in voltage is too much for stability in just one benchmark. Your CPU should thermal throttle at 87C anyway (You'll see the clock speed drop up and down).

 

It never will reach the temps you get in IBT. Nothing ever gets the CPU as hot as IBT in my experience. My 2500K at 4.6GHz (1.4v) runs IBT at 78C. However, in rendering, it only gets to 60C.

but at 1.225V prime95 runs only a few seconds before it stops. Is this safe to ignore? how do i factor this into my OC?

[scottyseng]

2 minutes ago, grasspuff said:

but at 1.225V prime95 runs only a few seconds before it stops. Is this safe to ignore? how do i factor this into my OC?

Hmm, it's a tough call indeed. At 1.225v, does it survive IBT at maximum? Usually I base my stress tests on IBT. It should be able to pass them all usually though...at least not crash that fast on Prime 95.

[Juggern0ttt]

6 hours ago, grasspuff said:

yes, unfortunately. Using a noctua nh-u12s. I'm running it on push-pull though, but with different fans.

Even on air that seems pretty warm. If you're really trying to push a decent oc maybe look into getting a aio. I've got my 6700k at 1.36v and after 24 hours on aida it never got above 75c same with  ibt 

[grasspuff]

17 hours ago, scottyseng said:

Hmm, it's a tough call indeed. At 1.225v, does it survive IBT at maximum? Usually I base my stress tests on IBT. It should be able to pass them all usually though...at least not crash that fast on Prime 95.

I re-did the stress tessts on IBT an it passed fine at 1.225V 

 

(oh, and I know I wrote 1.225, but actual readings my Vcore shows 1.25V under load since I'm using adaptive voltage. So at my current voltage settings of 1.26V my Vcore jumps to 1.3V(!) under load. Seems a bit much just for a 4.5 GHz OC)

 

Prime95 though, two,maybe three of the "workers" stops a few seconds into the test at 1.225 V. I'm using this as a guide though, http://overclocking.guide/stability-testing-with-prime-95/ 

so it's not "stopping" immediately during a blend test, but rather the 1344K, 30 min test. Quite concerned now with running it on higher voltage since I am on an air cooler, ideally I would like to go down to at least 1.23 V as that gives me temps I can stomach on IBT.

[scottyseng]

18 minutes ago, grasspuff said:

-snip-

hmm, yeah,1.5V is quite scary for under IBT load. 1.3 for regular load is the limit for me for skylake.

 

It should be fine as long as the temps are in check (Might be more noise for you...85C is what I'd call my limit for IBT). IBT is the worst case scenario for the most part, games and even rendering won't get that hot.

 

That 6700K sure does want a lot of power though.

[grasspuff]

16 minutes ago, scottyseng said:

hmm, yeah,1.5V is quite scary for under IBT load. 1.3 for regular load is the limit for me for skylake.

 

It should be fine as long as the temps are in check (Might be more noise for you...85C is what I'd call my limit for IBT). IBT is the worst case scenario for the most part, games and even rendering won't get that hot.

 

That 6700K sure does want a lot of power though.

LOL, typo. i meant to type out 1.25V for the IBT. so adaptive voltage offsets 25mv-ish more than what i keyed in in the UEFI. so 1.225 = 1.5V. 

 

So final thoughts:; stability in IBT ONLY or take prime95 into consideration?

[scottyseng]

34 minutes ago, grasspuff said:

LOL, typo. i meant to type out 1.25V for the IBT. so adaptive voltage offsets 25mv-ish more than what i keyed in in the UEFI. so 1.225 = 1.5V. 

 

So final thoughts:; stability in IBT ONLY or take prime95 into consideration?

hmm, ideally consider both, but to me, if it can survive IBT at the max preset for 20-30 runs, it's good.

 

Prime95 has kind of been a bit of controversy lately because some argue it uses too much power / stress the wrong parts of the CPU (Yeah, I'm not too well versed in the inner workings of CPU voltage regulation).

 

I would roll with it for now and see if you experience any bluescreens.

[grasspuff]

3 hours ago, scottyseng said:

hmm, ideally consider both, but to me, if it can survive IBT at the max preset for 20-30 runs, it's good.

 

Prime95 has kind of been a bit of controversy lately because some argue it uses too much power / stress the wrong parts of the CPU (Yeah, I'm not too well versed in the inner workings of CPU voltage regulation).

 

I would roll with it for now and see if you experience any bluescreens.

Yeah that's what i thought, too. Fuck prime95. Just re-did another IBT run, 20 loops at max @ 1.230 V (added an extra 5mV there just because) and it was stable.

 

Just gonna use it as is for a while and hope its good. Thanks for the feedback, man appreciate it.

[grasspuff]

3 hours ago, scottyseng said:

hmm, ideally consider both, but to me, if it can survive IBT at the max preset for 20-30 runs, it's good.

 

Prime95 has kind of been a bit of controversy lately because some argue it uses too much power / stress the wrong parts of the CPU (Yeah, I'm not too well versed in the inner workings of CPU voltage regulation).

 

I would roll with it for now and see if you experience any bluescreens.

Yeah that's what i thought, too. Fuck prime95. Just re-did another IBT run, 20 loops at max @ 1.230 V (added an extra 5mV there just because) and it was stable.

 

Just gonna use it as is for a while and hope its good. Thanks for the feedback, man appreciate it.

[grasspuff]

3 hours ago, scottyseng said:

hmm, ideally consider both, but to me, if it can survive IBT at the max preset for 20-30 runs, it's good.

 

Prime95 has kind of been a bit of controversy lately because some argue it uses too much power / stress the wrong parts of the CPU (Yeah, I'm not too well versed in the inner workings of CPU voltage regulation).

 

I would roll with it for now and see if you experience any bluescreens.

Yeah that's what i thought, too. Fuck prime95. Just re-did another IBT run, 20 loops at max @ 1.230 V (added an extra 5mV there just because) and it was stable.

 

Just gonna use it as is for a while and hope its good. Thanks for the feedback, man appreciate it.