Overclocking
Since I haven’t overclocked an MSI motherboard before, I did a little digging and found that I can use MSI’s Control Center software to help set the necessary parameters. I was curious about the software, so I downloaded and installed it on Mini Build. However, that little experiment lasted about ten minutes, since it crashed whenever I started making changes to the core voltage. So, off into the BIOS I went.
This time I went hunting for the magic numbers from high to low. I started at 4.4GHz, but Mini Build wasn’t as giving as Big Build. After some stress testing with Prime95, I ended up with a Core i5-4690K running at 4.2GHz and 1.24V.
For the graphics card overclocking, I had to register on EVGA’s website to download and install PrecisionX. Using the program, I went a little extreme with the settings at first, thinking that 3D Mark’s Fire Strike benchmark would be a good indicator for overclocking stability.
The highest overall score I got in Fire Strike was 12013, and that was thanks to some tinkering and a little advice from our other techs. At this level, the fans on Mini Build were running all-out and I was ready to start the first round of testing, including the multimedia, productivity and synthetic apps. Games were last.
Loud as it was, Mini Build finished most of the testing without a hitch. However, things turned the other way as I started to go through the game benchmarks. I ran into some major problems, especially after exiting Far Cry 3, Battlefield 4 and Arma 3. I saw driver crashes and restarts, along with an occasional blue screen and reboot. The reality of the situation reared its ugly head when I was trying to test Grid 2 and couldn’t pass the first 30 seconds of our benchmark without one of the three monitors going out on me, a precursor to a full system reboot.
At this point in the game, I chose to address the graphics overclocking and rerun all the game testing with more conservative settings. For example, where I once tried to pull off a graphics memory overclock of 600MHz, I dropped down to just 53MHz just to finish my testing.
So, what’s the lesson here? Choose how you interpret your benchmarking validation tools wisely.