6700k overclocking help

[sazrocks]

Everyone else seems to overclock with ease. Not me. For some reason I just cant get an overclock to work. I have run 24hour aida64 stress tests, only for the system to crash within minutes of normal operation. I have tried disabling turbo boost and speedstep, but then my overclock do nothing and my CPU just sits there at 4.0GHz. Help? All I want is for 4.6GHz @ 1.345v to be stable.

[Lord Nicoll]

6 minutes ago, sazrocks said:

Everyone else seems to overclock with ease. Not me. For some reason I just cant get an overclock to work. I have run 24hour aida64 stress tests, only for the system to crash within minutes of normal operation. I have tried disabling turbo boost and speedstep, but then my overclock do nothing and my CPU just sits there at 4.0GHz. Help? All I want is for 4.6GHz @ 1.345v to be stable.

Alright, whats your motherboard model, that helps, also get CPUz if you don't have it already as that helps a lot. 

[sazrocks]

Just now, Lord Nicoll said:

Alright, whats your motherboard model, that helps, also get CPUz if you don't have it already as that helps a lot. 

Sorry, tired right now. My entire build is in my sig. Already have cpu-z, hwinfo, etc.

[SquintyG33Rs]

it's very hard to know what you could have done wrong.

[sazrocks]

1 minute ago, SquintyG33Rs said:

it's very hard to know what you could have done wrong.

Is there any way I could help with that?

[Lord Nicoll]

3 minutes ago, sazrocks said:

Sorry, tired right now. My entire build is in my sig. Already have cpu-z, hwinfo, etc.

Ok, I have an Asus Z170-P and my personal board is the Maximus VIII Formula. Does AIDA64 run the CPU at the 4.6GHz OC, or does it stay at 4GHz as you mentioned, it's odd it crashes when you do something normal, so maybe AIDA64 isn't going to the OC state. On my board you need to enable turbo boost for the CPU to overclock as ASUS views it as turbo state, I'm not sure on the Z170 A but it might also do this. 

[SquintyG33Rs]

2 minutes ago, sazrocks said:

Is there any way I could help with that?

well you said you turnned off speedstep and turboboost. in general you don't need to turn off speedstep and turboboost is automatically disabled when you overclock.

(edit) apparently this turboboost behaviour is not the case for ASUS boards, refer to lord nicoll

I'm confused thought, your OC passes a 24h aida test? with no crash? and you stop it manually at the end?

but then it crashes with no load or almost.

[sazrocks]

9 minutes ago, SquintyG33Rs said:

well you said you turnned off speedstep and turboboost. in general you don't need to turn off speedstep and turboboost is automatically disabled when you overclock.

 

I'm confused thought, your OC passes a 24h aida test? with no crash? and you stop it manually at the end?

but then it crashes with no load or almost.

Ok, clearing my mind and focusing a bit here so I can actually give you guys information.

 

Yes, it passed a 24h aida64 stress test. However, as I said earlier, it crashed soon after resuming normal use (nothing strenuous). I believe the cause of this to be that something is noticing the drop in usage and thus dropping the voltages since the cpu is not being utilized much anymore. However, this shouldn't be happening, as I have set the voltage to manual in the bios. HWinfo as reports 1.344 as Vcore, which sounds about right; it could drop for a fraction of a second that I don't have the time to notice in hwinfo though.

9 minutes ago, Lord Nicoll said:

Ok, I have an Asus Z170-P and my personal board is the Maximus VIII Formula. Does AIDA64 run the CPU at the 4.6GHz OC, or does it stay at 4GHz as you mentioned, it's odd it crashes when you do something normal, so maybe AIDA64 isn't going to the OC state. On my board you need to enable turbo boost for the CPU to overclock as ASUS views it as turbo state, I'm not sure on the Z170 A but it might also do this. 

