## Doing the Math on ECO

On MSI’s website, a number of statements are made by their internal testing.

One picture shows a regular MSI mATX motherboard based on a low cost design drawing 29W. It does not state if this is idle, load or what CPU is used, but based on our testing it could easily be idle numbers with a reasonable CPU and a discrete GPU installed.

Next to the 29W number is an ECO motherboard indicating it uses 22W. Next to this is the same motherboard but with ECO Center software enabled, and 17W being used. MSI’s 40% claim of power reduction comes from the difference between 17W and 29W (~41%). Any company publishing numbers on its own products have a vested interest in making them sound as good as possible, so I took these numbers to run a regular working year analysis on them. Follow the math with me, and see if you agree with my conclusions.

• Normal motherboard uses 29W
• ECO motherboard uses 22W, which is 7W fewer
• ECO with ECO Center software uses 17W, which is 12W fewer

The standard working year is 1920 hours long (8 hours per day, 5 days per week, 48 weeks per year).

This means that each motherboard uses the following amount of energy:

• Normal motherboard at 0.029 kW for 1920 hours uses 55.68 kWh
• ECO motherboard at 0.022 kW for 1920 hours uses 42.24 kWh
• ECO motherboard with Software at 0.017 kW for 1920 hours uses 32.64 kWh

At a price of \$0.12 per kWh, this means that:

• The normal motherboard costs 55.68 * \$0.12 = \$6.68 per year to run
• The ECO motherboard costs 42.24 * \$0.12 = \$5.07 per year, a saving of \$1.61
• The ECO+S motherboard costs 32.64 * \$0.12 = \$3.92 per year, a saving of \$2.76

So by using MSI’s own numbers, an ECO motherboard with ECO center enabled would save \$2.76 per year. If we extend that to a 3 year upgrade cycle (based on warranty), that means that the ECO motherboard should only cost an \$8.28 maximum difference over the base board. MSI told me that the ECO range would cost 20% more than the standard range, and if that number is accurate, it means that the non-ECO boards should only be \$41.40 (\$8.28 / 0.2), and the ECO boards should be \$49.88 maximum. If MSI were to move on to a five year warranty and this would extend to \$13.82 in savings, a \$69.12 base price and an \$83.94 ECO price.

One quick blast to Newegg shows that the B85M ECO is on sale for \$73.

One of the big problems with marketing components like motherboards as ‘eco-friendly’ is that when we are dealing in sub 10W savings, the cost of the extra components can vastly out-weigh the savings unless the product cycle is extended and supported accordingly.

Another interesting point to note is MSI’s quoted effect of the ECO Center software at load. Here MSI quotes 52.1W for an ECO motherboard at load which drops to 50.1W when the ECO Center software is enabled. It comes across a small drop, but that it important when it comes to our numbers.

Our Numbers

While MSI doesn’t state which motherboard they used to get the 29W value, we recently reviewed the MSI Z97I AC and thought that would be a good starting point. Being a mini-ITX motherboard it should have fewer components, but it is in the \$140 price bracket and thus has a controller or two.

Normally we do our power readings with the same power supply across all motherboards. This means that while the numbers might be a bit inefficient at the lower part of the spectrum, they are all on the same efficiency line. Using our stupidly overpowered 1200W Gold power supply (for example) and an i7-4770K, the comparison between the Z97I AC and the B85M ECO gives:

 Peak Power Consumption w/GTX 770 Equipped MSI Z97I AC MSI B85M ECO Long Idle 56W 47W Peak CPU Load (OCCT) 154W 129W

Even if you ignore the power efficiency, we have a 9W difference at idle and a 25W difference at load, which both works out to around 16%.

If we extrapolate 9W out in our mathematics, this equates to \$1.84 a year, whereas 25W is \$5.76. The more you load up the ECO system, compared to another loaded system, the more savings are to be had is a perfectly valid qualitative conclusion. The ECO certainly scores top marks in our standard motherboard power tests, coming top of our Z97 series by a good margin.

We tested the B85M Eco with four CPUs and two different power supplies with and without a discrete graphics card to get an understanding of how the power savings work. The power supply here is a 500W Platinum unit from Rosewill. All numbers given are peak power readings at the wall for the system as a whole.

Our testing shows that the ECO mode affects a 1-2W difference in peak loading, but this correlates with MSI’s own numbers when using an ECO board vs. using the ECO board with the software enabled.

