Help controlling Phenom II X6 1090T voltage

[ET]

I recently bought a Phenom II X6 1090T. I'm running it at stock speed and underclocking it, to keep it as cool (and quiet) as possible when folding.

Using AMD OverDrive, the default voltage shows as 1.35V and I can lower it to 1.25V without problem. Where I have a problem is when I put the PC to sleep and wake it up (I don't normally fold 24/7), because the voltage is set back to 1.35V.

I tried going through the BIOS of my Gigabyte GA-MA790X-UD4 (AM2+ board), but the voltage shows there as 1.475V, which doesn't seem related to the 1.35V OverDrive is reporting.

Any ideas? Thanks.

 

[BD231]

I'd say stop using software to underclock your computer.

 

[ET]

That's what I'd like to do. My problem is that I trust OverDrive but don't trust the BIOS. I can change the value in OverDrive and I can see the effect on the temperature (and hear the change in the fan), but the value in the BIOS is unrelated to what I see in OverDrive and when I tried to change it I didn't see any change to what I saw in OverDrive.

What I'd be happy to hear is whether other people have experienced that kind of difference between BIOS and what's reported in software and how to deal with it.

 

[mrcmtl]

For clocking settings, I go inside the BIOS. But for my fan control, I use software because it has more precise control. I also have to reapply those settings in software after each wake up.

 

[BD231]

That's what I'd like to do. My problem is that I trust OverDrive but don't trust the BIOS. I can change the value in OverDrive and I can see the effect on the temperature (and hear the change in the fan), but the value in the BIOS is unrelated to what I see in OverDrive and when I tried to change it I didn't see any change to what I saw in OverDrive.

What I'd be happy to hear is whether other people have experienced that kind of difference between BIOS and what's reported in software and how to deal with it.

1.475v is the CPU voltage, Gigabyte motherboards report max turbo/load voltage that's all. Confused me a bit at first as well. Lower it by 1.000v and you're set. Just remember, when Thuban goes full load or turbo it jacks up the voltage, can't remember the exact amount but its something over 1.000v on the cpu, and .250v to .500v on the northbridge as well.

 

[LoneNinja]

Google PhenomMsrTweaker, it's a handy program for adjusting cpu multipliers and voltages. It'll work with CoolnQuiet and saves your settings for every time windows loads, no need to keep opening the program. You can even set 3 different presets for High Performance, Balanced, and Power Saver. I use it on both an X6 1090T and X4 965.

 

[thilanliyan]

^I was using PhenomMSRTweaker but 64-bit windows put a stop to that.

 

[exar333]

That's what I'd like to do. My problem is that I trust OverDrive but don't trust the BIOS. I can change the value in OverDrive and I can see the effect on the temperature (and hear the change in the fan), but the value in the BIOS is unrelated to what I see in OverDrive and when I tried to change it I didn't see any change to what I saw in OverDrive.

What I'd be happy to hear is whether other people have experienced that kind of difference between BIOS and what's reported in software and how to deal with it.

I think you have this backwards. BIOS should be much more reliable than any software OC application. Making any frequency change should always require stability testing, regardless of where it is made. OC is more than just checking the temp.

 

[ET]

BD231, thanks, but I tried it again, and changing the voltage value in the BIOS doesn't seem to do anything. I still get the same 1.35V when Windows loads.

As for PhenomMsrTweaker, first of all if it doesn't work for 64-bit Windows that's a problem. Secondly the problem isn't at PC load. Even OverDrive should theoretically do that (though it doesn't in the version I'm using, even though I set the option). The problem is on resume from sleep.

What I'm going to try is see what I can get by running programs when Windows resumes. Some googling showed me it's possible to run tasks at resume time, and at the very least I can have OverDrive run at that time. If I can find an API which will let me change the voltage in a program, then I could do it transparently, so I'll be looking for that.

Of course if I can get the BIOS undervolting to work that'd be better, but since I also have this problem when undervolting my graphics card I think it's worth looking into this solution.

 

[BD231]

BD231, thanks, but I tried it again, and changing the voltage value in the BIOS doesn't seem to do anything. I still get the same 1.35V when Windows loads.

If ur using aod to read your voltages youre wrong, aod reports default voltage of the chip not your motherboard. Thats why it says 1.35v even tho ur bios is @1.475v. U really need to get away from that software and figure out ur bios. Who knows, maybe u need to lower the voltage even further to replicate aod voltages. Either way aod works directly thru your bios, whatever aod can do so can ur bios.

Start over using only ur bios and cpu-z. Imo aod is confusing u. Good luck.

 

[Jovec]

The CPU Status tab on AOD is bugged as it shows the stock value regardless of setting. However, the Board Status tab pictured below will (should?) show the correct CPU voltage as VIN0:

 

[LoneNinja]

^I was using PhenomMSRTweaker but 64-bit windows put a stop to that.

I use to use K10stat but switched to Phenom MSRTweaker because K10stat wouldn't recognize all 6 cores of my Phenom II X6. Both computers I use PHenomMSRTweaker on have Win7 Pro 64bit. Maybe it was an older version that you had that doesn't work in 64bit. I just downloaded it a few months ago.

 

[thilanliyan]

^Hmm, I'll give it a shot, but it seems Cool 'n Quiet does a decent job anyway.

 

[ET]

I ran HWMonitor (from the guys that may CPU-Z) and it reflects what AOD shows. I set the CPU VID target to 1.25V and started folding, the CPU VCORE voltage in HWMonitor went briefly up to 1.475V (the BIOS value), then settled on 1.25V.

As for the board status page Jovec mentioned, it shows ~0.70V for VIN0, which looks unrelated to the CPU core voltage. Also the posted picture looks like it's from a very old version of AOD.

Frankly, after checking with HWMonitor I trust AOD even more, and the BIOS value looks like a maximum that should be kept that way (when I lowered it I experienced some instability) and not the "target voltage" that AOD controls.


Update:

I downloaded PhenomMsrTweaker, which has a 64 bit version and seems to work fine. It shows 5 states: P0 to P4. P1 is where the 1.25V is set, and P0 is where the 1.475V is set. Obviously P1 is what's typically used while folding, and P0 can't be lowered much or it causes instability. PhenomMsrTweaker looks like a good utility and I will try using it. Thanks for refering me to it.