Question When overclocking I'm having to set voltage high but temps are still low...

[phalkore]

So I'm new to overclocking and just playing around with a few basic settings at the moment (basically multiplier and Vcore).
So my issue is, lets say I set my Vcore at 1.35v and my clock at 5ghz then run either Prime95 or Aida64 my temps stay below 70c but I BSOD. Then I up the Vcore to 1.375 and keep the 5ghz. When I stress again it goes longer and my temps hover around 70c to 73c and again BSOD.

So what I'm confused about is I was under the impression that more Vcore means higher temps and from everything I've seen high temps are always the problem before voltage.
So because my temps are so low I thought I would have some more OC headroom, but voltage is definitely winning the race.
Am I missing something?

Thanks in advance, any info is appreciated!!

My build just in case:
8700k
Aorus Z390 Ultra
Noctura D-15
Corsair DDR4 3200
MSI 1070ti gaming
EVGA 750 psu

 

[AnnoyedGrunt]

The higher you over lock, the more voltage you need to maintain stability. With that higher voltage comes more heat. The exact voltage required to maintain stability will depend on how good your specific processor is. So, if you are crashing you will need to use incementally more voltage until you reach stability or until you are no longer able to keep the chip cool. At that point you would need to lower the overclock.

Typically you would overlock in small increments, gradually increasing the voltage and frequency each time and tracking the temps.

-T

 

[ehume]

Trouble may come from a problem with one or more of the temp sensors. I remember my i7 860 used to report temps that were below ambient. Check ALL the sensors.

 

[phalkore]

The higher you over lock, the more voltage you need to maintain stability. With that higher voltage comes more heat. The exact voltage required to maintain stability will depend on how good your specific processor is. So, if you are crashing you will need to use incementally more voltage until you reach stability or until you are no longer able to keep the chip cool. At that point you would need to lower the overclock.

Typically you would overlock in small increments, gradually increasing the voltage and frequency each time and tracking the temps.

-T

Yeah I understand all that. What I am confused about is I thought It was always heat that would cap your overclock limit and not voltage. I'm not having issues with heat so I thought I could push farther on my overclock but it seems voltage is becoming the issue. So I wasn't sure if there is a setting I could play with to correct this problem. It's just everything I've seen and read it is always high temps that stop an overclock. I've never seen it the other way around. I know delidding will lower temps but will do nothing for voltage right?

 

[phalkore]

Trouble may come from a problem with one or more of the temp sensors. I remember my i7 860 used to report temps that were below ambient. Check ALL the sensors.

When you say check all the sensors, do you mean run a program to check them or actually check them at the motherboard?

 

[ehume]

When you say check all the sensors, do you mean run a program to check them or actually check them at the motherboard?

Compare the core sensor readings. They should be within 10c of each other, and the average should be less than your package temp. The various Monitor Apps will show the individual core temp readings as threy run your benchmarks.

BTW -- a CPU can crap out if the multiplier is too high. Happens a lot with water cooling, where you have to stop raising your OC despite not reaching throttle temps.

 

[DrMrLordX]

Voltage should jack up temps like crazy once you get close to Fmax. If you aren't seeing temp increases associated with voltage increases, it may be due to vdroop of some kind. Your mobo may not be delivering as much voltage as advertised. A Z390 board should not be having that problem, though.

Also while it's possible that temp sensors may be off, the sensors standard to Intel CPUs/platforms are usually pretty accurate. If you were running certain AMD machines then I would suspect problems with your temp sensors. But Intel? Nahhh.

I own a D15 and I can tell you that, as good as it is, it isn't like you're running chilled water or something like that. In those setups (chilled water, dry ice, phase, LN2) you have so much cooling capacity that you may be able to shed +25-50% or more heat without seeing appreciable increases in core or package temps. With air coolers, you will notice it.

 

[ehume]

Voltage should jack up temps like crazy once you get close to Fmax. If you aren't seeing temp increases associated with voltage increases, it may be due to vdroop of some kind. Your mobo may not be delivering as much voltage as advertised. A Z390 board should not be having that problem, though.

Also while it's possible that temp sensors may be off, the sensors standard to Intel CPUs/platforms are usually pretty accurate. If you were running certain AMD machines then I would suspect problems with your temp sensors. But Intel? Nahhh.

I own a D15 and I can tell you that, as good as it is, it isn't like you're running chilled water or something like that. In those setups (chilled water, dry ice, phase, LN2) you have so much cooling capacity that you may be able to shed +25-50% or more heat without seeing appreciable increases in core or package temps. With air coolers, you will notice it.

Amen.

I used to test all kinds of heatsinks (including the D15) on the i7 8700k up to 5 GHz all cores. But then again, I did use Linpack+AVX2 with no reduction for AVX. At 5 GHz I needed a Vcore of 1.37v. The best heatsinks (air coolers) kept the beast under Tmax, but it wasn't 70c. It is just possible that OP had a late-build golden chip, his software did not fully exercise his chip, he did not use AVX2 and/or his ambient is low. Many variables.

I concur that it is not likely that the Intel sensors are off, but one can check them for consistency. My predecessor chip was an i7 4790k that had a core that read 10c cooler than the next.