desktop Nvidia 6150 power efficency
Moderators: NeilBlanchard, Ralf Hutter, sthayashi, Lawrence Lee
desktop Nvidia 6150 power efficency
Im looking at building a desktop that is running off solar power.
6150 motherboard and a AM2 X2 3800+. I have some questions..
- Does the desktop 6150 support all the same power modes as the 6150go? In partciular the DFI board.
- Does the SATA shutdown now bring the SATA power consumption down to PATA levels or is PATA still superior?
- What chipset and CPU voltages does the DFI 6150 AM2 board support? I know it supports CPU undervolting but no one manages to tell me by how much or if the chipset is able to be undervolted.
Also I am thinking of getting a 7300 GT with DDR3. Do these desktop cards support the power saving tools and modes of the mobile 7x00go versions?
I am really keen to sqeeze every W out of this system. I want to get idle <30w and peak <50w. Any other tips or suggestions would be helpful.
6150 motherboard and a AM2 X2 3800+. I have some questions..
- Does the desktop 6150 support all the same power modes as the 6150go? In partciular the DFI board.
- Does the SATA shutdown now bring the SATA power consumption down to PATA levels or is PATA still superior?
- What chipset and CPU voltages does the DFI 6150 AM2 board support? I know it supports CPU undervolting but no one manages to tell me by how much or if the chipset is able to be undervolted.
Also I am thinking of getting a 7300 GT with DDR3. Do these desktop cards support the power saving tools and modes of the mobile 7x00go versions?
I am really keen to sqeeze every W out of this system. I want to get idle <30w and peak <50w. Any other tips or suggestions would be helpful.
Re: desktop Nvidia 6150 power efficency
Don't know anything about the mobile counterpart. It's well known that ATI's chipsets use less power in general though. The mobo you're mentioning may be a good one, just remember that it's not micro ATX. It's 20 mm too wide. I'm really curious why DFI did it like that, and why they called it µATX until recently. They have 2 upcoming µATX AM2 mobos with ATI graphics, one with the old RS485, and also one with the brand new RS690. The latter have X700 graphics and probably uses more power, but I just wanted to tell you what's new.Phido wrote:- Does the desktop 6150 support all the same power modes as the 6150go? In partciular the DFI board.
If you don't need the integrated graphics then why buy the hottest one (nVidia)? Get a 945GM mobo (mobile chipset) and a mobile C2 instead.
Why do you need a separate graphics card anyway?
It's actually 89 W, even though there are 65 and 35 W low power models (almost) available too (but you knew that ).jaganath wrote:Unless you are using the AM2 35W X2 3800 that looks impossible; the CPU alone maxes out at 65W IIRC?I want to get idle <30w and peak <50w
-
- *Lifetime Patron*
- Posts: 1809
- Joined: Sat Apr 24, 2004 1:45 am
- Location: At Home
Re: desktop Nvidia 6150 power efficency
If you want to use a discrete VGA card I can’t see that it’s possible to achieve your desired power figures.Phido wrote:Also I am thinking of getting a 7300 GT with DDR3. Do these desktop cards support the power saving tools and modes of the mobile 7x00go versions?
I am really keen to squeeze every W out of this system. I want to get idle <30w and peak <50w. Any other tips or suggestions would be helpful.
I managed 50W at load with a Core Duo 1.66GHz using an IGP and this could be bettered significantly with a power supply that is more efficient at such low levels.
A PicoPSU would be a big help as at these low power levels the efficiency of the power supply is as important as just about anything else.
30W at idle will be harder to achieve I think. With integrated graphics, a PicoPSU and a Core Solo or Duo you should get pretty close.
Using a 2.5â€
-
- Posts: 140
- Joined: Wed Jan 18, 2006 4:57 am
- Location: France, Lyon
Re: desktop Nvidia 6150 power efficency
Why don't you want to use a laptop ?
They have lower power consumption by design, and anyway you will have to fill your desktop with laptop parts to reach your power consumption goal, so any complaint about low capacity hard-drive or expensive components is null.
Besides, you also have to add the LCD monitor consumption to your 30w-50w target, whereas a laptop can reach these figures even with the display taken into account I think. Not to mention the security (exploding battery jokes aside) provided by the battery in conjunction with a fluctuating solar power source (as I have no idea about your solar system).
