Max VCore - in BIOS or Windows ??

Stuart Gibson

Stuart Gibson

Stuart Gibson

Associate
Joined
25 Mar 2007
Posts
144
Location
NE England
I've been overclocking my rigs for years :D, but I've never been sure about which max vcore I should be working to. :confused:

I always seem to need more volts in the BIOS to get a certain overclock stable under load than other people are reporting. :( So are they reporting the BIOS setting or the vcore under load in Windows ? :confused:

The Intel specs don't tell you which one it should be :mad:, so can anyone here tell me which one is the right one ?
 
Don't really know the answer, but my guess would be they are reporting what they see in Windows. I have exactly the same as you, in that I need to put the vcore higher in the BIOS than other people are reporting. At least it looks better when I check it in Windows afterwards. I have no idea why you get different voltages reported though.

I'll wait for the more experienced people to comment - would be interested to know.
 
Yea, most people say what they see in windows. If they're saying BIOS voltage they tend to say "I got it to xGHz with these settings in the BIOS..."
 
A bit more to it than this, though it is indeed windows (cpu-z, mostly) people work from. Two cases, with load line calibration (sometimes called vdroop) on or off.

With llc off / vdroop on, the reading set in the bios is a maximum. It is what the cpu will see when idle. So when in windows, under some load, it is always below this. The current source is not perfect so the voltage across the cpu drops when the current drawn increases, and the current increases with processing load.

Will llc on / vdroop off, the voltage set in the bios is a target/minimum. In windows, under various loads, the board will strive to maintain this voltage across the cpu. It does this by increasing/decreasing the supply voltage to try to keep the end result after vdroop at the set value. So, the bios reading *should* match the windows one, calibration issues aside. A side effect of this is the voltage going far over the value set in the bios, briefly but frequently. Opinion varies as to whether this matters or not. I personally don't want my cpu to see the bad side of 1.8V, however briefly.

The short answer to the question is indeed "windows" however.
 
There's the issue :(

It's really hard to find someone who's connected one of these to an oscilloscope, and I haven't done so. I believe everest has a high enough resolution to see the changes in voltage when changing load. Anandtech posted the main source for this, and the graphs aren't very usefully labelled.

My conviction comes straight from fourier analysis, which I freely confess I am not good at. The idle to load change on my P5Q was close to 0.1V, the damned thing dropped from 1.4 to 1.31V. I don't know if this is standard, I'll check what my UD5 does if you like. It's around 5% from my experience with the P5Q, and sounds reasonable. The issue I have is I don't know how fast it changes from idle to load, any ideas? The ringing is dependent on the size of the step, which isn't too bad here, and the time period over which the voltage changes. If it took a second or so, no worries. However I fear a processor may change from idle to load and back very quickly, and do so often.

Any ideas for where to find some numbers for this? I've seen ringing on an oscilloscope, and from fourier series, however don't know where to get the time period from to estimate it. For the time being I'm playing it safe by sticking to intels specification; I'd be very interested if you can help with the calculation here.
 
You must log in or register to reply here.
Back
Top Bottom