Will I be able to undervolt the CPU?

[Mikael]

I'm pretty set on upgrading my system and I'm thinking of buying the parts below (I have the rest of the parts that I need):



What I'm wondering about is the motherboard (ASUS M2V). I've been trying to find any info about its undervolting capabilities, but I haven't found anything. I'm hoping to be able to run the CPU at 1.1-1.2V. I'd also like to be able to set this voltage in the BIOS and not having to use software.

Any other thoughts about this motherboard? Would it be better to buy an ASUS M2N-E? I saw the M2N-E mentioned in a thread here, saying that it supported 0.8-1.525V. No artificial limit at 1.1V on that board then?

[smilingcrow]

Try downloading the manual and seeing if it mentions this feature. This is just indicative as features come and go depending on BIOS versions.

In case you aren’t aware, when you under-volt in the BIOS you lose the ability to change the Vcore dynamically using Cool ‘n’ Quiet or 3rd party software. I’m not aware of any motherboards that allow these two features to work together!
So if you need to run a VCore higher than 1.1V, you will be penalised at idle with higher CPU temps and power consumption.
The 1.1V limit seems to be imposed by Cool ‘n’ Quiet, you cans et what you like in the BIOS.
I’ve found CrystalCPUID to be a solid performer with AMD systems for under-volting.

[Mikael]

Try downloading the manual and seeing if it mentions this feature. This is just indicative as features come and go depending on BIOS versions.

Already did that. Nothing of interest...

In case you aren’t aware, when you under-volt in the BIOS you lose the ability to change the Vcore dynamically using Cool ‘n’ Quiet or 3rd party software. I’m not aware of any motherboards that allow these two features to work together!
So if you need to run a VCore higher than 1.1V, you will be penalised at idle with higher CPU temps and power consumption.
The 1.1V limit seems to be imposed by Cool ‘n’ Quiet, you cans et what you like in the BIOS.
I’ve found CrystalCPUID to be a solid performer with AMD systems for under-volting.

Yeah, I know. I somehow prefer having it all configured with static settings, though. The difference at idle between a C'n'Q enabled 3200+ and a 3200+ @ 1.1V will probably be very small anyway.

[Tibors]

I saw the M2N-E mentioned in a thread here, saying that it supported 0.8-1.525V. No artificial limit at 1.1V on that board then?

The artificial limit of 1.1V is caused bu the CPU's not by the boards.

Also double check how the person who wrote that undervolted. In the BIOS or with some software tool. A lot of people don't seem to make that distinction.

[Mikael]

The artificial limit of 1.1V is caused bu the CPU's not by the boards.

The limit can not be caused by the CPU, since it is the board that applies the voltage to the chip. Sure, the chips may not like sub-1.1 voltages, but the chips themselves can't stop us from applying said voltages.

[Tibors]

If that is the case, then please explain why we only see this problem with E6 stepping CPU's.

The number of boards which actually have undervolting settings in the BIOS is fairly low. Most of the talk about undervolting you see in the last year is using software like CrystalCPUID. This software sends signals to the CPU, which in turn asks the mobo for a certain voltage. An E6 stepping CPU never asks the mobo for anything lower than 1.1V, no matter what the software tells it.

But you don't have to believe me. Just browse these forums.

[smilingcrow]

If that is the case, then please explain why we only see this problem with E6 stepping CPU's.

The number of boards which actually have undervolting settings in the BIOS is fairly low. Most of the talk about undervolting you see in the last year is using software like CrystalCPUID. This software sends signals to the CPU, which in turn asks the mobo for a certain voltage. An E6 stepping CPU never asks the mobo for anything lower than 1.1V, no matter what the software tells it.

I think you are both talking at cross purposes here! Mikael is interested in under-volting in the BIOS, in which case the limitation is purely down to the motherboard and has nothing to do with the CPU. At least it’s my understanding that E6 revisions can under-volt through the BIOS just as D0’s can. Please correct me if I’m wrong.
Whereas you are talking about the 1.1V limit with E6 revisions using CrystalCPUID.
So you both are right.

[Mikael]

If that is the case, then please explain why we only see this problem with E6 stepping CPU's.

The number of boards which actually have undervolting settings in the BIOS is fairly low. Most of the talk about undervolting you see in the last year is using software like CrystalCPUID. This software sends signals to the CPU, which in turn asks the mobo for a certain voltage. An E6 stepping CPU never asks the mobo for anything lower than 1.1V, no matter what the software tells it.

But you don't have to believe me. Just browse these forums.

It may be a problem with the E6 CPUs because AMD has asked the motherboard makers to not allow lower voltages. What exactly do you think stops the mobo from applying the lower voltages? The CPU has to take what it gets from the motherboard.

It might be as smilingcrow said, though, that the limit is something to do with CrystalCPUID and not the BIOS. I might have confused those two, thinking there is a BIOS limit of some kind when there isn't...

[jaganath]

it’s my understanding that E6 revisions can under-volt through the BIOS just as D0’s can. Please correct me if I’m wrong.

In my experience even if you set the Vcore lower than 1.1V in the BIOS for an E6 chip, the Vcore refuses to go below 1.1V. However QuietOC got an E6 2500 down to 0.88V @ 1.4GHz (IIRC), so maybe it is something to do with the CnQ implementation (2500 doesn't have CnQ, all the chips with the apparent Vcore lock have Cnq)? QuietOC did quite a lot of experimentation on this issue, a search through his old posts would probably shed more light on this issue.

[widman]

The artificial limit of 1.1V is caused bu the CPU's not by the boards.

The limit can not be caused by the CPU, since it is the board that applies the voltage to the chip. Sure, the chips may not like sub-1.1 voltages, but the chips themselves can't stop us from applying said voltages.

it's caused by CPU, not the board.

I have MSI Neo2 Platinum. On this board I can set Clawhammer chip to 0.85V by bios or software. But on 3000+ Venice E6 and X2 3800 manchester, it refuse to set below 1.1V by bios and software.

somehow cpu can instruct board to set the voltage.

[Mikael]

it's caused by CPU, not the board.

I have MSI Neo2 Platinum. On this board I can set Clawhammer chip to 0.85V by bios or software. But on 3000+ Venice E6 and X2 3800 manchester, it refuse to set below 1.1V by bios and software.

somehow cpu can instruct board to set the voltage.

Yeah, it's caused by the CPU in a way, and in a way it's not. The voltage is controlled by a voltage ID (VID). As far as I know, the CPU has hard-wired minimum and maximum values for the core voltage stored as VID values.

This wouldn't be a problem if it wasn't for the fact that the voltage regulator on most motherboards seems to be fed these VID values directly from the CPU pins. This means that there is no way to by-pass the CPU. If the motherboard instead was designed so that the BIOS sent the VID to the regulator, this wouldn't be a problem. We could choose whether to allow the CPU to control the minimum or maximum values or to set the voltage to anything within the regulator's range.

So, whether this is a CPU issue or not is probably up to debate.

[jaganath]

the voltage regulator on most motherboards seems to be fed these VID values directly from the CPU pins. This means that there is no way to by-pass the CPU.

I'm sure there is some way of connecting up the VID pins with wire to get the Vcore below 1.1V, but I'm happy with my Sempy 3000 as it is (1.6GHz @ 1.1V), so no incentive to find out.