Can you elaborate a bit on what you plan to do with the computer, because the CPU choice aswell as the discrete graphic card choice seem unoptimal as reflected by previous messages.
>smilingcrow
Are you sure the power brick associated with the picoPSU would have a better efficiency (even at such a low load) compared to a regular fan-cooled PSU ?
They have lower power consumption by design, and anyway you will have to fill your desktop with laptop parts to reach your power consumption goal, so any complaint about low capacity hard-drive or expensive components is null.
Besides, you also have to add the LCD monitor consumption to your 30w-50w target, whereas a laptop can reach these figures even with the display taken into account I think. Not to mention the security (exploding battery jokes aside) provided by the battery in conjunction with a fluctuating solar power source (as I have no idea about your solar system).
Can you elaborate a bit on what you plan to do with the computer, because the CPU choice aswell as the discrete graphic card choice seem unoptimal as reflected by previous messages.
>smilingcrow
Are you sure the power brick associated with the picoPSU would have a better efficiency (even at such a low load) compared to a regular fan-cooled PSU ?
-
- *Lifetime Patron*
- Posts: 1809
- Joined: Sat Apr 24, 2004 1:45 am
- Location: At Home
Re: desktop Nvidia 6150 power efficency
If the issue is noise related, then typically only laptops using an ULV Intel CPU are silent under full load and only then if they are fanless.Le_Gritche wrote:Why don't you want to use a laptop?
You make the point that I failed to, which is that the efficiency of the power brick used with a PicoPSU is critical in the scenario that I described. Thanks.Le_Gritche wrote:>smilingcrow
Are you sure the power brick associated with the picoPSU would have a better efficiency (even at such a low load) compared to a regular fan-cooled PSU ?
Yeah, that's the way it used to be. But now I'm not sure. Since AMD uses individual TDP for both S939 and AM2, and also cherrypicks low TDP AM2 CPU's for their EE lineup it's hard to tell. Does this mean that the average S939 X2 have a lower individual TDP than the AM2 counterpart?jaganath wrote:Yes, the TDP for the model type is 89W, but I assumed the max power consumption for the lowest-clocked model would be less than the nominal TDP (as with all other AMD chips).It's actually 89 W, even though there are 65 and 35 W low power models (almost) available too (but you knew that
Do you know?
I am moving out to a property where there is no grid power. While that is reason enough, I also will be adding a carputer and building several projects that require oddshaped computers.
This particular computer will be:
-My main work box
-My main gamming box
-run as a HTPC
I am going to underclock. I am looking at underclocking/undervolting both the chipset/CPU. Given that these basic components are used unchanged from mobile versions I don't see why desktop consumption would be much higher than a laptops except in areas of larger CPU power circuitry.
All this is going to be fed by 12v dc (from the solar and battery banks) through a 12v dc-dc psu.
I'm seeing diffrent arguments with regards to what platform.
Coreduo seems to offer lower peak power. But the Igp is slower, the idle is worse, and the price is more.
AMD seems to offer lower idle power (more important overall). Sometimes.
Amd boards also offer clock/voltage options for chipset and cpu which could tip the balance.
As this is going to be my main box the IGP has to atleast be useable. While I may wait for the ati x700 IGP, theres no proof it will actually be as fast as a x700 desktop, nor as power efficent as most other IGP.
I may be better off choosing a intel IGP with 6200tc than a new power hungery IGP like the ati.
Basically if you wanted to build the most power efficent desktop. What components would you use.
I already have a laptop. The laptop doesn't offer the flexability, responsiveness, price etc I am looking for. As there is also no internet, I want large local storage. Nor are laptops quiet either. This is going completely fanless with a passive water cooler.Why don't you want to use a laptop
This particular computer will be:
-My main work box
-My main gamming box
-run as a HTPC
I am going to underclock. I am looking at underclocking/undervolting both the chipset/CPU. Given that these basic components are used unchanged from mobile versions I don't see why desktop consumption would be much higher than a laptops except in areas of larger CPU power circuitry.
All this is going to be fed by 12v dc (from the solar and battery banks) through a 12v dc-dc psu.
I'm seeing diffrent arguments with regards to what platform.
Coreduo seems to offer lower peak power. But the Igp is slower, the idle is worse, and the price is more.
AMD seems to offer lower idle power (more important overall). Sometimes.
Amd boards also offer clock/voltage options for chipset and cpu which could tip the balance.
