lowest vcore

Soldato
Joined
22 Mar 2009
Posts
7,754
Location
Cornwall
i have read a lot about if you go over 1.4 vcore then you can damage the chip, but is this the same for lowering the vcore?
i still have a stock cooler so im trying to get the temps down. on stock settings under load it was going over 80deg, i lowered the vcore to 1.10 and it dropped to 77deg under load. i then bought 2 8mm case fans for the rear of my pc and got it to 74deg under load. so i then lowered vcore to 1.00 and its now 72deg under load. dont want to go down further if its going to cause problems or damage my chip.
 
Nope. If you lower it too far it'll become unstable, needing you to lower clock speeds to regain stability. So stress test when you lower the voltage to check it's still stable.

Undervolting is popular with people who like quiet. Undervolting with overclocking can be done if you've got a particularly good chip, but it'll never go as fast as overvolting with overclocking.

You need to get rid of the stock cooler basically.
 
You need to get rid of the stock cooler basically.

yeah, thing is i have a small case so im limited on what i can get, and it only has 2 8mm case fan holes (i think, the are some holes that are slightly further out, so not sure if a 12mm fan is supposed to bolt to them) so potentially that rules out the H50.
 
you mean 80mm, right? The next step up is 92mm. By two holes, do you mean two places to mount fans?

Measuring mine, if the holes are 102mm apart diagonally, that's 80mm fans. If 118mm apart diagonally, that's 92mm fans. If 147mm apart diagonally, that's 120mm fan. Accurate to +/-1mm or so, just get a ruler out and check your one.

If the two fans are right next to each other, I've got a radiator that takes two 92mm fans with a fan spacing of 10mm, i.e. the distance between the holes closest to each other is 10mm. If yours happens to have this spacing you could fit a water cooling loop without modification.
 
yeah i meant there are 2 places to mount the fans, 92mm :( so that rules out the H50. dont want to do wc as i dont know enough about it and dont want to spend a fortune on it.
 
Related to this, is there any way to check VID on the i3/i5/i7 chips? Might be a good reference point at where voltages are 'recommended', and working down from there.

I'm aware that the VID is all over the place with these chips due to load and idle volts, however there should be something right?

From what I can see, CPU-Z and Everest don't show VID on the newer CPUs.
 
Vid is the voltage it requests from the bios when at stock settings. The bios probably shows you a value for this, otherwise turn off load line calibration or on vdroop to stop these interfering and boot with otherwise stock settings. Cpu-z reported vcore at idle is taken as your vid.

Not that vid means very much, as the correlation between low vid and overclocking potential seems to be pretty weak. It'll probably be somewhere in the range of 0.9 to 1.3V.
 
a side question to under volting, forgive me if this sounds silly; aren't you causing a higher current draw for a given frequency and thus increasing the electron migration process?
 
I never thought about current increasing, but would it necessarily? I mean resistance increases as temperature increases, but if you undervolt temp drops so shouldn't resistance should drop? Or have I got that backwards?
 
clearly not a uni student quoting the 'pedia :D

but yeah, i would've thought at the same frequency, if the voltage is lowered the current draw would increase

anyone care to elaborate/correct?
 
Increasing the voltage applied to anything increases the current running through it at a given temperature. If this higher voltage (and consequent higher current) raises the temperature, and if the resistance rises as a consequence, then the current will decrease. If resistance increases fast enough with temperature you can get into a position where higher voltage results in less current.

The resistance of copper increases with temperature. The resistance of doped semiconductor decreases with temperature (the resistance of a superconductor is zero below a critical temperature, and greater than zero above this temperature). Semiconductor can show positive feedback, where current makes it hotter, higher temp lowers resistance causing higher currents making it hotter still. This is called thermal runaway. Processors are a mix of copper tracks and semiconductor, and do throttling and similar, so the resistance/temperature curve will be complicated. However they definitely draw more current with higher voltage, and consequently will draw less current as voltage decreases.

The fallacy in the above argument is the assumption that power consumption is constant. If it was, then increasing voltage would decrease current and vice versa. However power is not constant, and decreasing voltage will decrease power used.

In the case of a processor, power consumption increases linearly with frequency and quadratically with voltage increase. Electromigration is a function of current density, which is approximately proportional to voltage.
 
Last edited:
na, as jon has said, the lower voltage in a cpu will cause lower current draw at a given frequency, so all it's doing really is lowering heat output :)
 
dont want to do wc as i dont know enough about it and dont want to spend a fortune on it.

If you can build a PC from components - you have more than enough confidence to put together a loop from the 4 basic parts - waterblock(s), pump, radiator, reservoir.
 
If you can build a PC from components - you have more than enough confidence to put together a loop from the 4 basic parts - waterblock(s), pump, radiator, reservoir.

finance i believe is more of the issue than skill. i'm a complete noob at pc stuff, and so am still on a learning curve, but i've put together 2 loops so it's not beyond everyone, but money is more of a factor :)
 
I'm not sure about watercooling being easy. In terms of combined res/pump, cpu block, radiator in series it's every bit as simple as plugging ram into a motherboard. Trying to optimise it once there are multiple blocks/radiators/pumps involved is pretty difficult.
 
I'm not sure about watercooling being easy. In terms of combined res/pump, cpu block, radiator in series it's every bit as simple as plugging ram into a motherboard. Trying to optimise it once there are multiple blocks/radiators/pumps involved is pretty difficult.

getting a basic setup with decent components isn't too difficult tho, with decent results.

as you say, squeezing ever drop of performance out of it is the hard part
 
Back
Top Bottom