Articles: Mainboards
 

Bookmark and Share

(3) 
Pages: [ 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 ]

Power Consumption

We perform our power consumption measurements with an Extech Power Analyzer 380803. This device is connected before the PSU and measures the power draw of the entire system (without the monitor), including the power loss that occurs in the PSU itself. In the idle mode we start the computer up and wait until it stops accessing the system drive. The mainboards are sorted in alphabetic order. The results of the Gigabyte G1.Sniper 5 are colored differently for the sake of readability.

We feel ashamed now that we criticized the ASUS Z87-Deluxe for its higher power consumption in comparison with similar mainboards. The PLX PEX 8747 switch built into the Gigabyte G1.Sniper 5 provides performance benefits for multi-GPU configurations but does that at the expense of high power draw.

For all their downsides, Haswell-based CPUs should be given credit for requiring less power in idle mode in comparison with their LGA1155 counterparts. Unfortunately, we can’t see that when the mainboards work at their default settings, so we have an additional test mode called Eco. It means the same default settings but we manually switch all options referring to Intel’s power-saving technologies from Auto to Enabled in the mainboards’ BIOSes. This helps improve the results, lowering the power consumption of our configurations considerably.

We want to remind you that we install an AMD Radeon HD 7970 graphics card into our test configurations. If we instead used the CPU-integrated graphics core, the overall power draw would be lower than 30 watts (perhaps not with the G1.Sniper 5, though). Haswell-based CPUs are indeed very economical when idle, so it is a shame that the mainboards do not ensure this advantage by default. You have to correct some BIOS options for that.

For power consumption tests under high load we run the LinX utility, which is a graphics shell for Intel’s Linpack benchmark and supports AVX instructions. It is heavier on the CPU than ordinary applications, yet it is just an application nonetheless. It is quite possible that there are some other programs that can be just as heavy. That’s why we stick to using LinX for the purpose of checking the computer out for stability and when measuring its power consumption.

The numbers are high, but they are close to the highest power consumption possible at all. To measure the power draw of our configurations in typical applications, we used the Fritz benchmark. It doesn’t really matter which exactly application you use for that purpose. Any ordinary program that can run on all four CPU cores will produce the same or comparable results. So it turns out that we shouldn’t worry about the high power draw under the AVX-using LinX. The typical power consumption is about 100 watts whereas the most economical mainboards need even less.

By the way, you have to count in the graphics card’s power draw to calculate the overall system consumption. We use CPU loads in our power consumption tests, but if we load the AMD Radeon HD 7970, the total power draw will be close to 250 watts at default settings and even higher at overclocking.

Now let’s see how much power our configurations need when overclocked.

We don’t have a special Eco mode for the overclocked configurations because we always use as many power-saving technologies as possible while overclocking. Instead, we show you a summary diagram with the power consumption of each mainboard at different settings. The results are paradoxical. The overclocked systems (working at increased CPU voltage) need as much power as in the Eco mode and less than at the default settings. It just proves once again how important it is to enable all power-saving technologies if you want your computer to be energy efficient.

The overclocked systems need much more power at high loads compared to themselves at the default settings. The high frequencies and increased voltages show up then.

The Gigabyte G1.Sniper 5 needs much more power than the others in idle mode, but the gap narrows at high loads and when the mainboards are overclocked. We can also remind you that LGA1155 mainboards with PLX PEX 8747 bridge have a typical power draw of over 80 watts when idle. LGA2011 mainboards with comparable PCIe sharing capabilities need even more power. Although we tested LGA2011 products quite a long time ago and our testbed configuration has changed since then, we are absolutely sure that the Gigabyte G1.Sniper 5, although more voracious than regular LGA1150 mainboards, is more economical than older products with comparable functionality.

 
Pages: [ 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 ]

Discussion

Comments currently: 3
Discussion started: 12/06/13 10:28:51 AM
Latest comment: 02/09/14 12:21:02 PM

View comments

Add your Comment