Power consumption

We conducted an experiment to determine power consumption of this system in various modes.

The motherboard was powered by the Corsair CMPSU-620HX PSU. Four TRMS multimeters were daisy-chained between the PSU and the motherboard. Those were used to measure the direct current in each power lane: +3.3VDC, +5VDC, +12V1DC, +12V2DC. The readings were multiplied by the voltage values to get system power consumption in each lane. After that, the results were summed up and published on the diagram.

The maximum power consumption reached 35W. And the power applied to the input of the CPU voltage regulator (the rough estimate of the CPU power consumption) was only a little higher than 4W. The rest was consumed by the chipset, memory, hard drive, and CPU cooler. Theoretically, the entire system with one or two hard drives will do fine with a typical 65W PSU.

Then we compared the power consumption of the competing miniature platforms based on integrated chipsets: Intel 945GC and NVIDIA ION. Power consumption was measured in the following modes:

• Idle -- Windows XP SP2 with the default list of background processes.
• Full CPU and GPU load -- The maximum system load generated by the Prime95 and ATITool utilities.
• Full CPU load -- The maximum CPU load without additional load on the graphics subsystem generated by Prime95 (mt=4).
• Full GPU load -- The maximum load on the graphics subsystem generated by ATITool.

The NVIDIA ION platform consumes less power than the platform based on Intel 945GC. We should also mention that ION's graphics core is more energy-efficient, as it offers higher performance and lower power consumption than its competitors.

Noise level

Test conditions

VShV-003-M3 sound level meter and A-weighting are used. The sound pressure sensor is located 90cm above the center of the test subject, which itself stands on the soundproof sheet of porous rubber. All tests are conducted in a soundproof room with the typical noise level of 20 dB(A). All electric appliances in the room are switched off during tests. Note that we do not assume that manufacturers conduct their own tests under similar conditions.

Subjective loudness scale

We assume that:

• 35 dB(A) -- Low noise that can be drowned by sounds of a movie and background noises.
• 40 dB(A) -- Moderate noise that can be drowned by a loud movie OST, etc.
• >45 dB(A) -- Loud noise that can drown a movie OST. To be near the test object is uncomfortable.

Thus, the loudness can be separated into the following ranges:

• <35 dB(A) -- Very quiet
• 35-40 dB(A) -- Quiet
• 40-45 dB(A) -- Moderately loud
• >45 dB(A) -- Very loud

This loudness scale is rather subjective and rough. It implies certain exceptions, like when the test object is located far enough from listeners, so its noise is less discomforting. It's solely based on our method of measuring noise.

Test results

Unfortunately, this motherboard has a noise problem. It doesn't allow controlling fan speed (neither by means of BIOS nor by means of utilities), so its noise level exceeds 50 dB(A) at full speed. This is very loud, and it becomes unbearable even after a couple of minutes. Perhaps, our motherboard sample had a defective fan, but we cannot say for sure. Keep this issue in mind, if you want to buy this motherboard.

Performance

Testbed configuration:

• CPU: Intel Atom 330, 2 physical cores + 2 virtual cores, 1.60 GHz, 533 MHz FSB
• Memory: 2 x 1GB Corsair CM2X1024-9136C5D, tested as DDR2-800 at 4-4-4-12-2T
• Graphics card: integrated GeForce 9300, 256MB of shared memory
• HDD: Seagate Barracuda 7200.10, SATA, 7200 rpm
• OS: Windows XP SP2

Our comparison table includes the results of the ASUS AT3N7A-I, as well as Zotac ION-ITX (with the same chipset) and Intel D945GCLF2D (i945GC), which helps us compare performance of different solutions with the same Intel Atom 330 processor.

ION-based motherboards turn out to be a tad slower than Intel solutions in archiving and encoding tests, but it will be difficult to notice the difference in real life -- resource-intensive applications are very slow on this platform anyway. In return, ION is much better in games, leaving the i945GC way behind. If not for the weak processor, this motherboard would have done fine in modern games with at the lowest settings.

As for the hardware-assisted video decoding, it's nearly perfect. The motherboard does fine with all of the test videos:

• XviD, 640x480, 2000 Kbps
• x264, 640x480, High profile Level 4.1, 2000 Kbps
• x264, 1280x720, High profile Level 4.1, 5000 Kbps
• x264, 1280x720, High profile Level 4.1, 50 fps, 5000 Kbps
• x264, 1920x1080, High profile Level 4.1, 10000 Kbps
• x264, 1920x1080, High profile Level 5.1, 15000 Kbps
• MPEG-2, 1920x1080, 30000 Kbps

The most difficult step here is to enable DXVA. It's easier to do so in Windows Vista or Windows 7, while Windows XP has many more problems with it. In our case it was Windows Vista SP2 (32-bit) and The KMPlayer with four profiles of hardware-assisted decoding -- Nero, ArcSoft, MPC-HC, and CoreAVC. We only used EVR as a renderer. If these conditions are met, all of the H.264 and MPEG-2 videos are played without frames being dropped. In fact, we had no doubts about it -- as we reviewed Zotac motherboards with the same chipset, we tested the hardware decoder with H.264 (High@L4.1 profile) and VC-1 (AP@L3 profile) videos with the average bitrate of 30 Mbps, peak bitrate - 35 Mbps. ION coped with them without any problems.

It's interesting to look at the results of hardware acceleration of video encoded with non-standard settings: H.264, 1280x720, High profile Level 4.1, 50 fps, 5000 Kbps and H.264, 1920x1080, High profile Level 5.1, 15000 Kbps. The fact is, most modern "hardware" HD players (such as Dune, TViX, and many other popular solutions) are designed to play only "correct" files, while these video clips are not quite standard. Increased fps should not be a problem for advanced devices, but the High@L5.1 profile and the increased number of reference frames can easily stupefy a player, as it does not comply with DXVA standards and does not allow to use hardware acceleration. However, NVIDIA ION, as well as other systems with modern graphics from NVIDIA or ATI, does not suffer from these problems: the developers of video drivers and codecs consider what users really need, so the non-standard videos can be played without any problems.

We should add a few words about the Elephant's Dream (1920x1080, VC-1) WMV file downloaded from Microsoft's website. The fact is, it's the only video that the ASUS AT3N7A-I failed to play without frame drops. WMP 11 loaded the processor by 15-20%, which is an apparent sign that hardware decoding was enabled successfully. However, dropped frames were noticeable, especially in dynamic scenes. On the other hand, motherboards with NVIDIA ION easily cope with much "heavier" content in the VC-1 format (but in a different container -- M2TS). Thus, this blunder is a platform problem, but it's theoretical rather than practical, since there is no paid VC-1 content in the WMV container anyway, and it's really difficult to create such video at home.

Conclusions

ASUS rolled out a good motherboard with a high-quality audio codec, lots of USB and SATA ports, built-in Bluetooth, and a PCI expansion slot. The cooling system has to be improved though, and the LAN controller offers only a one third of its official parameters. Finally, the price of this product is relatively high. There are many solutions based on NVIDIA ION and Intel Atom, so you should definitely take a look at alternatives.

The motherboard was provided by NVIDIA.

Write a comment below. No registration needed!