As this is going to be my main box the IGP has to atleast be useable. While I may wait for the ati x700 IGP, theres no proof it will actually be as fast as a x700 desktop, nor as power efficent as most other IGP.
I may be better off choosing a intel IGP with 6200tc than a new power hungery IGP like the ati.
Basically if you wanted to build the most power efficent desktop. What components would you use.
unless some major OEM starts making SFFs with EE X2's, then i really doubt it will make any significant difference atall. are these even available in retail now? as iirc, they weren't last time they came up..Does this mean that the average S939 X2 have a lower individual TDP than the AM2 counterpart? Confused
i would think the ATI chipset would be a good move; ATI 'traditionally' have had low powered (fanless) chipsets... and i don't see why the x700 would be any less powerful than the separate version (atleast, than a 'TC' version of it)- if the GPU (not memory) was less powerful, then surely they'd call it the x650 or something; similarly to what nV have done with the 6100/6150.
Would it work OK if u undervolted the entire computer?? i mean; if a dc-dc board with a 12v input was given 11.5v? -as PSUs are only required to be within 5% right?
There is alot of power circuitry on boards etc these days so dropping your 12v line from the PSU will have very little effect on other parts of the system. With a lower voltage, the amp draw most likely will increase and you could burn out traces on your mobo.
These PICO PSU I think pass the 12v directly onto the mobo and the 3.3 and 5v lines are fed off little regulating diodes. Why the hell are they so expensive?
These PICO PSU I think pass the 12v directly onto the mobo and the 3.3 and 5v lines are fed off little regulating diodes. Why the hell are they so expensive?
-
- Posts: 53
- Joined: Sun Aug 06, 2006 2:35 pm
Diodes? No, the Picos use DC-DC buck converters. $50 is a good price for what's involved. The parts and engineering skill to make these things so small, powerful and efficient do not come cheap.Phido wrote: These PICO PSU I think pass the 12v directly onto the mobo and the 3.3 and 5v lines are fed off little regulating diodes. Why the hell are they so expensive?
I have ruled out the 35W X2. It is too expensive. For what is just a undervolted AM2. I can undervolt myself and save my money.
which brings me back to:
http://www.silentpcreview.com/article313-page5.html
Its a shame Conroe measurements werent directly avalible.
Also its clear that the AMD cores are severly limited by the Asus platform with no undervolting options.
I suppose I am favouring the nv6150 and a AMD x2 over the i950 and a e6300. I feel that I could proberly live with the 6150 performance while the 950 would require another video. Althought promise of a IGP with x700 performance makes me hesitant to jump but is a argument in AMD's favour.
However its clear that laptops have lots of little power saving features which add up to be a conciderable advantage. So-dimm, 2.5" HD, enhanced bios's, enhanced power circuitry and overall design. Of course the more of these you use the more your basically building a laptop and the constraints of that.
I don't think there is enough information on the ground for this one.
which brings me back to:
http://www.silentpcreview.com/article313-page5.html
Its a shame Conroe measurements werent directly avalible.
Also its clear that the AMD cores are severly limited by the Asus platform with no undervolting options.
I suppose I am favouring the nv6150 and a AMD x2 over the i950 and a e6300. I feel that I could proberly live with the 6150 performance while the 950 would require another video. Althought promise of a IGP with x700 performance makes me hesitant to jump but is a argument in AMD's favour.
However its clear that laptops have lots of little power saving features which add up to be a conciderable advantage. So-dimm, 2.5" HD, enhanced bios's, enhanced power circuitry and overall design. Of course the more of these you use the more your basically building a laptop and the constraints of that.
I don't think there is enough information on the ground for this one.
It's most likely not only undervolted, but also cherry picked for it's low power consumption. Some people say that they actually are Turion cores, but I don't know anything about that, and CPU-Z doesn't reveal any useful info. But still, they're overpriced, especially compared to the 65 W 4600+.Phido wrote:I have ruled out the 35W X2. It is too expensive. For what is just a undervolted AM2. I can undervolt myself and save my money.
What Asus platform?Phido wrote:Also its clear that the AMD cores are severly limited by the Asus platform with no undervolting options.
What about the current ATI IGP? It draws less power than the nVidia counterpart.Phido wrote:I suppose I am favouring the nv6150 and a AMD x2 over the i950 and a e6300. I feel that I could proberly live with the 6150 performance while the 950 would require another video. Althought promise of a IGP with x700 performance makes me hesitant to jump but is a argument in AMD's favour.
