Articles: CPU

Bookmark and Share

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

Power Consumption

Our tests showed that I-series processors yield to “regular” processor modifications in computational performance. This is the side effect of the lower power consumption, resulting from a number of measures including lowering the clock frequencies, too. However, until this moment we used unofficial specifications to speculate mostly theoretically about the low power consumption thresholds. Now we should check out their practical energy-efficiency.

The graphs below show two power consumption readings. The first one is the full power draw of the computer (without the monitor) measured after the power supply. It is the total power consumption of all the system components. The second one is the individual power consumption of the CPU alone. In both cases the PSU's efficiency is not taken into account, because our measuring devices are installed after the PSU and read the currents along 12, 5 and 3.3 V lines. During this test session the processors are loaded by running the 64-bit LinX 0.6.4 utility. We also used FurMark 1.9.1 to load the graphics cores. We enabled all the power-saving technologies including Turbo Boost to ensure that we measure computer's power draw in idle mode correctly.

We start noticing the differences in power consumption between regular and energy-efficient processors in idle mode already. Even in idle mode 45 and 35 W CPUs can help you save about 1-2 W of power due to lower processor Vcore guaranteed by activated Enhanced Intel SpeedStep technology.

Things get very interesting when only one processor core is utilized. Here 35 and 45 W Core i5-2500T and Core i5-2390T processors do not show any advantages in terms of energy-efficiency. And the reason behind it is extremely aggressive implementation of Turbo Boost technology. If they are not fully utilized, they rapidly increase their clock frequency exhausting the entire thermal envelope and catching up with 95 W and 65 W models that don’t dare overclock that seriously. As far as Core i3-2100T and Pentium G620T processors are concerned, they do not support Turbo Boost that is why they consume fewer watts of power than Core i3-2100T and Pentium G620T with 65 W TDP.

The results are also pretty interesting when the processor’s computational cores are fully utilized. Overall, systems built around T-series processors consume noticeably less power than platforms with standard CPUs of the same class. But nevertheless, we can notice a few cool irregularities in the practical results. For example, Core i5-2500T processor with the maximum declared TDP of 45 W in the end needs more power than Core i3-2120 with a 20 W higher official TDP. Of course, it happens because these two processors have different number of cores inside, but the fact remains undeniable. This is also why Core i5-2390T consumes more power than Pentium G850.

All this indicates that T-series processors are not always more energy-efficient tan their “regular” brothers. They indeed offer better performance-per-watt, but even when the absolute power consumption readings are identical, they may still lose to significantly slower CPUs with higher declared TDP. So, please keep it in mind.

When we ran the tests with additional graphics load, nothing unexpected happened. Intel HD Graphics 2000 core needs considerably less power than the computational cores that is why the manufacturer didn’t really invest much effort in optimizing this part of their processors. As a result, we see only minor discrepancies in our measurements. The only one that stands out is Core i5-2500T, which graphics core may be overclocked to a higher frequency than in all other cases.

We see a similar situation when the processors are used as part of a media center. High-definition video decoding tasks hardly disturb the picture showing very similar power consumption in systems with 95, 65, 45 and 35 watt processors.

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


Comments currently: 8
Discussion started: 08/25/11 05:36:31 PM
Latest comment: 04/12/12 03:28:16 PM

View comments

Add your Comment