• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.
  • The forums have been upgraded with support for dark mode. By default it will follow the setting on your system/browser. You may override it by scrolling to the end of the page and clicking the gears icon.

x99 overclocking without increasing idle power consumption

Joined
Nov 27, 2006
Messages
91 (0.01/day)
Location
Slovakia
Processor Core i7 5820k
Motherboard MSI x99s Mpower
Cooling Zalman Reserator 1, XSPC RX120, Koolance QDCs, Koolance MVR-40, EK Supreme HF, EK vga supreme HF
Memory 4x8GB crucial 3000MHz
Video Card(s) R9 290x 4GB with EK fullcover waterblock.
Storage Intel SSD 330 180GB, Samsung HD103UJ 1TB, Seagate ST31500341AS 1,5TB, Seagate ST4000DM000 4TB
Display(s) HP LP3065 30"
Case Silverstone Fortress FT02B
Audio Device(s) ATI HDMI audio device -> Yamaha RX-V1600
Power Supply Corsair AX860 80plus platinum
Software Windows 8.1 prof. x64 eng retail.
Hi.

I'll try to by as brief as possible. I want to buy new computer with Haswell-E and i intend to overclock it. My concern is power consumption in idle. Basically I want to pick for my next rig motherboard witch is is capable to overclock the CPU without rising power consumption at idle. Cadaveca's tests showed that after overclocking the power consumption in idle greatly increases. Do you have any information if overclocking without increasing idle power consumption is possible? If yes, with which motherboard?

FYI. look at my current system specs :) I still use C2D as that platform supported software VID control through 3rd party software like Rightmark CPU clock. So you could manually tune CPU voltage for every frequency step. I know that such software voltage control is not possible on current platforms but maybe there exists a Motherboard witch is flexible enough to setup something similar in UEFI BIOS.

Thanks in advance.
 
Last edited:
depends on how you overclock.

if you overclock via multi while leaving voltages on stock, idle shouldnt go up much.

best bet is to buy a wall meter and measure it yourself as you go.
 
I've done what you describe by using a negative core voltage offset in tandem with a positive boost voltage offset. This has the following impact:
  • Drops voltages at all vid levels via negative core voltage offset, pulling stock freq and idle freqs VID levels lower by the amount of the offset.
  • Increases boost voltage by boost offset with respect to the core voltage offset.
These two things will do the following: Drop idle and non-boost voltage while only increasing boost voltages. This is how I overclock using boost with additional voltage while I still under-volt the rest of the CPU.

Basically, if your non-boost loaded voltage is 1.3v (like mine is,) you can do something like a negative offset of -0.090v which would bring that to 1.21v, but if you still needed 1.3v for boost, you would give boost +0.090v or more if you want the machine to have more umfph under load and sips power while not which is why I like it.

I posted screenshots about this some time ago in another thread: How to save electricity on computer using Windows 8.1

Edit: I should note that I'm not sure how analagous this process is for Haswell or Haswell-E, but I suspect it's not too far off the mark wrt core voltage.
 
Last edited:
Leave power savings on and overclock like normal... No fancy offsets or anything required.

EDIT: At idle voltage and clocks will drop. At load, it will ramp up. Eazy breezy lemon squeezy.

EDIT2: I see this forum caught the VID = Vcore bug too...
 
Last edited:
I've done what you describe by using a negative core voltage offset in tandem with a positive boost voltage offset. This has the following impact:
  • Drops voltages at all vid levels via negative core voltage offset, pulling stock freq and idle freqs VID levels lower by the amount of the offset.
  • Increases boost voltage by boost offset with respect to the core voltage offset.
These two things will do the following: Drop idle and non-boost voltage while only increasing boost voltages. This is how I overclock using boost with additional voltage while I still under-volt the rest of the CPU.

Basically, if your non-boost loaded voltage is 1.3v (like mine is,) you can do something like a negative offset of -0.090v which would bring that to 1.21v, but if you still needed 1.3v for boost, you would give boost +0.090v or more if you want the machine to have more umfph under load and sips power while not which is why I like it.

I posted screenshots about this some time ago in another thread: How to save electricity on computer using Windows 8.1

Edit: I should note that I'm not sure how analagous this process is for Haswell or Haswell-E, but I suspect it's not too far off the mark wrt core voltage.
the most useful post ever!!! thank you man! now i know that asus boards support what i want.
Could anyone examine if this is possible on MSI boards?
 
Leave power savings on and overclock like normal... No fancy offsets or anything required.

EDIT: At idle voltage and clocks will drop. At load, it will ramp up. Eazy breezy lemon squeezy.

EDIT2: I see this forum caught the VID = Vcore bug too...

Oh boy, here we go.

The core voltage is based off of the supplied VID. Offsets apply the difference in voltage that comes off the VID as opposed to manually setting a specific voltage which will force the CPU to use strictly one VID level to maintain that voltage all the time, including at idle, even if the CPU still drops it's frequency to reduce power it won't be able to drop the voltage. The point of using offsets is to lets the CPU under volt itself when it's idle in addition to speedstep dropping the frequency and the offset gets applied to every voltage at every VID level so you can essentially under-volt at all the power states except for the turbo where it can be adjusted to your overclock, at least that's the behavior that my P9X79 gives me.

...but no, I know what the VID is and I'm not implying that it's the same as the core voltage, however the core voltage is directly influenced by the the VID voltage at any given time and it's the VID that determines what voltage the CPU receives and depending on what power state or VID level the CPU is in, that dictates what voltage the CPU gets assuming a manual voltage isn't set.

Next time I'm unclear, let me know, but you don't need to be a mean about it.
 
I don't believe I was mean at all. My apologies regardless! Thanks for clarifying what you meant. I have seen all too many people use VID/Vcore interchangeably and they are not.

I can tell you that, if I just touch the override voltage and the multiplier and leave all power saving/sleep/power management functions alone (auto) in both the bios and windows, I can set whatever voltage and clocks I want, then on idle both the multiplier and the voltage drop. This is how the MSI X99S Gaming 9 AC and the Asrock X99 Fatality Killer both behave.
 
I don't believe I was mean at all. My apologies regardless! Thanks for clarifying what you meant. I have seen all too many people use VID/Vcore interchangeably and they are not.

I can tell you that, if I just touch the override voltage and the multiplier and leave all power saving/sleep/power management functions alone (auto) in both the bios and windows, I can set whatever voltage and clocks I want, then on idle both the multiplier and the voltage drop. This is how the MSI Gaming 9 AC and the Asrock Fatality Killer both behave.

So it's probably a difference between ASUS and MSI boards, granted the MSI board in my gateway gives you control over the VID which is nifty but as you change it, it will change the range of core voltages that are available to you as well.

Either way, the point is that it depends on the board. On mine, you can't set a manual voltage and expect it to under-volt when speedstep kicks in. It just doesn't work like that.
 
Blame ASUS :p. Also note I edited and I am working on X99 boards. It looks like you are using X79 so there may be a difference there too.

...with that thinking, I believe my X79 ASUS board Hero(?) works as I described it (multi + override + not touching power mgmt = idle voltage/multi drops) as well... I do not recall as I leave it at full tilt anyway. If anyone cares I can check tonight...?

@ OP - I would try my method first and see if it works for you. Seems like it is a bit less effort than the other completely valid method Aquinas posted. :)
 
Back
Top