-
- *Lifetime Patron*
- Posts: 1809
- Joined: Sat Apr 24, 2004 1:45 am
- Location: At Home
Looking at the figures in the SPCR review that you linked has me scratching my head why you would go with an AMD X2 and a 6150 motherboard!Phido wrote:which brings me back to:
http://www.silentpcreview.com/article313-page5.html
I suppose I am favouring the nv6150 and a AMD x2 over the i950 and a e6300. I feel that I could proberly live with the 6150 performance while the 950 would require another video. Althought promise of a IGP with x700 performance makes me hesitant to jump but is a argument in AMD's favour.
The single core Athlon 64 1.8 using the 6150 mobo, IGP, 1GB RAM & a notebook drive gives these figures for idle/load – 48/61W
In contrast, I tested a Core Duo 1.66GHz with 2GB RAM, 250GB desktop drive & a Nvidia 7600GS using the same utility to get the load figures (CPUBurn) – 49/62W
Taking into account the extra RAM and desktop drive would put the Core Duo at a lower power level and with the advantage of much better VGA performance. And this is just comparing to a single core Athlon 64. If you add the extra wattage of even the 35W X2 and the difference becomes quite large.
If the 7600GS consumes too much power for you whilst gaming, you could always go with a 6200TC or 7300LE, which should be less power hungry whilst gaming and they are all within 2W of each other at idle.
I'm not so sure about that. You can see that the actual 3000+ they're using in the review have a TDP = 44 W.smilingcrow wrote:And this is just comparing to a single core Athlon 64. If you add the extra wattage of even the 35W X2 and the difference becomes quite large.
The 35 W X2 uses 35 W maximum, but the individual TDP can actually be lower, just like 67 W is max for the 3000+.
-
- *Lifetime Patron*
- Posts: 1809
- Joined: Sat Apr 24, 2004 1:45 am
- Location: At Home
I chose to look at the actual wattages recorded by SPCR. Using these figures for idle/load:Mats wrote:I'm not so sure about that. You can see that the actual 3000+ they're using in the review have a TDP = 44 W.smilingcrow wrote:And this is just comparing to a single core Athlon 64. If you add the extra wattage of even the 35W X2 and the difference becomes quite large.
The 35 W X2 uses 35 W maximum, but the individual TDP can actually be lower, just like 67 W is max for the 3000+.
Athlon 64 1.8GHz = 4.8/20.5W
Athlon X2 2.0GHz = 5.6/41.4W
Both of the chips tested are at the lower end of the scale in terms of having a lower individual TDP, so these are best case scenarios really. So a standard X2 3800 is going to add 1W/21W to the previous system setup that I listed at idle/load.
If you add this extra wattage along with that for bringing the spec up to the same as for the Core Duo system you get:
Load
CPU 21W
RAM 1W
Hard Drive 5W (difference between laptop and desktop drives at idle)
Total 27W
Idle
CPU 1W
RAM 1W
Hard Drive 5W
Total 7W
This gives the X2 system figures of 55W/88W versus 49/62W for the Core Duo with a 7600GS.
As for the 35W X2 chip, since it’s already under-volted to 1.025/1.075V, I think it’s reasonable to say that it won’t have much headroom to under-volt. Being generous, let’s say it can go as low as 25W actual power consumption.
This then gives a 41.4 – 25 = ~16W advantage over the stock X2 at load and if it halves the idle power consumption it gets a 3W benefit at idle. Which equates to:
X2 = 52/72W versus S479 = 49/62W (7600GS) or 39/48W (IGP)
I've been looking at the SPCR figures and the one from LC over at:
http://www.lostcircuits.com/cpu/low_e/5.shtml
They are showing the "35w" 3800+x2 to be totally kicking arse in power idle/peak/power per w..
A 1/3 of intel at idle, simular at peak. Holding its own in benchmarks against the e6300..
Somethings not right... I belive the LC figures were swaying me towards AMD...
The Asus platform is the asus 6150 boards which do not feature any undervolting/underclocking bios options.
http://www.lostcircuits.com/cpu/low_e/5.shtml
They are showing the "35w" 3800+x2 to be totally kicking arse in power idle/peak/power per w..
A 1/3 of intel at idle, simular at peak. Holding its own in benchmarks against the e6300..
