VcoreA voltage varies

  • Thread starter Thread starter Mark
  • Start date Start date
M

Mark

Hi,

I have a Gigabyte GA-965P-DS4 m/b with Intel C2D E6420 CPU. According
to the supplied utility Vcore A varies between 1.210 and 1.370V.

Is this normal or indicative of a problem?

TIA, Mark
 
In message <[email protected]> Mark
I have a Gigabyte GA-965P-DS4 m/b with Intel C2D E6420 CPU. According
to the supplied utility Vcore A varies between 1.210 and 1.370V.

Is this normal or indicative of a problem?
Click to expand...

I wouldn't trust the motherboard's voltage reading until proven
accurate, although I can't speak to that motherboard specifically.
 
Hi,

I have a Gigabyte GA-965P-DS4 m/b with Intel C2D E6420 CPU. According
to the supplied utility Vcore A varies between 1.210 and 1.370V.

Is this normal or indicative of a problem?

TIA, Mark
Click to expand...

I think Speedstep could be dynamically changing the core voltage in
line with CPU usage.

- Franc Zabkar
 
Franc said:
I think Speedstep could be dynamically changing the core voltage in
line with CPU usage.

- Franc Zabkar
Click to expand...

I think RMClock has a FID/VID display, and it will show Speedstep activity.
Correlate Speedstep with the monitor chip readings.

http://cpu.rightmark.org/download/rmclock_225_bin.exe

Picture of a typical display.

http://img105.imageshack.us/img105/6184/vidchange2hl.jpg

There are two kinds of voltage variation. There is the variation caused
dynamically by Speedstep, where the processor writes a different VID value
to the regulator. But there is also the "load line" for the processor.
The more current the processor draws, the lower the voltage drops - the
slope of that curve is the output impedance of the regulator. And by
design, the impedance is not made to be zero, but some finite value.
Checking the data sheet for the processor, will show a plot of current
versus voltage, along with two lines representing the highest and lowest
load line allowed.

So just looking at the instantaneous voltage reading from the monitor
chip, in isolation from all else, doesn't tell you a lot.

Paul
 
I think RMClock has a FID/VID display, and it will show Speedstep activity.
Correlate Speedstep with the monitor chip readings.

http://cpu.rightmark.org/download/rmclock_225_bin.exe

Picture of a typical display.

http://img105.imageshack.us/img105/6184/vidchange2hl.jpg
Click to expand...

Thanks for that.
There are two kinds of voltage variation. There is the variation caused
dynamically by Speedstep, where the processor writes a different VID value
to the regulator. But there is also the "load line" for the processor.
The more current the processor draws, the lower the voltage drops - the
slope of that curve is the output impedance of the regulator. And by
design, the impedance is not made to be zero, but some finite value.
Checking the data sheet for the processor, will show a plot of current
versus voltage, along with two lines representing the highest and lowest
load line allowed.
Click to expand...

FWIW, the Vcore for my Athlon XP 2500 CPU drops from 1.65V at idle to
1.63V at full load. This represents a 1% variation. The OP is seeing
about 12%. Of course the reading depends on where it is taken, ie
whether the test point is closer to the regulator or to the CPU, so
the comparison may be inconclusive.

I'm using CPUIdle and CPU Burn-in for my tests.
So just looking at the instantaneous voltage reading from the monitor
chip, in isolation from all else, doesn't tell you a lot.

Paul
Click to expand...

Here is an Intel application note that describes a reference design
for a Vcore regulator:
http://www.intersil.com/data/an/an1126.pdf

AIUI, Vcore is regulated right at the CPU (see page 6), with current
sensing occurring immediately upstream. I don't see any intentional
series impedance, ie I can't see any load line, although I expect that
the copper traces would present some finite resistance.

- Franc Zabkar
 
I was wrong. This is VID not Vcore.

M
Click to expand...

Even better. The VID consists of several electronic switches, either
ones or zeroes, which are the CPU's way of telling the control chip in
the Vcore regulator what voltage it wants. So if the Vcore is varying,
then it's because the CPU is doing it.

- Franc Zabkar
 
Back
Top