Opteron Dual Core: Voltage and Temps, limits of...

[OatMan]

In an attempt to some up information thats all over the place and link it ultimately to the appropriate "unofficial" thread, I ask this, cause everyone wants to know and it is really hard to make sense of all the meandering tech advice...

So

What are the safe/reasonable limits for voltage and Temps for Dual Core Opterons?

Points to consider...
1. I know there is much speculation and opinion involved, so opinions will vary
2. There is the apparant voltage differential issue between CPU and RAM voltages, that is a topic for other threads (and has also meandered all over the place.
3. CPU, Case, PWMIC, Chipset temps should all be considered and have different thresholds.
4. The lines of safe, risky and nutty are blurred
5. Opterons may be more robust in many ways (mem controller, tolerances) compared with A64x2s...

So please I beseach the big brains of the AT OCing community to chime in to help make life easier for many new dual core opty OCers for what I think is a pretty critical subject.

THanks!

 

[OatMan]

I have good air cooling and never get much past 50C at Load for any test so far.

I have not tried more than 1.65vCore. I would rather not go past 1.50 vCore, which I figure is probably low risk. My CPU doesn't have to last 10 years or even 5.

I use MBM5 which I have found to be rather twitchy with the DC optys. I use CPUz 1.31. I test with OCCT, P95, Everest, All flavors of 3DMark, Play FEAR for a while, memtest, and SuperPi.

Now please someone who knows what they are talking about add some actually substantial/supportable info. THanks!

OM

 

[onn2000]

I do some circuit design on CMOS processes. I don't know what the limitations of AMD's 0.09u SOI process is, but from my experience, they normally have a 0.2V leeway for the breakdown voltage. For example, for a certain 0.25u process, the breakdown voltage is 2.5V, but 2.7V is the absolute maximum that they'd allow designers to reach. So, assuming that you leave 0.05V as leeway for fluctuating voltage, I'd say that you shouldn't go more than 0.15V in Vcore for any processor. In the case of Opteron Dual Cores, I think that maximum voltage (not including fluctuations) is 1.45V. Having said that, since X2s have a Vcore of 1.35-1.4V, even though they also use the process as the DC Opterons, you could probably go to about 1.50V-1.55V (not including fluctuations) with Opterons, just like the X2s.

As for temperature, AMD specifies the max to be about 65 degrees celcius, I think. Take them seriously.

Note: Breakdown voltage is where once a transistor reaches that voltage, it will never return to functioning the same way ever, and its behaviour will be non-predictable thereonwards.

 

[racinjimy]

one thing to be careful of on the NF4 DFI boards (not expert) is overvolting on dual cores

Hi all. I have a request for anyone using a Dual Core that has a multimeter. I noticed that while CPUZ reported my Vcore as 1.552V, my actual idle Vcore meaured by a DMM was 1.609V. Disturbingly, I found that it jumped to 1.659V under load! This is nearly .11V greater than reported by software monitoring programs! We know these DFI boards underreport voltages in Windows, but this discrepency approaches dangerous levels.

If anyone could check their idle/load Vcore's (as well as reporting VID and VID multiplier from BIOS), I would be extremely appreciative. We need to find out if this is isolated to my rig or is endemic to Dual Cores generally.

For more info, see HERE. Thanks for your help.

my results

dual core 175


Bios 1.35 X 110%
CPUZ 1.424
idle DMM 1.493
dual p95 DMM 1.524


so if one is too assume that safe voltage to be 1.6V or so, 1.45-1.5V should be close to the max your run ON A DFI

when in the 1.45V vcore range you should be good to at least 3.2V vdimm

PWMIC temps can get quite high I see 55-60C with active cooling (fan blowing on the ram/pwmic) while dual priming running at the 1.35X110% vcore and 3.2V vdimm

 

[fbrdphreak]

Originally posted by: racinjimy
one thing to be careful of on the NF4 DFI boards (not expert) is overvolting on dual cores

Hi all. I have a request for anyone using a Dual Core that has a multimeter. I noticed that while CPUZ reported my Vcore as 1.552V, my actual idle Vcore meaured by a DMM was 1.609V. Disturbingly, I found that it jumped to 1.659V under load! This is nearly .11V greater than reported by software monitoring programs! We know these DFI boards underreport voltages in Windows, but this discrepency approaches dangerous levels.

If anyone could check their idle/load Vcore's (as well as reporting VID and VID multiplier from BIOS), I would be extremely appreciative. We need to find out if this is isolated to my rig or is endemic to Dual Cores generally.

For more info, see HERE. Thanks for your help.

my results

dual core 175


Bios 1.35 X 110%
CPUZ 1.424
idle DMM 1.493
dual p95 DMM 1.524


so if one is too assume that safe voltage to be 1.6V or so, 1.45-1.5V should be close to the max your run ON A DFI

when in the 1.45V vcore range you should be good to at least 3.2V vdimm

PWMIC temps can get quite high I see 55-60C with active cooling (fan blowing on the ram/pwmic) while dual priming running at the 1.35X110% vcore and 3.2V vdimm

:Q:Q:Q:Q:Q:Q

*Goes and turns DFI board down from 1.55 + 1.04%*

EDIT I can check on my DFI nF4 with a multimeter, where do I measure?