low power SNB build - which cpu/mobo

[ddutta]

I am going to upgrade to SNB mainly due to the hope of low power. I am no gamer. I need a 24/7 low idle power box that will act my server and will do photoediting and computational stuff occasionally.

Seems SNB chips have 3W idle power but all reviews show an idle power of 70-80W. There was one review with a H67 Intel board which showed a 27.2W idle power.

Any ideas which cpu-mobo combo will be the best for my needs? Is a H67 going to give me lower power than a P67 + Nvidia 430 (6W idle)? Is a m-atx board going to consume less power due to less PCI slots etc?

 

[DaRuSsIaMaN]

SNB? Do you mean sandy bridge? Everyone else here has been calling it "SB," so just making sure.

Also, I think your confusion stems from the fact that the reviews measure idle power of the entire system. That is common to both the [H] review and the reviews at Tom's Hardware. The chip itself may be 3W idle, but how would one measure that? That review with H67 board probably achieved that low idle number due to not having a discrete graphics card in the system? Just a guess, though; I don't know what review you're citing, obviously.

Smaller mobos with fewer components will probably consume less power but the difference will probably be very minuscule, so it might not be worth the consideration. The cpu and choice of graphics are the biggest factors in power consumption. You may want to consider the T and S models of the new SB chips. I'm not sure what they are, exactly. But Tom's mentions them in passing in a few places, such as here. Apparently, they will be lower TDP variants, from what I gather.

 

[ddutta]

Ok SNB = SB .....

I am ok with either integrated graphics or with a fermi with low idle power (e.g. GT430). The Intel H67 board in the same review from Tom had an idle system power of around 30W with integrated graphics. Do you think we will get the T and S in the retail channels? Then that would be awesome (paired with one of these Intel boards)

They had a H55 board which consumed 30W on idle with a i3-530 while my Gigabyte i3-530 used to sip 55W .... with a 80% PSU. Seems one needs to buy Intel desktop mobo parts for low power .....

SNB? Do you mean sandy bridge? Everyone else here has been calling it "SB," so just making sure.

Also, I think your confusion stems from the fact that the reviews measure idle power of the entire system. That is common to both the [H] review and the reviews at Tom's Hardware. The chip itself may be 3W idle, but how would one measure that? That review with H67 board probably achieved that low idle number due to not having a discrete graphics card in the system? Just a guess, though; I don't know what review you're citing, obviously.

Smaller mobos with fewer components will probably consume less power but the difference will probably be very minuscule, so it might not be worth the consideration. The cpu and choice of graphics are the biggest factors in power consumption. You may want to consider the T and S models of the new SB chips. I'm not sure what they are, exactly. But Tom's mentions them in passing in a few places, such as here. Apparently, they will be lower TDP variants, from what I gather.

 

[DaRuSsIaMaN]

I am ok with either integrated graphics or with a fermi with low idle power (e.g. GT430).

Btw, why would you go for a fermi? I am under the impression that recent generations of AMD beats nVidia hands down in terms of power usage. Certainly the higher end 5000 series and the 6000 series outperform nvidia in both load and idle scenarios... although I don't remember comparisons specifically between the low end parts. But I'm pretty sure the AMD advantage would carry over since it's all the same architecture.

They had a H55 board which consumed 30W on idle with a i3-530 while my Gigabyte i3-530 used to sip 55W .... with a 80% PSU. Seems one needs to buy Intel desktop mobo parts for low power .....

.... I seriously doubt that the practically 100% difference in power consumption can be attributed to just a different motherboard! No way! Did you check the rest of their test system and how it compares to yours?

Do you think we will get the T and S in the retail channels?

I have no idea about that heh.

 

[ddutta]

Btw, why would you go for a fermi? I am under the impression that recent generations of AMD beats nVidia hands down in terms of power usage. Certainly the higher end 5000 series and the 6000 series outperform nvidia in both load and idle scenarios.

I run Linux and AMD drivers are generations behind Else I would have probably gone for AMD. GT 430 is 6W idle which is ok for my uses.

.... I seriously doubt that the practically 100% difference in power consumption can be attributed to just a different motherboard! No way! Did you check the rest of their test system and how it compares to yours?



I have no idea about that heh.

The rest of the system is identical except for the graphics stuff. Also check this article on efficiency
http://www.guru3d.com/article/msi-p67gd65-motherboard-review/9 in addition to
http://www.tomshardware.com/reviews/sandy-bridge-efficienct-32-nm,2831-7.html

One way to explain this is that the lower end matx Intel boards are better optimized for power under normal use and they have less irrelevant (to non OCers) chips. Note the difference of 12W as you go from the matx Intel board to the atx board with the same GPU and the CPU.

