ASUS Striker II Extreme: Mucho Bang, Mucho Bucks

Test System Configuration and Methodology

Each motherboard was flashed with the latest available BIOS - some of which were still labeled beta at the time of testing - and any unneeded functionality was disabled (unused serial ports, onboard HD audio, IDE controllers, etc.). This streamlined the OS and driver installs for optimum performance.

Prior to each round of motherboard testing we loaded a fresh install of Windows Vista Ultimate (x64), followed by platform drivers, then Service Pack 1 (SP1) including all patches and hot fixes currently offered by Microsoft's online Windows Update service. We made sure that our hard disk's write-back caching was enabled and the prefetch folder was cleared before each test run.

The latest WHQL-certified video drivers for our pair of MSI 8800GTS 512MB (G92) cards were then installed and all NVIDIA control panel settings were left as-is with the exception of SLI functionality, which was selectively enabled and disabled depending on which round of benchmarks were being run. Additionally, rather than deal with comparison issues arising from GPU core, shader, and memory clocks speed discrepancies seen from card to card, depending on the vendor, we chose to flash our pair of graphics cards to the NVIDIA reference clocks as listed on their site. Therefore, if some of our scores seem low to you it may be because the cards you have been working with come factory-programmed with higher clock frequencies.

Our test suite consists of Futuremark 3DMark 2006, a popular 3D benching program, Futuremark PCMark Vantage (x64), Crysis (32-bit executable), Unreal Tournament 3, Company of Heroes: Opposing Fronts, and Lavalys' EVEREST Cache and Memory Benchmark tool. All game benchmarks are run at a higher resolution (1920x1200) and a lower resolution (1280x1024), always on "High Quality" settings - with the exception of Crysis that was run at "Medium" settings - and in each case results are reported with and without SLI enabled, whenever possible. In addition, these tests used our Core 2 Duo E8500 dual-core CPU as well as our quad-core QX9770. Finally, we show scores at both stock processor frequencies and rated bus speeds and when overclocked to 4.00GHz (8x500MHz for the E8500 and 10x400MHz for the QX9970).

We ran each test five times, threw out the high and the low score, and then averaged the remaining three. These results fully qualify the advantages (or not) of purchasing a more powerful CPU - whether that means higher frequencies or more cores - along with any real-world benefits of NVIDIA's SLI Technology.

There is one exception to this: the ASUS P5E3 Premium. Although a worthy contender, the P5E3 Premium is not SLI-capable due to drivers, and as such we do not provide SLI results. Future testing using Crossfire-capable ATI/AMD HD 3800-series cards will be completed in the next roundup, but for now we're concentrating on SLI. We've really added the P5E3 Premium to the mix just to get a feel for how X48 stacks-up against NVIDIA's 750i, 780i and 790i Ultra when it comes to single-card graphics performance. This will be good information for those that are not necessarily in the market for an SLI-capable motherboard or feel gaming may not be their top priority.

Before we get into the specifics let's look at overall system performance. We chose PCMark Vantage for this task as it does an excellent job of documenting the performance of nearly every subsystem and then combines the results in a meaningful way. This allows us to conduct a broad sweeping check of the system as a whole; then, if the score is unusually low, we can decide whether investigation and further testing may be prudent.