P5B-E Vcore

I have a 6850 in an ASUS P5B-E. I haven't messed around with it too
much but it is stable at 3.825 GHz.

In BIOS, I have Vcore set to 'AUTO'. CPU-Z reports the voltage as
1.225, while ASUS probe reports it at 1.54V.

Which is correct?

TIA

Lowly Engineer wrote:
> I have a 6850 in an ASUS P5B-E. I haven't messed around with it too
> much but it is stable at 3.825 GHz.
>
> In BIOS, I have Vcore set to 'AUTO'. CPU-Z reports the voltage as
> 1.225, while ASUS probe reports it at 1.54V.
>
> Which is correct?
>
> TIA

CPUZ could be reporting the nominal voltage (like from a register
inside the processor). Asus Probe is more likely to be using the
hardware monitor. I'd say that Asus Probe is giving you the real scoop.

Why the voltage got that high, depends on how you are overclocking.
If you use manual overclocking techniques, then it is up to you
to turn it up that high. If using some of the automatic overclocking
techniques, then the BIOS has probably turned it up automatically,
in response to the high clock.

If it was my system, I'd overclock it manually, only applying as
much voltage as is necessary, to run at the frequency. For example,
run at stock voltage, increase FSB in small steps, checking stability
with something other than my Windows hard drive. A Knoppix or Ubuntu
CD, for example. Or even memtest86+. When the processor shows signs
of instability, apply a small voltage increment, then go back to
small frequency increases. From the trend you get from those data
points, you should get some idea how much voltage is needed to get
to "X" frequency.

Eventually, you hit a wall, where a lot more voltage is needed.
And that is telling you, it is time to stop. Then back off to
a set of conditions, that you think can be applied for the life
of the processor (so-called "everyday overclock" value).

As a final check, boot from Knoppix or Ubuntu, go to mersenne.org,
and get a copy of Prime95. Use the Torture Test option, and see
if it runs error free for hours on end. There is also the Orthos
program, which runs multiple copies of Prime95. For a multicore
processor, you want to run a Prime95 thread per core, to get
the maximum degree of testing. If you can pass Prime95 under those
conditions, only then would I plug my Windows hard drive back in
and try booting from it.

A Windows hard drive can be corrupted, from overclocking experiments,
so you run the risk of losing the boot drive. Which is why you
need to find a bulletproof bootable OS, such as a Linux Live CD.
By booting from a CD, and having no hard drive present, there is
nothing to corrupt.

Paul