Power consumption of mainboard

[Hudu]

I first have to blame myself to ask this question Mr. Nielsen about the Asus P9D-M in a private conversation.
(I recognize today that it is against the forum rules. :( )
I search the Internet several days about the topic, but did not find any answer.

So I want to ask:

What is the power consumption at idle state without drives of the following mainboards:

1. Asus P9D-M/P9D-MV
2. Intel S1200V3RP
3. Supermicro X10SLH-F
4. Supermicro X10SL7-F (with and without enables LSI 2308)

Background: In Germany power consumption is not negligible, so I want so plan the server well.

 

[cyberjock]

I doubt anyone has that information.. and there's now way to test a mainboard by itself.

Things such as:

1. Quantity of RAM
2. CPU used
3. BIOS settings
4. PSU used
5. OS used

All have significant factors into how much power it draws idle. So your comparison is somewhat apples-to-oranges without somewhat normalizing for those.

 

[SmallGuy]

I have do a little of maths:
Imagine there is a difference of 10W between the better and the worst (witch is very much), with a 0.0883 Euros per KWh without tax (french price), you are talking about 10 euros tax include per years cost saving, so less than 1 Euros per month, equal to a coffee price.:)
I don't really know the price of the electricity in Germany, but I think this is the order of magnitude.
So, I'm not convinced this factor has to be taken into account as prior argument to make your motherboard choice: Buy the most powerful and drink less coffee. :D

 

[HoneyBadger]

Hi Hudu, I recognize you from the other thread about optimal RAIDZ2 sizing.

If low power consumption is a major concern, I would look into one of the newer Avoton Atom based motherboards like the ASRock C2550D4I

http://www.asrockrack.com/general/productdetail.asp?Model=C2550D4I

Max TDP (thermal design power) on the embedded Atom processor is 14W.

Several users have boards based on this and have no problem saturating single gigabit interfaces, even when using drive encryption.

 

[Hudu]

I have do a little of maths:
Imagine there is a difference of 10W between the better and the worst (witch is very much), with a 0.0883 Euros per KWh without tax (french price), you are talking about 10 euros tax include per years cost saving, so less than 1 Euros per month, equal to a coffee price.:)
...

In Germany you pay this year 0.27 Euros per kWh. So 10 Watt power consumption costs 23.1 Euro per year, if you use it 24/7.
But if all the motherboards a within a 10 Watt difference, I would not think about it, since i have to calculate my work evaluating it.
My fear is that I decided to use motherboard, which uses 50 Watt instead of one, which uses 20 Watt. So we are talking about 70 Euro per year without advantage.

 

[Hudu]

Hi Hudu, I recognize you from the other thread about optimal RAIDZ2 sizing.
If low power consumption is a major concern, I would look into one of the newer Avoton Atom based motherboards like the ASRock C2550D4I
http://www.asrockrack.com/general/productdetail.asp?Model=C2550D4I
Max TDP (thermal design power) on the embedded Atom processor is 14W.
Several users have boards based on this and have no problem saturating single gigabit interfaces, even when using drive encryption.

This board is a good example. I know this board. The ASRock C2750D4I was tested by the German paper "ct" and they
found, that Asrock used an old and energy hungry pcie chip to connect the two additional sata chips. As a consequence the
Asrock C2750D4I has the same power consumption like a Supermicro A1SAi-2750F with a IBM M1015.
(I will prefer the second. The C2550 variant is on my list.)

 

[Hudu]

I doubt anyone has that information.. and there's now way to test a mainboard by itself.

Things such as:

1. Quantity of RAM
2. CPU used
3. BIOS settings
4. PSU used
5. OS used

All have significant factors into how much power it draws idle. So your comparison is somewhat apples-to-oranges without somewhat normalizing for those.

But you can normalize the value, if you know the configuration:

Memory per Slot uses (at IDLE) less than 1 Watt.
The new Haswell-CPUs have at IDLE roughly the same power consumption. The Pentium is 1 Watt higher, because it does not use the C7-State.
OS (and drivers) used matters. It is feasible that the administrator optimized the BIOS (and OS and drivers) hopefully.
PSU is quite unpredictable, because at low power consumption the PSU are very different efficient.

The power consumption should not be exact on digits after the comma, but you get a feeling how the motherboard behaves and what is achievable.
(And you see what to avoid...)

 

[Hudu]

So bought some parts:

Case: Fractal Design DEFINE R4 Black Pearl
PSU: 360 Watt Seasonic G Serie 360 Non-Modular 80+ Gold
Harddisk: 5x 3000GB WD Red WD30EFRX 3.5"
Memory: 2x 8GB Kingston ValueRAM DDR3L-1600 ECC DIMM CL11 Single (KVR16E11)
SSD: 128 GB Samsung 840 Pro

For the motherboard I am still double-minded:
I think I go the Avoton (C2550) track and have to decide between the Supermicro A1SAM-2550F (234 Euro, IDLE: 15 Watt) and the ASRock C2550D4I (295 Euro, IDLE: 22 Watt).
Both are compatible with the memory, have the ability to saturate a 1GB-network link and can transcode two PLEX-stream simultaneously.

Since the price of both boards are different, I think I will buy the Supermicro and supply a IBM M1015 (IDLE: 7 Watt), if I need more SATA ports. With the price and
the IDLE-power I will get roughly to the same values for both board.

Is this a good decision?

 

[SmallGuy]

Your researches are good, why would you we take a decision for you?

 

[Hudu]

Your researches are good, why would you we take a decision for you?

I have to make the decision alone, but I do not deny the power other people. :)

 

[warri]

Since the price of both boards are different, I think I will buy the Supermicro and supply a IBM M1015 (IDLE: 7 Watt), if I need more SATA ports. With the price and
the IDLE-power I will get roughly to the same values for both board.

Is this a good decision?

In this forum a lot more people seem to trust a Supermicro board combined with a M1015 instead of an ASRock board. So I'd say it is a good decision ;)