I do seem to need to enable turbo boost to get more than 4.0GHz. The Aida64 stress was done quite a while ago at something like 1.325V, so significantly lower than what I'm at right now. I do remember having cpu-z open during the stress test and that the CPU was at 4.6GHz (this was before I disabled turbo boost). With turbo boost disabled I have only stressed the cpu with the cpu-z stresser, and it still sticks at 4.0GHz.

 

Ugh, I don't have the brain capacity left to deal with this tonight. I'll check back in tomorrow. And sorry about the incoherency/lack of clarity, I'm pretty tired right now and am usually way better than this.

[SquintyG33Rs]

1 minute ago, sazrocks said:

Ok, clearing my mind and focusing a bit here so I can actually give you guys information.

 

Yes, it passed a 24h aida64 stress test. However, as I said earlier, it crashed soon after resuming normal use (nothing strenuous). I believe the cause of this to be that something is noticing the drop in usage and thus dropping the voltages since the cpu is not being utilized much anymore. However, this shouldn't be happening, as I have set the voltage to manual in the bios. HWinfo as reports 1.344 as Vcore, which sounds about right; it could drop for a fraction of a second that I don't have the time to notice in hwinfo though.

I do seem to need to enable turbo boost to get more than 4.0GHz. The Aida64 stress was done quite a while ago at something like 1.325V, so significantly lower than what I'm at right now. I do remember having cpu-z open during the stress test and that the CPU was at 4.6GHz (this was before I disabled turbo boost). With turbo boost disabled I have only stressed the cpu with the cpu-z stresser, and it still sticks at 4.0GHz.

 

Ugh, I don't have the brain capacity left to deal with this tonight. I'll check back in tomorrow.

there is such a thing as a voltage drop causing a stall because of a sudden under voltage in electronics but i don't think that happens on modern computers. they have built in protection.

[Lord Nicoll]

Just now, sazrocks said:

I do seem to need to enable turbo boost to get more than 4.0GHz. The Aida64 stress was done quite a while ago at something like 1.325V, so significantly lower than what I'm at right now. I do remember having cpu-z open during the stress test and that the CPU was at 4.6GHz (this was before I disabled turbo boost). With turbo boost disabled I have only stressed the cpu with the cpu-z stresser, and it still sticks at 4.0GHz.

 

Ugh, I don't have the brain capacity left to deal with this tonight. I'll check back in tomorrow.

Yeah, I have a similar problem right now of an i7 7700K that was fine at 5.2GHz 1.44v now suddenly hitting 100 degrees on 1 core (that before only hit 80) with a delid and custom loop, so you and me both. I don't want to have to reapply thermal paste every week, so I'll just leave it at 5GHz on 1.27v and order some Conductanaught and not have to deal with drying thermal paste, although Antec Formula 7 should do better than this, 

[WereCat]

Go to BIOS and locate LLC (Load Line Calibration), set that to extreme and try if it gets stable.

[WereCat]

14 minutes ago, sazrocks said:

Ok, clearing my mind and focusing a bit here so I can actually give you guys information.

 

Yes, it passed a 24h aida64 stress test. However, as I said earlier, it crashed soon after resuming normal use (nothing strenuous). I believe the cause of this to be that something is noticing the drop in usage and thus dropping the voltages since the cpu is not being utilized much anymore. However, this shouldn't be happening, as I have set the voltage to manual in the bios. HWinfo as reports 1.344 as Vcore, which sounds about right; it could drop for a fraction of a second that I don't have the time to notice in hwinfo though.

I do seem to need to enable turbo boost to get more than 4.0GHz. The Aida64 stress was done quite a while ago at something like 1.325V, so significantly lower than what I'm at right now. I do remember having cpu-z open during the stress test and that the CPU was at 4.6GHz (this was before I disabled turbo boost). With turbo boost disabled I have only stressed the cpu with the cpu-z stresser, and it still sticks at 4.0GHz.

 

Ugh, I don't have the brain capacity left to deal with this tonight. I'll check back in tomorrow. And sorry about the incoherency/lack of clarity, I'm pretty tired right now and am usually way better than this.

