Aerothorn on 28/6/2011 at 01:42
I have a machine I build about two years ago that has an AMD Phenom II 940 BE as a processor, a very overclock-friendly CPU. I've previously avoided overclocking out of caution more than anything else, but given that I current have a well-ventilated PC with a single extra-cool graphics card, I'd figure it wouldn't hurt to overclock it a bit.
The challenge is that I (somewhat foolishly) opted for a 380 watt power supply back when I built this, which was more than enough for the base components, but I didn't factor in capacitator aging and overclocking potential back when I did that.
So I looked at Tom's Hardware and found that you don't really have to increase the voltage for lower overclocks. At the stock clock of 3 GHz under load (i.e. not running Cool & Quiet) the voltage was 1.39 or so. It operates fine there up to 3.4 GHz, and it only has to be raised to 1.4 for 3.6.
Hunky dory, I think. But out of curiousity I load up CPU-Z to see what my current stock config looks like...and found that I had a core voltage of 1.54 (and this is with it maybe running cool and quiet - it's unclear as it keeps oscillating between 1.8 and 3 GHz).
I thought this might be an error, but ran CPUID's Hardware Monitor and came up with the same result, a fixed vcore of 1.54. Why is this so much higher than normal? Is there a reason behind it? Is it fixable? If that's what it has to be at stock I certainly can't overclock!
Edit: What's weird is that in the BIOS at startup, it lists the VCORE as a normal 3.7. So either it's increasingly drastically once Windows boots or both the CPUID programs are wonky (and looking it up, I've seen no such problems reported with it and this CPU).
My CMOS battery seems to be failing (won't remember date/time) and I'm replacing it tomorrow. Could it be related to that?
baeuchlein on 16/7/2011 at 16:58
Quote Posted by Aerothorn
At the stock clock of 3 GHz under load (i.e. not running Cool & Quiet) the voltage was 1.39 or so. It operates fine there up to 3.4 GHz, and it only has to be raised to 1.4 for 3.6.
Hunky dory, I think. But out of curiousity I load up CPU-Z to see what my current stock config looks like...and found that I had a core voltage of 1.54 (and this is with it maybe running cool and quiet - it's unclear as it keeps oscillating between 1.8 and 3 GHz).
I thought this might be an error, but ran CPUID's Hardware Monitor and came up with the same result, a fixed vcore of 1.54. Why is this so much higher than normal?
Did you verify vcore with CPUID and the CPU running at
stock speed (not overclocked)? If it's the same vcore (1.54) then, I would assume that CPUID's readings are faulty.
Quote Posted by Aerothorn
Edit: What's weird is that in the BIOS at startup, it lists the VCORE as a normal 3.7.
Do you mean 3.7 Volts? That would not be very normal. Or did you want to say, "it lists the normal voltage [which would be around 1.39 Volt, as you wrote earlier] for a CPU speed of 3.7 GHz"?
Anyway, usually the BIOS is more reliable concerning sensor readings than any program. For example, Linux just tells me that the 3.3 V line of my mainboard is oscillating between 2.97 and 3.40 V; I rather think that the values read from the hardware aren't that reliable. Sometimes, the BIOS can partially compensate for this. Furthermore, the BIOS can be programmed to ignore some values reported by the monitoring chip; quite often, the chip has some connectors not connected to anything on the mainboard.
If your CPU's voltage is far too high, the CPU temperature should also go up. So, if the voltage was too high in Windows (with no programs running, meaning the CPU is idle most of the time), you should see a notable decrease in CPU temperature
over time once you reboot, enter BIOS, and look at the CPU temperature values there for a few minutes.
If the CPU temperature sinks several degrees over time, Windows may really select a wrong CPU voltage (although I don't know why it should do that - but that's common with Windows, if you ask me...). If, however, the CPU temperature does not change much once you have entered the BIOS, the CPU did not heat up in Windows, thus a drastically higher CPU voltage in Windows is unlikely.
There
may be an immediate temperature jump when you switch from looking at the temperatures with a program to looking at them with your BIOS. But that can be an illusion, since a program may apply corrections to the raw temperature value reported by the hardware monitoring chip, and the BIOS may apply a different correction algorithm.
