When we tested Mini-ITX platforms for Socket FM1 processors we made two assumptions. First we assumed that the users shopping around for compact mainboards prefer to use not 100 W processors, but more energy-efficient models with lower power consumption. Therefore, for our test session we chose a quad-core Llano processor with 65 W TDP. It was AMD A8-3800 with the nominal clock frequency of 2.4 GHz. However, it was capable of overclocking to 2.7 GHz in Turbo mode.
Secondly, we didn’t install an external graphics card into our testbed and decided to rely on the Radeon HD 6550D graphics core integrated into AMD A8-3800. Of course, all mainboards participating in our today’s test session can work with an external graphics accelerator, but we believe that integrated graphics is a more reasonable and popular solution for a Mini-ITX system.
Taking into account these two assumptions we put together the following testbed:
- Processor: AMD A8-3800 (Llano, 4 cores, 2.4/2.7 GHz, 4 MB L2).
- CPU cooler: AMD’s default boxed cooler.
- ASRock A75M-ITX (BIOS version P1.40);
- ASUS F1A75-I Deluxe (BIOS version 0602);
- Gigabyte A75N-USB3 (BIOS version F2);
- Zotac A75-ITX WiFi (BIOS version A199PA10).
- Memory: 2 x 4 GB, DDR3-1866 SDRAM, 9-11-9-27 (Kingston KHX1866C9D3K2/8GX).
- Hard drive: Crucial m4 256 GB (CT256M4SSD2).
- System case: Thermaltake Element Q (VL52021N2U).
- Operating system: Microsoft Windows 7 SP1 Ultimate x64.
- AMD Catalyst 12.1 Driver;
- AMD Chipset Driver 12.1.
We used Futuremark PCMark 7 benchmark to estimate the average platform performance. It measures the execution speed for typical applications widely spread in everyday usage models.
To test the performance during data archiving we used the benchmark built into the WinRAR 4.0 utility.
Final rendering speed was measured in Cinebench 11.5.
The diagram below shows the results obtained in x264 HD Benchmark 4.0, where a short video is encoding in two passes and the entire process is then repeated four times. We are offering you average results of the first as well as second iteration.
We measured the performance in Adobe Photoshop using our own benchmark made from Retouch Artists Photoshop Speed Test that has been creatively modified. It includes typical editing of four 10-megapixel images from a digital photo camera.
Gaming 3D tests portion starts with 3DMark Vantage with the “Performance” profile.
Since the graphics core integrated into AMD’s APU supports DirectX 11, we also ran a few tests in 3DMark 11 suite, also with “Performance” profile selected.
To investigate the platforms performance in real games we selected a few titles, including Far Cry 2, Crysis 2 and Aliens vs. Predator 2. These 3D shooters boast contemporary graphics and load the graphics sub-system pretty seriously. Of course, we can’t claim that the graphics core integrated into AMD A8-3800 is capable of delivering good gaming performance here, but these games will serve the comparison purposes just fine. We measured the fps rate in 1280x800 resolution and set the image quality to High or Medium.
We suddenly discovered that there is quite noticeable difference in performance between tested Socket FM1 systems. The gap between the fastest and the slowest of the four tested systems may sometimes reach 6-7%. It means that the mainboard makers have applied different optimizations, primarily to the memory sub-system. Therefore, the performance difference between the platforms is greater in those applications that are sensitive to the memory sub-system speed. This is also true for the tests that involve the graphics sub-system, because Socket FM1 based systems use part of the system memory for the needs of the graphics sub-system.
As for the specific mainboards we tested today, I have to stress that the best results belongs to ASUS F1A75-I Deluxe. ASRock A75M-ITX was a little bit slower, and Gigabyte A75N-USB3 and Zotac A75-ITX WiFi were the lowest of the four.