We set our benchmarks up to get the fairest results possible, going to great lengths to make sure power savings features are enabled and Intel Turbo Boost ratios are at factory defaults. Some manufacturers enable Turbo Boost “Enhancements” when XMP is enabled, so we even test with XMP disabled while using a memory set that defaults to our desired DDR3-1600 baseline. While small differences can be attributed to luck of the draw or slight variation in base clock, large performance gains tend to indicate cheating via unintended overclocks. Large losses usually indicate configuration problems, and we like to expose both issues whenever they occur.
A small PCMark Work loss for Gigabyte’s Z97X-Gaming GT might have gone unnoticed, except that it’s also reflected in Sandra’s bandwidth-intensive Encoding/Decoding test and Sandra Memory Bandwidth. We’ll keep an eye out for how that affects any real-world tests.
Except for a small bump for the Z97 MPower Max in Far Cry 3’s high-quality test, we don’t see any suspicious gaming results. While that one test doesn’t have much impact on performance averages, we’ll keep an eye out for additional Far Cry 3 oddities in our next round-up.
The same MSI motherboard that encountered a gain in one of our games also face an unexpected performance loss in one of our applications, Adobe Premiere. We rechecked the data to confirm its accuracy.