It was interesting to see that when the integrated graphics were in play, the i7-4770K downclocked to 3.3 GHz to hit the 85W mark. If we directly compare the POV-Ray results of each CPU in normal and ECO mode, the T/S processors stay within 2% but the 4770K reduces significantly enough to be noticeable:

 POV-Ray Results POV-Ray BIOS Defaults POV-Ray in ECO Mode i3-4130T 605 596 i5-4570S 1205 1186 i7-4790S 1336 1333 i7-4770K 1541 1412

MSI claims 100% performance at all times, but our i7-4770K CPU did reduce in frequency to hit the power point. We passed our information on to MSI although we are waiting on an official response. Chances are this behavior will be fixed in a future BIOS update.

MSI B85M ECO Overview, Visual Inspection, Board Features MSI B85M ECO BIOS

• #### simonpschmitt - Wednesday, November 26, 2014 - link

I just found the board online for 68€ witch means one would break even after 2.2 Years.
• #### miksmi - Wednesday, November 26, 2014 - link

For servers, I moved to the mini-ITX form factor and am interested in an ECO version. I keep servers 8-10 years.
• #### hojnikb - Wednesday, November 26, 2014 - link

For office (and other non demanding uses) wouldn't it make more sense to go j1900 route rather than eco 1150 + celeron/pentium/i3 cpu ?
It will use way less power than 1150 + fanless by default (even more power savings).
• #### Ian Cutress - Wednesday, November 26, 2014 - link

For most office work, you're probably right. But a socketed platform does offer a potential upgrade path if the dynamics of the work change to something more computationally intensive. Also going down the LGA1150 route offers faster response times, which some businesses might argue is important when continuously dealing with emails and so on. It really depends on the scenario.
• #### Folterknecht - Wednesday, November 26, 2014 - link

"One of the points in the review was the inability to select a lower CPU voltage. Both voltage and frequency have a role in total system power consumption, but when full performance is still needed, voltage is the only variable left to modify. I posed this question to MSI, and received the following response:

“We actually did try to do some testing with lower CPU voltage settings. The reason why we didn’t include it into the current BIOS is because we think Intel’s current FIVR architecture puts too many limits inside their design and we [would] rather use Intel’s integrated power saving features like C-State (Up to C7) and also SVID power. But it’s still a good suggestion that we can request our R&D to do more testing and check if we can fine tune better settings to enhance the power saving ability.”"

Lazy excuse in my book, considering that its still possible to undervolt current generation Intel CPUs quite a bit, at least when it comes to load voltage susually something between 0.1 - 0.2 V. As a MB manufacturer I can imagine that it would even be possible to play around with everything between idle and full load voltage, something a normal user cant do.

So instead of waisting their time on hot marketing air, develop something along the lines of auto-OC software or as an option in BIOS, but instead of overclocking let it undervolt the CPU automatically until it fails. But please no "1-2-3 click ready nonsense" of predefined values, more along the lines of a small stress test, which lowers the voltage by 0.02V or something like that after every sucessfull pass.
The perfect end result would be a bios voltage table (or in software), which fits the cpu installed - we all know the silicon lottery here. In an approach like this, lies much potential for saving energy.
• #### andychow - Wednesday, November 26, 2014 - link

I've worked in the cubicles of many large corporations, and most people just log off their session or lock their screens at the end of the day, they don't turn the computer off. So cost savings would be even more interesting in these scenarios.
• #### piasabird - Wednesday, November 26, 2014 - link

So do unused ports and slots use power? Like if you use just 2 ddr3 SLOTS do the empty slots use power? Same with SATA and PCIE? So if the case is so why not use a MITX motherboard?
• #### piasabird - Wednesday, November 26, 2014 - link

You cant just look at the cost to run the motherboard. What about an eco friendly Monitor? Then there is the heat that is created to use the motherboard. During the summer or in say a server room something is cooling off the hot air.
• #### just4U - Wednesday, November 26, 2014 - link

Hi Ian,

I hadn't really noticed that Anand wasn't reviewing business class motherboards. I picked up a H97 GAMING 3 MSI board for my wife a few weeks back.. certainly doesn't look like a business board.. but it does come with all the software. Maybe you will get a chance to review that one in coming months..

Anyway, on this one I almost thought it was a Sniper board at first geez.. GREEN.. Waiting to see your matx x99 review.. should be interesting!
• #### Daniel Egger - Wednesday, November 26, 2014 - link

I LOLed when seeing the TÜV Logos. There're only few certifications like GS that follow a normed procedure. Other than that you can basically specify the test procedure and criteria, deliver the products (and a boatload of cash) and they will certify you that your products passed test procedures by the criteria you've specified. Very useful...

The important point of information I'm missing here is: What were the tests? What were the passing criteria? Is there any competition which underwent the same certification and if so what were the results?