SPCR

As usual the great german magazine c't issue 17/06 page 162.

All in watt of the whole system idle/cpu full load. CnQ setting as in BIOS defaults. The MSI boards had CnQ disabled by default. It is unclear to me wether the CnQ of the Asus board was enabled. In a table it's disabled, in the text it is reported to be enabled.

ULi M1697:
Asrock AM2XLI-eSATA2 78/147 W

ATI Crossfire 3200 (RD580/SB600):
Elitegroup KA3 MVP 78/146 W

nforce 500:
MSI K9N Neo F 88/157 W CnQ off

nforce 570 SLI:
ABIT KN9 SLI 76/142 W
Gigabyte GA-MS67SLI-S5 86/149 W
MSI K9N SLI Platinum 86/153 W CnQ off

nforce 590 SLI:
Asus M2N32-SLI Deluxe 97/168 W

Conclusion: No difference between ULi, ATI and nvidia here. Stay away from the 590 SLI. Gigabyte uses more power, MSI should be in line with enabled CnQ. Abit has the least full load power.
Benchmark performance was about the same. The USB issues with ATI southbridges are solved. This time, all four nforce 500/570 SLI boards lagged behing by around one third.

Good stuff posting this info, thanks.


As the ati 3200 chipset has all of the features of the 590SLi it can be seen as the more efficient chipset.

i think implimentation has alot more to do with things than the chipsets themselves.

Case and point, notice the huge varience in power usage in just the 3 boards using the nforce 570 SLI. I bet if you took a wider sampling of the other chipsets you would find similar variences within the same chipsets.

We should get some SPCR insider in a big computer store with time on his hands to do a test like this with a wider range of mobos. There seems to be no simple way of guessing which mobo is more efficient, you can guess that if it's a 590 it's going to use alot, if it's a VIA it'll use less, and if it has more power regulation phases (within reason) it will be more efficient.

But this abit board surprises, maybe it's got to do with better cooling of VRM's by their SilentOTES system. If that is the case, it will largely depend on your cooling setup. I wonder what the Asus M2N-SLI would look like in the 570 comparison as it has a 4-phase VRM while the ones tested here are 3-phase.

For the moment finding an efficient mobo is a black art. Hopefully SPCR's upcoming article on VRM efficiency will help make it more scientific. I can't wait

Wait a sec! I just looked at the Abit website, and it appears that the KN9 is based on the nForce 4 Ultra, not the 570 SLI. Or maybe you meant the KN9 SLI...

Funny, I thought the 5xx series could only have been an improvement over the past generation in terms of power consumption (I thought they were 90nm like the nForce410/430)

cAPSLOCK wrote:Wait a sec! I just looked at the Abit website, and it appears that the KN9 is based on the nForce 4 Ultra, not the 570 SLI. Or maybe you meant the KN9 SLI...

It was the SLI, sorry for the mistake. Corrected.
I wrote the author an email asking him about the Asus CnQ setting.

Thanks. 10W is a pretty significant difference, now why does nobody in Switzerland have Abit boards???? I might just have to import it, but the import duties kill you here...

cAPSLOCK wrote:Thanks. 10W is a pretty significant difference

The MSI with enabled CnQ should have the same numbers as the Abit.

So the conclusion then is Gigabyte sucks?

The load power levels seem to indicate that the gigabyte board is more efficient though. The question now is: is the gigabyte just more efficient at full load and crap at low load, or has the gigabyte board got a good VRM, but loads of "extra" chips that use a noticible amount of power all the time.

Sorry to annoy you, but the gigabyte board you mentioned doesn't exist, these are the two likely candidates:
-GA-M57SLI-S4 [nForce 570 SLI] (there is no S5 version)
-GA-M59SLI-S5 [nForce 590 SLI] (this would explain the higher consumption at idle, and the 4-phase VRM would explain the smaller difference load makes)

If it is the GA-M59SLI-S5 (I really hope it is ) then there is a bit of science to this: the 570 chipsets use less juice than the 590, and a 4-phase VRM is more efficient than a 3-phase.

Looking over your post, and the pictures of the motherboards concerned, I have another, more plausible theory:
The gigabyte board is the GA-M57SLI-S4. If it was the 590, the power consumption difference between it and the Asus board in your post is too much. The reason it is less good than the Abit and MSI counterparts is because it has no heatsinks on the VRM mosfets.

On second thought, you would think that the heatsinks would make more of a difference with a heavy load.... ARRRRGGGGGGGG

Gigabyte boards are quite notorious for their additional power requirements.
The test says about the Gigabyte board that CnQ is not adjustable in BIOS. In past years, CnQ was enabeld by default on Gigabyte boards.

nforce4 Gigabyte board is an par with competition: http://forums.silentpcreview.com/viewto ... 833#215833

KT400 /K8T800/Nforce3 uses more power: http://forums.silentpcreview.com/viewto ... =7250#7250
http://forums.silentpcreview.com/viewto ... 2127#52127

Thanks for this thread. I was looking for AM2 ATI chipset boards with the SB600 (particularly the MSI K9AGM), but the only ones I can find right now in the U.S. are the ECS and MSI CF boards($).

The MSI K9NU with the ULI chipset looks very promising. The performance and features are awesome for a very low price. It's also a compact ATX, so it should be easier to mount in an Antec NSK4400.

Thanks again and keep an eye on that magazine for us.