Michel R. Carleer wrote:
> Well, in my case, I can tell you that the ~0.1V variation has nothing to do
> with noise. It is indeed clearly due to the C2D going into idle power saving
> mode:
> - As long as I do nothing with my PC, the voltage is around 1.14 +/- 0.01 V.
> As soon as I start using the CPU heavily (Orthos or some other intensive
> prog), the voltage raises to 1.22 +/- 0.01 V.
> - It happens completely in sync with the clock multiplier going from x 6 to
> x 9. And back when going back to idle.
> OK, this is all with an Asus P5W DH mobo.
> Speedfan gives me the exact same figures.
There are two major components to voltage variation.
1) Variation due to Speedstep (EIST). OS changes the FID/VID on the fly,
based on processor loading. This can be disabled (a couple different
ways).
2) Load line variation. Popularly referred to as "droop". The amount of
droop is specified in the processor datasheet. There is a min and a
max load line for the processor. Removing all the voltage droop, would
actually violate the processor spec. A certain amount of droop is
normal, and the acceptable range of load lines is documented by Intel.
For example, on my processor, I might expect to see the measured voltage
drop by 0.15V, if the processor goes from no load to full load. That is
perfectly normal and since VID is fixed on my processor (no EIST), the
0.15V change is due to the load line. Each family of processors has
its own load line, so the number for Core2 Duo could be a different one.
You can see min, max, and nominal load lines, on page 21 here, for Core2.
A motherboard is supposed to stay between min and max, when the regulator
is designed for Vcore.
http://download.intel.com/design/pro...s/31327804.pdf
Paul
>
> "Jack R" <> wrote in message
> news:Xkbgi.11$%...
>>> "Jack R" <> wrote in message
>>> news:ZXVfi.32$...
>>>> What utility/program do you folks use to measure the CPU core voltage?
>>>> I've been trying an assortment and getting varying results, so I don't
>>> know
>>>> what to trust, and which to use as a guide for OC settings.
>>>>
>>>> I get:
>>>>
>>>> CPU-Z 1.40: 1.213V (under stress testing, jumps to 1.440V and back)
>>>> CoreTemp 0.95: 1.3250V
>>>> BIOS: 1.48V
>>>> PCProbeII, SpeedFan and FreshDiag: 1.52V
>>>>
>>>> This is on an ASUS P5N-E SLI mb, with an Intel C2D E6600, set at +.100
>>>> in
>>>> the BIOS.
>>>>
>>>> Thanks for help and suggestions,
>>>>
>>>> Jack R.
>>>>
>>>> (Currently running at 3.3GHZ, mem at 856MHz, video +15% OC)
>>>>
>>>>
>>> "M. R. Carleer" <> wrote in message
>>> news:f5r6o7$908$...
>>> The E6600 has a feature by which it saves power when idle. It does this
>>> in
>>> two ways that I know of: lowering the internal clock multiplier (from x9
>>> to
>>> x6) and lowering the requested VCore by about 0.2V.
>>> This explains what you read in CPU-Z, unless you specifically disabled
>>> the
>>> power saving feature in the BIOS.
>>> As for the other progs, I don't use them so cannot comment (except for
>>> speedfan which I use only to measure temps).
>> Thanks for your reply.
>> I have all of the power saving features turned off.
>> My guess is the variation is just measurement 'noise', not actual voltage
>> variation.
>> It only occurs during 100% stress tests, not in any static situation.
>> Jack R.
>
>
Similar Threads
Linear Mode
