Many either reviewers don’t pay any attention to processor power consumption, or have a chapter dedicated to it somewhere closer to the very end of their review. However, for me the power consumption of the Socket FM2 platform was nearly the most shocking experience throughout the entire review. Therefore, I decided to really invest some time and effort into determining which of the system components made the most significant contribution into its very high power numbers. To accomplish this I measured the system power consumption in the nominal mode when idle and under different types of operational loads: only CPU, only the graphics core, and combine load. Then we took the same measurements in overclocked mode when we overclocked only computing cores, only memory, only graphics core and during complex system overclocking. In this chapter we are going to offer you the obtained results with the necessary commentary. And further in the review we will apply the same principles to power consumption of the entire system as a whole and its individual components in particular, so that we could analyze detailed results and draw proper conclusions.
We performed our power consumption measurements using 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 system up and wait until it stops accessing the hard disk. Then we use LinX to load the CPU, Cinebench 11.5 - to load the graphics core, and LinX and MSI Kombuster working simultaneously - to create complex heavy load. We decided on these particular programs because they demand almost the same levels of power consumption as the system would have during regular everyday use in real applications. The results on the diagrams are sorted out in ascending order. Let’s start with the chart showing the power consumption in idle mode.
Of course, we immediately noticed that system power consumption in idle mode increases during overclocking, but we believed that it was caused by the increased processor Vcore. It came as a surprise that it was the integrated processor graphics core that was almost entirely responsible for the increase in power consumption in idle mode. The difference between 28 and 31 W doesn’t strike as dramatic, but in reality it is quite substantial, over 10%. And keeping in mind that any computer is not fully loaded most of the time and the processor is in one of its power-saving states, then the increase in power consumption like that may be quite tangible.
We have already mentioned that we couldn’t measure the actual power consumption of the system under heavy load, because “Turbo Core” technology lowered the processor frequency below the nominal level. Therefore, we had to find a compromise and disable the “Core Performance Boost” parameter in the mainboard BIOS. This is a pretty serious allowance, which changes the system behavior dramatically. In this case the processor clock frequency doesn’t increase beyond 3.8 GHz, its Vcore doesn’t increase either, and therefore the power consumption appears lower than actual. Disabled “Core Performance Boost” doesn’t have any effect on the power consumption of the system in idle mode and it remains at 28 W. However, as the processor utilization increases, the difference becomes more and more prominent. For example, as you can see in the diagram above showing single-threaded mode, the power consumption of the system in nominal mode is 65 W, while in reality with “Turbo Core” technology in place it is about 10 W higher. We can’t say how much power the system needs when it is fully loaded and the CPU is 100% utilized, because in this case the processor frequency drops and this is exactly the reason why we had to resort to special allowances. However, it would be fair to assume that in this case the power consumption might be 10 W or even 20 W higher.
Despite this fact, we believe that the obtained results are quite relevant and may be used for comparisons. The thing is that LinX utility loads the processor extremely heavily, much more than a normal real application would even under most stressful conditions. As a result, even with disabled “Core Performance Boost”, the measured system power consumption in burn mode created by LinX turns out almost the same as we saw in real usage scenarios when we ran regular applications and had “Turbo Core” technology enabled. We used the same principles when we selected utilities for loading just the graphics part of the CPU and for creating complex burn mode for the entire system. We selected programs for loading given system components based on their ability to emulate the same levels of power consumption as the regular real applications would in every-day user experience.
We are going to draw final conclusions a little later, when we discuss the performance numbers, however, the already obtained results call for some intermediate verdict. Overclocked processor graphics core has a great effect in idle mode, but its contribution to the overall system power consumption remained at the same 10% or even drops to 4-8%, depending on the type of operational load. The increase in the memory frequency hardly has any effect on the power draw, the maximum we ever saw was 6%, but in most cases it is between 1 and 3.5% at the most. As we have expected, the highest power consumption increase occurs when the processor part of the A10-5800K is overclocked: it reaches as high as 20-40%. And complex system overclocking boosts its power consumption by astonishing 50-75%. Now let’s see how much of a performance boost we get during overclocking and where all the energy goes to.