Motherboard testing comes down to one thing: making sure no one cheats. It's easy for manufacturers to slip hidden enhancements and frequency boosts into the BIOS to give their products an edge. Thankfully, that's rare these days.
Large performance gains are usually due to behind-the-scenes cheats, while significant lagging comes from configuration conflicts. We like to explore both possibilities using our testing suite. Meanwhile, small differences are usually attributable to slight variances in clock rate. We enable all power-saving features, including SpeedStep, and set the RAM to its default rate of 9-9-9-24 at 1333 MT/s.
Synthetic benchmarks are perfect for motherboards because they're written to take advantage of an entire platform. Even if your favorite games don't use every optimized instruction set, synthetics typically do, so they reveal whether any subsystem is artificially boosted or handicapped.
We start off with PCMark 8. The scores fall in as expected, with less than one-percent deviation from the average. This is what we want to see.
Next up is SiSoftware's Sandra suite. MSI dips ever so slightly in GIPS, but it's well within the margin of error. The story continues through the Multimedia, Cryptography and Memory Bandwidth modules, with all the boards within one percentage point of each other. The B85M-DGS takes a small lead in the sequential read portion of the file system bandwidth test, but also lags slightly behind in sequential writes. Neither result was repeatable, and more runs brought the average more in line with expected results.
The Cinebench scores are nearly flat. Differences of one point can come from rounding errors or the Asus/MSI board's slight BCLK advantage.
3D And Gaming
Since these are entry-level motherboards equipped with low-end graphics cards, our 3D testing is also constrained to synthetic metrics. The purpose isn't to demonstrate playable performance in any one title, but rather to identify suspect results. As such, synthetic 3D tests serve our goals well.
3DMark's Cloud Gate is designed for lower-end gaming systems, making it a great fit. The more demanding Skydiver scene is meant for systems a little above this level. Still, our GeForce GT 730 delivers solid performance. The B85M-DGS establishes an ever-so-slight lead over the other three boards, but it's hardly notable.
Next up are two Unigine tests. Heaven was the first DirectX 11 benchmark, and it's still useful now. Valley is newer, and it's more graphically demanding due to its wide vistas, forested hills and weather effects. Less than a single frame per second separates the slowest from the fastest performers across both tests.
Heat, Power And Efficiency
Every power-saving feature in each board's BIOS is turned on, fan speed is set to the Standard setting and Windows 8.1 Balanced power profile is enabled.
Both ASRock boards burn more power at idle, but the H81M-HDS' result is noticeably higher. Asus' H81M-E sips power by comparison, and the MSI H81M-E34 is our champion when you're not using your computer.
Saving power doesn't help the H81M-E keep its temperature down though; it's the hottest board under load. As I mentioned in the overclocking section, I'm not sure how the VRMs get so toasty when they're underneath the fan. ASRock's B85M-DGS ends up as our winner in the thermal measurements.
It seems somewhat silly to compare average performance and average efficiency across these boards, since they deliver almost identical benchmark numbers. But we want to stay consistent across our reviews. Dividing a board's score by the average of all the scores for a particular test gives a percentage of how much higher or lower that board scored compared to the others. Averaging them all yields an overall performance score above or below the average. As you can see, less than half a percentage point separates the contenders, and that's what I like to see.
Comparing average performance to average power turns into an efficiency rating as a percentage in relation to the average efficiency of all four boards we tested. Because they do roughly the same amount of work, this basically comes down to the motherboard with the lowest average power draw. ASRock's H81M-HDS loses fairly convincingly due to its higher consumption compared to the other three. Asus' H81M-E, on the other hand, wins with MSI's H81M-E34 finishing a close second. ASRock's other board, the B85M-DGS, is exactly in the middle, and that's not necessarily a bad place to be.
Really, these are all low-power boards with only a few watts separating them. The G3258 is a 53W chip. The average full system draw for all boards during torture tests was only 81.6W, which is less power than a Radeon R7 260X alone under normal gaming loads. Adding any kind of mid-range gaming graphics card would make the delta between our winner and loser almost insignificant.