Somethings not right... I belive the LC figures were swaying me towards AMD...
The Asus platform is the asus 6150 boards which do not feature any undervolting/underclocking bios options.
You still need software to undervolt the lower C'n'Q speeds from stock. Besides, using anything else than default Vcore in BIOS often prevents software undervolting in nVidia systems, not sure about this actual board though.Phido wrote:The Asus platform is the asus 6150 boards which do not feature any undervolting/underclocking bios options.
I guess you've seen this review?
They're actually not that low, even for being standard models.smilingcrow wrote:Both of the chips tested are at the lower end of the scale in terms of having a lower individual TDP, so these are best case scenarios really.
3000+ 44.1 W, thermal profile B: 20.6 - 67 W.
3800+ 65.6 W, thermal profile P: 21.9 - 89 W.
Those figures says it all, in the end you're interested in total power draw.X2 = 52/72W versus S479 = 49/62W (7600GS) or 39/48W (IGP)
I still think it's hard to beat the low power figures of Merom, which also cost a lot less. If you buy a system with a HOT IGP (nVidia) you're stuck with that, but if you're running a 945GM + graphics card system you can always remove the graphics card later on if the sun won't shine on you.
Lower power draw, cost less, better graphics.
BTW do you really need dual core?
Personally, I find it pretty hard to get my head around some of the Opteron figures posted from the X2 TDP poll thread. These indicate that there are some 2,000MHz+ AMD64 chips out there with a TDP of just 35W at the standard voltage of 1.35V! Undervolt one of these (and possibly underclock slightly) and I'd have thought you ought to be able to get it running under 20W at full load.
The problem is, of course, that it's impossible to tell just what TDP any standard non-EE AMD CPU has without buying the damn thing first and then running TCaseMax!
Personally, I was thinking about building a Core2Duo system but I'm now moving towards the tightwad route - i.e. get an AM2 motherboard + a cheap Sempron or Athlon64, hoping for one with a TDP in the region of 35-40W which I can then undervolt (I'm aiming for a passive or semi-passive system and performance isn't of vast importance). I believe that Sempron X2 is due to be released on the 65nm process during the first half of next year so I can always go for one of those if I'm looking for improved performance in six months time. When you consider how well AMD have done with power consumption on the 90nm process I have great hopes for their 65nm chips!
The problem is, of course, that it's impossible to tell just what TDP any standard non-EE AMD CPU has without buying the damn thing first and then running TCaseMax!
Personally, I was thinking about building a Core2Duo system but I'm now moving towards the tightwad route - i.e. get an AM2 motherboard + a cheap Sempron or Athlon64, hoping for one with a TDP in the region of 35-40W which I can then undervolt (I'm aiming for a passive or semi-passive system and performance isn't of vast importance). I believe that Sempron X2 is due to be released on the 65nm process during the first half of next year so I can always go for one of those if I'm looking for improved performance in six months time. When you consider how well AMD have done with power consumption on the 90nm process I have great hopes for their 65nm chips!
-
- *Lifetime Patron*
- Posts: 1809
- Joined: Sat Apr 24, 2004 1:45 am
- Location: At Home
The SPCR review shows a Core Duo idling at 1.5W so there’s not much room for a real gain there. The pricing for Core Solo is so close to Core Duo that’s it seems redundant. The advantage of the dual cores are that when using multi-threaded applications it will not only performance quicker but also more efficiently in the power sense. So it has a small downside and a big upside. I do like a big upside.Mats wrote:BTW do you really need dual core?
I stand corrected.Mats wrote:They're actually not that low, even for being standard models.smilingcrow wrote:Both of the chips tested are at the lower end of the scale in terms of having a lower individual TDP, so these are best case scenarios really.
I also just noticed that the S939 motherboard used in the SPCR review was limited in its under-volting potential and could go no lower than 1.18V, as opposed to the normal limit of 1.1V. This would close the gap between the two platforms.
Some of the LC figures look very dubious to me. They show an E6300 CPU as consuming 52.8W when running Prime95. I measured a whole system with the faster E6400 as only consuming 77W when running dual Prime95.Phido wrote:I've been looking at the SPCR figures and the one from LC over at:
http://www.lostcircuits.com/cpu/low_e/5.shtml
They are showing the "35w" 3800+x2 to be totally kicking arse in power idle/peak/power per w..
A 1/3 of intel at idle, simular at peak. Holding its own in benchmarks against the e6300..
