CPU input voltage

[lexidobe]

Hello everyone, I have been folding for the last month and I am wondering if I can reduce power consumption and heat by reducing my Core i7-6850K's input voltage. I'm running a -0.08 volt offset on my core voltage and I set my input voltage to 1.5 volts instead of the default 1.8 volts, although I can likely go a bit lower if that would help me accomplish my goal of reducing power consumption and heat. I have good cooling, but every little bit helps especially when it is less than free. I know that decreasing the core voltage will decrease power draw and therefore heat output, but is the same true for input voltage as well? Or will the CPU's power consumption (wattage) stay the same and current draw increase due to the lower voltage being supplied to the CPU? Not much info is out there on the input voltage, most of the info out there is the core voltage which I already know about. Thanks in advance!

[Fasauceome]

Lower voltage will always reduce thermals

8 minutes ago, lexidobe said:

default 1.8 volts

1.8 volts is a pretty insane amount of voltage for a CPU, are you totally sure that's what it was working with?

[lexidobe]

1 minute ago, Fasauceome said:

Lower voltage will always reduce thermals

1.8 volts is a pretty insane amount of voltage for a CPU, are you totally sure that's what it was working with?

Yup, that what my BIOS says the default CPU input voltage (VCCIN) is. The core voltage (VCore) was around 1.14 volts stock. I think you are confusing the input voltage and the core voltage. The input voltage is the voltage the motherboard's VRMs supply to the CPU, the core voltage is the amount of voltage the CPU's internal voltage regulators supply to the cores. 

[Fasauceome]

17 minutes ago, lexidobe said:

Yup, that what my BIOS says the default CPU input voltage (VCCIN) is. The core voltage (VCore) was around 1.14 volts stock. I think you are confusing the input voltage and the core voltage

Yeah I was thinking of vcore

[lexidobe]

On 4/10/2020 at 2:28 PM, Fasauceome said:

Yeah I was thinking of vcore

Do you happen to know about vccin? So far I have been able to get mine down from the stock 1.8 volts to 1.45 volts and the system is still perfectly stable and I am wondering if lower vccin means less heat and power draw. Thanks 

[Fasauceome]

8 hours ago, lexidobe said:

Do you happen to know about vccin? So far I have been able to get mine down from the stock 1.8 volts to 1.45 volts and the system is still perfectly stable and I am wondering if lower vccin means less heat and power draw. Thanks 

Lower vccin means lower vrm temps, and that'll improve vcore performance because less heat = less resistance, but as far as I'm aware, the impact is fairly minimal, especially if you've got a motherboard with a great vrm solution. What motherboard are you working with?

[lexidobe]

2 hours ago, Fasauceome said:

Lower vccin means lower vrm temps, and that'll improve vcore performance because less heat = less resistance, but as far as I'm aware, the impact is fairly minimal, especially if you've got a motherboard with a great vrm solution. What motherboard are you working with?

Thank you for explaining. Will the CPU draw more current due to the lower input voltage, or will current draw stay the same? My concern is if the CPU's input voltage drops but its power consumption doesn't drop a proportionate amount more current will be drawn since volts X amps = Watts, and if more current is being drawn that would be counterproductive since the current draw through components is what creates heat. Thanks 

[Fasauceome]

55 minutes ago, lexidobe said:

My concern is if the CPU's input voltage drops but its power consumption doesn't drop a proportionate amount more current will be drawn since volts X amps = Watts, and if more current is being drawn that would be counterproductive since the current draw through components is what creates heat

On a very basic level, current is proportional to clock speeds and voltage is dynamic based on BIOS settings. Lower voltage reduces power consumption, as opposed to power consumption being compensated for by higher current. Undervolting always reduces power draw, hence why it's very commonly recommended for laptops with no room to breathe and Vega graphics cards.

[lexidobe]

2 hours ago, Fasauceome said:

On a very basic level, current is proportional to clock speeds and voltage is dynamic based on BIOS settings. Lower voltage reduces power consumption, as opposed to power consumption being compensated for by higher current. Undervolting always reduces power draw, hence why it's very commonly recommended for laptops with no room to breathe and Vega graphics cards.

Thank you for your help, I will reduce input voltage as much as I can then. Since current draw is proportional to clock speeds, if I reduce input voltage by 10% then I should see a 10% decrease in the wattage the CPU draws, correct? Or does that only apply to the core voltage? 

[Fasauceome]

1 minute ago, lexidobe said:

if I reduce input voltage by 10% then I should see a 10% decrease in the wattage the CPU draws, correct? Or does that only apply to the core voltage? 

There are more factors behind the scenes, I don't believe it's an exact reduction of power draw but it would be substantial to be sure. Things like IMC voltage matter here too.

[lexidobe]

10 hours ago, Fasauceome said:

There are more factors behind the scenes, I don't believe it's an exact reduction of power draw but it would be substantial to be sure. Things like IMC voltage matter here too.

That makes sense, thank you. I will keep the voltages at the minimum level possible then. 

[lexidobe]

On 4/12/2020 at 2:09 PM, Fasauceome said:

There are more factors behind the scenes, I don't believe it's an exact reduction of power draw but it would be substantial to be sure. Things like IMC voltage matter here too.

I did a quick test and I found that with an input voltage of 1.9V, my i7-6850K runs at around 48 degrees under load at 3.8 GHz with a core voltage of 1.05V and the system draws 222W from the wall (0.93A X 239V = 222W). With an input voltage of 1.42V the temp is the same but power consumption is down by 12 watts (0.88A X 239V = 210W). So from my testing the input voltage does affect power draw and therefore heat output, but not nearly as much as the core voltage does. 

[Fasauceome]

1 minute ago, lexidobe said:

I did a quick test and I found that with an input voltage of 1.9V, my i7-6850K runs at around 48 degrees under load at 3.8 GHz with a core voltage of 1.05V and the system draws 222W from the wall (0.93A X 239V = 222W). With an input voltage of 1.42V the temp is the same but power consumption is down by 12 watts (0.88A X 239V = 210W). So from my testing the input voltage does affect power draw and therefore heat output, but not nearly as much as the core voltage does. 

I was just messing with a 5960X on an X99-E WS and the system was not happy with an input voltage of less than 1.9. That said, I was able to get a sub 1V vcore for a 3.9GHz OC. Hope you can get the same, but this does seem to be some lucky silicon.

[lexidobe]

Just now, Fasauceome said:

I was just messing with a 5960X on an X99-E WS and the system was not happy with an input voltage of less than 1.9. That said, I was able to get a sub 1V vcore for a 3.9GHz OC. Hope you can get the same, but this does seem to be some lucky silicon.

That's interesting, thanks for sharing. My 6850K needs 1.05V to run at 3.8 GHz, but I think that's pretty decent considering Broadwell-E is known for not overclocking nearly as well as Haswell-E chips did. I was surprised that I was able to get the input voltage down so low, I dropped it by 0.38V below the stock 1.8V and it is perfectly stable.