Aida64 likes to force CPU to use a bit more voltage during stress tests, that's why it may have been stable 24/7 in Aida64.

However, I never bother stressing more than 20 minutes in Aida64 as from my experience you can pass these test for countless of hours and crash 5 minutes in any game. That's normal.

Still, I would bet that you have voltage droop so go and play with that LLC.

[sazrocks]

12 hours ago, WereCat said:

Go to BIOS and locate LLC (Load Line Calibration), set that to extreme and try if it gets stable.

Dont have an extreme option, only have 1-7, so I worked my way up from 4, and at 6 I cant see any visible droop on the Vcore in HWinfo, so I think its working! Stressing now, will report back. I think you may have solved this all! 

 

Edit: Just to be clear, I'm using the x264 stresser since that stresses harder than aida64.

[sazrocks]

@WereCat Unfortunately it crashed after a few mins of x264. Upping LLC to 7(max) and trying again. For reference CPU is set at 4.6GHz and 1.35V in bios. Temps barely break low 80s. I guess I just didn't win the silicon lottery...

Dang Linus was lucky when he got 4.7GHz on 1.325v.

 

Edit: HWinfo is reporting 1.36V. I guess that's level 7 LLC.

[WereCat]

4 minutes ago, sazrocks said:

@WereCat unfortunately it crashed after a few mins of x264. Upping LLC to 7(max) and trying again. For reference CPU is set at 4.6GHz and 1.35V in bios. Temps barely break low 80s. I guess I just didn't win the silicon lottery...

Dang Linus was lucky when he got 4.7GHz on 1.325v.

 

Edit: HWinfo is reporting 1.36V. I guess that's level 7 LLC.

I can do 4.6 at 1.3V with LLC to max, on default 4.4 is as far as it goes

 

good luck

[sazrocks]

6 minutes ago, WereCat said:

I can do 4.6 at 1.3V with LLC to max, on default 4.4 is as far as it goes

 

good luck

Darn it BSOD after 15 min. Upping voltage to 1.36v. Can't go any higher than that b/c 212 evo.

[WereCat]

3 minutes ago, sazrocks said:

Darn it BSOD after 15 min. Upping voltage to 1.36v. Can't go any higher than that b/c 212 evo.

This will probably not help but bump up VRIN from default 1.80V to 1.85V or 1.9V (it is safe up to about 2V on air, not really necessary to go that high).

[sazrocks]

2 hours ago, WereCat said:

This will probably not help but bump up VRIN from default 1.80V to 1.85V or 1.9V (it is safe up to about 2V on air, not really necessary to go that high).

Where is that setting? I can't find it in my bios.

[WereCat]

Just now, sazrocks said:

Where is that setting? I can't find it in my bios.

Not sure where it is located in your BIOS, it is CPU Input Voltage, check if it is not called that.

It will be 1.8V by default.

[sazrocks]

28 minutes ago, WereCat said:

Not sure where it is located in your BIOS, it is CPU Input Voltage, check if it is not called that.

It will be 1.8V by default.

Dug around for a while, still noting. In other news, it crashed after 2 hours of 1.36v when I turned off the stresser... so on to 1.37ish

[sazrocks]

@WereCat

a

Aaaannd it crashed again. Screw it, I'm dropping down to 4.5 until I can get an nh d15.

 

Edit: Stable 4.5GHz @ 1.35v. Don't care/have time to try lower voltages.

 

[WereCat]

Just now, sazrocks said:

@WereCat

a

Aaaannd it crashed again. Screw it, I'm dropping down to 4.5 until I can get an nh d15.

 

Edit: Stable 4.5GHz @ 1.35v. Don't care/have time to try lower voltages.

 

That's a bummer. Well, at least you now know what to expect from your CPU. I have no idea what mine is capable of as I am now limited by temperatures, I would need to delid.