Two Different Voltages on Each CPU Core

[Jimmy Question]

For several weeks I have been aware of the fact that one core consistantly fails Prime95 testing before the other. At first I thought that one core was just "weak", as I have heard that this can happen sometimes and I assume there's nothing much to be done about it.

 

But today while looking at SiSoft Sandra Lite's report, it shows each core as having a different voltage. Not only that, but the higher of the two is actually a half volt less than what the BIOS is set for.

 

[ATTACH]5506[/ATTACH]

 

As you can see, one core averages 1.26 volts, the other averages 1.34 volts.

 

That's almost a full volt difference between the two!

The vcore in BIOS is set for 1.375 volts. How accurate can I expect this to be ? If BIOS is different from Sandra, which should I believe ?

 

I have many questions.

 

First, could this be some problem with Sandra Lite, or can I rely on these readings ?

 

If the readings are correct, would the problem be attributable to the Motherboard, or the CPU ?

 

If it's the Motherboard, is this something I can tweak on the Motherboard, or are we getting into RMA territory ?

 

Does this happen to other people, or is this something "rare" ?

 

I wonder if this may mean I might be able to really tweak performance out of this CPU, as I have been considering getting the Ultra D. Maybe if I can RMA the RS482 I can do that. Or I could be okay with keeping it if I can fix it, as I like the board and would like it to function with the CPU at max performance.

 

Any opnions, suggestions & expertise welcome, & thanks in advance.

 

Jimmy Q

[MACarter02]

At first I thought that one core was just "weak", as I have heard that this can happen sometimes and I assume there's nothing much to be done about it.

 

One core will always fail. having both fail simultaneously would be like winning the lottery three times in a row.

[Poisonsnak]

First off I completely agree with MACarter02 about the weak core strong core thing, my 4400+ is a case in point one is stable to 2.5 at 1.45 the other 2.7 and possibly higher, I just haven't bothered going farther with it.

 

To be honest I've never seen an app that measures the core voltages independently, and I don't really think it's even possible, the 939 package probably has a lot of pins assigned for core voltage but I don't see how they could be designated for one core or another since the same package is used for single core CPUs.

 

Also there's no way it could be measuring on the CPU itself, the measuring chip is on the motherboard.

 

My guess is that Sandra knows you have 2 CPUs, and found 2 voltages from the monitoring chip that look like they're in the right range for CPU voltage and reported them as such. One of them is probably chipset voltage.

 

Note also the CPU2 fan speed in your picture. I doubt you have 2 CPU fans 1 at 2k RPM the other at 5k. One is just a different fan reading from the monitoring chip and Sandra just basically guessed that it was on your other CPU (it's assuming 2 CPUs not 1 dual-core).

 

I just installed SiSoft SANDRA Lite XI (11.17) and under "Mainboard" I have one CPU voltage only at 1.31V set to 1.35V in the BIOS.

Note that I'm using an ASUS A8N-SLI Premium though but once my RS482 comes in I'll (probably forget to) try it out.

 

As for the measured voltages being lower than set voltages, that's normal, I probably would attribute it to the low accuracy of motherboard monitoring chips, just take a voltmteter and test your +12V and you'll see what I mean.

[Jimmy Question]

In order to verify the voltages, I ran Everest to see if it would give a separate voltage for each core. It doesn't, but it does give two temperatures.

 

[ATTACH]5507[/ATTACH]

 

Everst shows 46 deg. C for Core #1, 41 Deg. C for Core #2, which seems about right in line with what Sandra says the voltages are.

 

Again, I am wondering if this corroborates the idea that there is a problem with the Motherboard, and if so can I fix/tweak it right or RMA it ?

 

Are there any test points on the M/B for vCore voltage ?

[CPDMF]

It's not a board problem, software doesn't read what CPU voltages are correctly. Both cores of the CPU operate off of the same voltage source, they do not get seperate voltages..

[noneed4me2]

Opteron 64 are built on a single die versus the older Pentium Duals before conroe on 2 seperate, so your voltage 1.34 is the CPU the other as previously said is most likely your NB chipset voltage 1.26 sounds about right at stock I have mine at 1.4 but check to see what your Bios voltage settings against what Sandra says and you will probably find they match and I have noticed I get slight fluctuations depending on the benching tool.Also I found raising the NB voltage gave me a little better OC.

[Jimmy Question]

Any way to read the vCore voltage on the motherboard ?

[Jimmy Question]

SiSoft Sandra lite also reports 2 cores, each running at significantly different temperatures.

[HSCHOUEST]

i never rely on sandra for temp readings