How We Tested
I assembled all three builds in an open case. All of the components except the motherboards were the same. Windows, its drivers and the benchmarks were reloaded for each board.
Test System Components
|CPU||Intel Pentium G3258|
|CPU Cooler||Boxed Cooler|
|RAM||Mushkin Enhanced Stealth 996995S 8GB DDR3-1600 Dual-Channel Kit|
|Graphics||Gigabyte GV-N730D5-2GI GT 730 902MHz Core 5000 MT/s 2GB GDDR5|
|Hard Drive||Mushkin Enhanced Chronos G2 MKNSSDCR120GB-G2, 120GB SSD|
|Sound||Integrated HD Audio|
|Network||Integrated Gigabit Networking|
|Power||be quiet! Straight Power 10 500W CM|
|OS||Microsoft Windows 8.1 with Update x64|
|ASRock H81M-ITX||ASRock H97M-ITX/ac||Gigabyte B85N Phoenix WiFi|
|PCB Revision||Not visible||1.02||2.0|
|100.0 MHz BCLK||99.94 (-0.06%)||99.94 (-0.06%)||99.76 (-0.24%)|
A colleague, also testing motherboards, provided some of his results. They obviously took a lot of time and effort to generate. But looking at them did not discourage me from skipping benchmark sections in future motherboard articles. They were too similar to base a buying decision on, and that's what I hope to do here. Still, some basic benchmarking is required to check for anomalies.
|PCMark 8||Version: 2.3.293 Work, Home, and Creative Benchmarks|
|SiSoftware Sandra||Version: 2015.01.21.15 Memory Bandwidth|
|Crystal DiskMark 3.03||Sequential Read|
|Unigine Heaven 4.0||Version 4.0, Built-in Benchmark Basic: DirectX 9, Low Detail, 1280x720, 2xAA, No Tessellation Custom: DirectX11, High Detail, 1280x720, 0xAA, No Tessellation|
All tests were run with the CPU at stock settings. However, I used Intel XMP settings for the RAM, getting as close to the labeled settings as the boards (or CPU) allow. I used a Kill-A-Watt meter rather than my UPS' read-out for power. With the system off, the UPS on its own draws about 6W.
I selected the Heaven 4.0 Custom Settings to try to improve appearance without killing performance on the test system's basic hardware. Removing AA seems to have helped. I ran only the sequential benchmark in CrystalDiskMark because I was interested in testing the chipset SATA and USB 3.0 throughput, not the attached drives. Similarly, I looked for bandwidth differences in the RAM.