 

[DaRuSsIaMaN]

Also, you might get more answers asking around on some other forums, like Tom's or maybe silentpcreview.com. SPCR forums might be best. I get the impression most people here don't give a shit about having a low power system. I myself care about these things and, like you, am not really a gamer. But vast majority of people here are gamers and are mostly interested in building monster gaming PCs.

 

[ddutta]

Also, you might get more answers asking around on some other forums, like Tom's or maybe silentpcreview.com. SPCR forums might be best. I get the impression most people here don't give a shit about having a low power system. I myself care about these things and, like you, am not really a gamer. But vast majority of people here are gamers and are mostly interested in building monster gaming PCs.

Yes I think I will try elsewhere too for these answers. Though I have found this forum very useful for my other needs e.g. display etc.

 

[needmorecarnitine]

Any ideas which cpu-mobo combo will be the best for my needs?

All you are going to get now is guesses. Why not wait a few weeks for a good answer...

 

[pxc]

Low power versions coming February 20 for LGA 1155:

Core i3 2100T 2c/4t 2.5GHz(no turbo) 35W
Core i5 2390T 2c/4t 2.7/3.5GHz 35W
Core i5 2400S 4c/4t 2.5/3.3GHz 65W
Core i5 2500T 4c/4t 2.3/3.3GHz 45W
Core i5 2500S 4c/4t 2.7/3.7GHz 65W
Core i7 2600S 4c/8t 2.8/3.8GHz 65W

All are a bit more expensive than the regular power versions. Unless you have a SFF case, IMO it's not worth it especially if your computer will mostly be running a light load or is otherwise mostly idle as with typical use.

 

[DaRuSsIaMaN]

The rest of the system is identical except for the graphics stuff. Also check this article on efficiency
http://www.guru3d.com/article/msi-p67gd65-motherboard-review/9 in addition to
http://www.tomshardware.com/reviews/sandy-bridge-efficienct-32-nm,2831-7.html

One way to explain this is that the lower end matx Intel boards are better optimized for power under normal use and they have less irrelevant (to non OCers) chips. Note the difference of 12W as you go from the matx Intel board to the atx board with the same GPU and the CPU.

Hmm... well the guru3d article seems to support what I was saying, though. Assuming everything else stayed equal, the system with the MSI P67 GD65 motherboard used only 1 more Watt over the system using the Intel DP67BG motherboard: 76W vs 75W at idle. (I'm comparing the middle pair of bars with the second from the right.) You're saying a difference of 30W ... at idle. I still don't see how that could possibly be due to just the motherboard alone even if the board is better optimized and with fewer features. I'd say a maximum of 5W difference would be possible to that factor alone. Like how the ASUS board is giving 79W idle vs. 75W idle on the intel board. Maybe the way you measured idle wasn't truly idle? Or maybe your PSU is just that much less efficient than whatever PSU they used in those reviews. Remember, PSU quality and its maximum wattage can make a huge difference at those low wattage numbers because PSU efficiency drops off a lot below like 10% of its maximum rating.

Also, what are you referring to that shows 12W difference? The 152W vs 144W? That's only 8.

Also also ... which mATX board you referring to? That model of Intel board, the DP67BG, is a normal ATX.

 

[DaRuSsIaMaN]

Low power versions coming February 20 for LGA 1155:

Core i3 2100T 2c/4t 2.5GHz(no turbo) 35W
Core i5 2390T 2c/4t 2.7/3.5GHz 35W
Core i5 2400S 4c/4t 2.5/3.3GHz 65W
Core i5 2500T 4c/4t 2.3/3.3GHz 45W
Core i5 2500S 4c/4t 2.7/3.7GHz 65W
Core i7 2600S 4c/8t 2.8/3.8GHz 65W

All are a bit more expensive than the regular power versions. Unless you have a SFF case, IMO it's not worth it especially if your computer will mostly be running a light load or is otherwise mostly idle as with typical use.

Hey thanks, that's helpful. I was wondering about those models too.

So you're saying that the lower power would really only be lower at load? If comparing the 2600S to the 2600, for instance, (stock clock) you would not see much difference at idle? I know obviously we can't say for sure right now but is that your guess?

 

[pxc]

Hey thanks, that's helpful. I was wondering about those models too.

So you're saying that the lower power would really only be lower at load? If comparing the 2600S to the 2600, for instance, (stock clock) you would not see much difference at idle? I know obviously we can't say for sure right now but is that your guess?

Yes, because at idle the clock speed/voltage is decreased for both normal power and low power models (unless of course power saving is disabled). There should be little power difference at low load or idle. At full load is where you would see significant power savings.

Typical use will have the computer running at different loads and overall the low power version will be more energy efficient. I just can't justify that savings with the price difference. Buy the lower power versions if you must due to small form factor or if you need to run it mostly under heavy load.