Power Consumption

We used Extech Power Analyzer 380803 for our power consumption measurements. This device is connected before the system PSU, i.e. it measures the power consumption of the entire system without the monitor, including the power losses that occur in the PSU itself.

This time we have slightly changed our approach to measurements, in order to eliminate some inconsistencies and be able to provide more precise readings. We have no issue with the readings that we used to take off the systems in idle mode. When we took the power readings in idle mode, the system was completely idle: we waited until all activity after boot-up is over and there were even no requests sent to the hard drive anymore. We also were totally satisfied with the graphics card load created by FurMark test. In this case the load was even higher than in many computer games. However, as for the system power consumption readings that were obtained after a single run of Fritz Chess Benchmark utility turned out lower than the actual power consumption of a working system. And in fact, it was much lower.

Here is an example. Power consumption of EVGA X58 SLI Classified mainboard during CPU overclocking measured following our initial methodology involving Fritz Chess Benchmark utility varies between 353W and 370W depending on the overclocking mode. However, during stability tests with Prime95 utility power consumption proved way higher and reached 415-418W. as you can see, the difference is significant, although Prime95 is far not the most resource-consuming application.

The problem is that it is very inconvenient to measure power consumption with Prime95. This program loads the system in recurring cycles when the load increases and then decreases, so it takes a lot of time to get trustworthy readings. Besides, it is pretty hard to ensure equal testing conditions for different mainboards. Power consumption depends not only on the load but also on the temperature. The higher the temperature, the higher is the power consumption. And I am not talking about room temperature here, because the difference between 22°C and 25°C is really insignificant. However, when the microchips temperature skyrockets from 22-25°C to 60-90°C, this is where the differences start to matter. Therefore, you should take the readings as soon as possible, to prevent the temperature differences from contributing to the power consumption measurements error.

Luckily, we did find a more suitable utility for our power consumption tests. New version of LinX program already knows to load Intel Core i7 processors with eight computational threads. This utility offers simple and convenient interface for Intel Linpack benchmark, which is very famous for loading the CPUs extremely heavily. As a result, we get pretty high system power consumption readings within a very short period of time, which are closer to what we see in real life.

Well, let’s check out the “correct” power consumption readings in different operational modes for two systems in question: one based on EVGA X58 SLI Classified and another – on Asus P6T:

As we have expected, EVGA X58 SLI Classified mainboard consumes much more power. The power consumption of Asus P6T mainboard is comparable with what other solutions have to offer and acceptable from the consumer standpoint only in the nominal mode. Because of specific issues that we have already discussed before Intel processor power-saving technologies do not work on this mainboard when the base frequency increases past 152MHz. therefore, I would expect Asus board to lose during overclocking, however, it turns out more energy-efficient in all modes. Even during CPU overclocking on EVGA X58 SLI Classified by raising the base frequency to 181MHz or 186MHz, when all Intel processor power-saving technologies remain up and running, it consumes way more power. The reason is obvious: the use of hot energy-hungry Nvidia NF200 controller.

As we know, EVGA X58 SLI Classified mainboard as well as the additional EVGA ECP panel have a special set of jumpers that allows to selectively disable those PCI Express x16 graphics cards slots that aren’t used at the moment. Just in case we checked it out as well, but even when we disabled all slots except the first one with the graphics card in it Nvidia NF200 controller didn’t disconnect and the power consumption didn’t go down. Frankly speaking, we are extremely curious to see how big of an advantage this controller will bring us one day. And how long will it take for the new graphics card generations to appear? Six months? A year? A year and a half? I believe that there will be new mainboard models by then equipped with an Nvidia NF200 or similar controllers. However, all this time, our mainboard will waste tens of watts of energy and turn them into heat. I am afraid this integration might have been a little too early and the hypothetical future benefits cannot justify extremely high power consumption of EVGA X58 SLI Classified today.