Somethings not right... I belive the LC figures were swaying me towards AMD...
Admittedly, my results are using an under-volted CPU and theirs aren’t. But if you’re looking for a low power system it’s the under-volted figures that count.
I agree. I was asking the question to Phido though, because he seems to prefer AMD. A Sempron clould be an alternative. While I'm writing this I realize it's getting harder and harder not to recommend C2, either mobile or desktop.smilingcrow wrote:The SPCR review shows a Core Duo idling at 1.5W so there’s not much room for a real gain there. The pricing for Core Solo is so close to Core Duo that’s it seems redundant. The advantage of the dual cores are that when using multi-threaded applications it will not only performance quicker but also more efficiently in the power sense. So it has a small downside and a big upside. I do like a big upside.Mats wrote:BTW do you really need dual core?
About 65 nm, I'm note sure that we will see it used for low end dual core first. T64 and FX CPU's will get it first, where AMD can't keep up with Intel and the need is much bigger. My guess is that we have to wait at least 9 months for regular X2's.
-
- *Lifetime Patron*
- Posts: 1809
- Joined: Sat Apr 24, 2004 1:45 am
- Location: At Home
Don’t AMD traditionally release a new and smaller process using the lower speed chips first! This makes sense from the perspective of yields as they may struggle to get good yields for the higher clocked parts initially.Mats wrote:About 65 nm, I'm note sure that we will see it used for low end dual core first. T64 and FX CPU's will get it first, where AMD can't keep up with Intel and the need is much bigger. My guess is that we have to wait at least 9 months for regular X2's.
I agree that AMD really need to address the higher end, but unless they are very confident that they’ve fully mastered the 65nm process it’s a bit of a risk. Also, most of their sales will be with the low to mid range clock speeds, so that’s where they need to improve their margins. With the top end parts, the margins are so high that the manufacturing cost is much less of an issue.
When you add in AMD’s need to ramp production to feed the Dell monster, I think 65nm will come at lower clock speeds from introduction.
For low cost, a Sempron looks good with the right motherboard. Or even a Dothan sourced from eBay, where they can go for peanuts might be attractive if you can get a suitable motherboard for a reasonable price.Mats wrote:I agree. I was asking the question to Phido though, because he seems to prefer AMD. A Sempron clould be an alternative. While I'm writing this I realize it's getting harder and harder not to recommend C2, either mobile or desktop.
That's the main problem at the moment - motherboards which support Intel mobile chips are generally pretty expensive. The AM2 motherboards are cheaper but current microATX AM2 motherboards don't seem to support undervolting which is obviously not ideal for silencing!smilingcrow wrote:For low cost, a Sempron looks good with the right motherboard. Or even a Dothan sourced from eBay, where they can go for peanuts might be attractive if you can get a suitable motherboard for a reasonable price.
I do find it very disappointing that Intel have again ensured that upgradability of motherboards which support their mobile CPUs isn't possible by rewiring the sockets - bizarrely we now have two different socket-479s from Intel, one of which supports Dothan/Banias and the other of which supports Yonah/Merom, but never the twain shall meet! AM2 is at an advantage here as we have been told that the future AM3 socket chips should be backwards compatible, making an upgrade from single to dual-core on the same motherboard easily possible.
Yes I really want dual core. I was always planning to get a 2xsocket opteron before deciding to go bush and away from 240v outlets. Much/all of the software I use is multi threaded. I also want to do a few major tasks at once.
I already have a AXP mobile 2600+, which will be a 2nd computer for my GF. So single core I already have a okay single core cpu. I can clock it right down to 600Mhz and very low volts for a Geode style power consumption.
The DFI 6150 board seems to support extensive underclocking options of both the chipset and the CPU/Memory. It would seem to be the best platform for lowest AMD power use.
AMD also seem to have the bigger choice in mATX at the moment. While I would like smaller form factors its not a deal breaker.
I already have a AXP mobile 2600+, which will be a 2nd computer for my GF. So single core I already have a okay single core cpu. I can clock it right down to 600Mhz and very low volts for a Geode style power consumption.
The DFI 6150 board seems to support extensive underclocking options of both the chipset and the CPU/Memory. It would seem to be the best platform for lowest AMD power use.
AMD also seem to have the bigger choice in mATX at the moment. While I would like smaller form factors its not a deal breaker.