For comparison I have chosen the popular Asus A7N8X, the Soyo KT400 Dragon Ultra, the Epox 8K5a and the Albatron KM18G. I don't expect to see much, if any, variation in performance. KT400 and KT333 performance is shown for reference.
Sisoft Sandra 2003 Pro is one of the most valuable benchmarking suites available, since it has the facilities to test almost every aspect of a system's performance. I have axed the CPU tests, as any variation in performance is almost invariably due to slight variations in the FSB.
As you can see there is little difference across the board, with the nFORCE 2’s synthetic memory performance only marginally quicker than the KT333s.
PCMark 2002 is a benchmark from Futuremark and gives an arbitrary score based on synthetic subsystem performance. Although it seems a little ambiguous, it does provide reproducible and accurate results - at least when comparing the same platform. The nFORCE 2 flexes its muscles over the older VIA boards in this test, with the spread being over 12%. The CPU tests are more or less the same across the board.
The classic 3DMark 2001SE comes up next (yawn). The tests still look the business and I STILL want to play the matrix game even after seeing the scene literally thousands of times. 3Dmark uses DirectX technology and tests all aspects of system performance including the CPU, memory bandwidth, AGP and the graphics card. As you can see the nFORCE board up by around 700 marks over the older VIA chipsets.
Quake III, even after all these years is a very handy memory and CPU subsystem test and even today, provided you play it with full features and settings, still looks better than many new titles.
At lower resolutions the nFORCE 2 is clearly faster than the VIA solutions, with the difference vanishing as it increases.
Serious Sam 2 is a very useful OpenGL benchmark, as you can run special scripts to make sure the detail setting on all platforms is the same. The options are also endlessly modifiable and it returns a report detailing average, peak, low and average without peaks results. The average without peaks is the scores shown below.
Clearly this test is mainly graphics-card dependant, with all boards performing similarly.
Sysmark 2002 tests all aspects of a machines performance and this edition uses all the latest versions of popular office applications. Although the test is not a good tool for comparing across Pentium / Athlon platforms, it is excellent comparing between chipsets within each group.
As you can see, the difference is minimal across the board, although the NF7 and the Albatron are noticeably quicker than the A7N8X by a reproducible amount. Overall then, there is nothing to report in terms of performance, the NF7 is a superb board in this respect, but then so are all of the nFORCE 2 boards. What sets them apart are features, layout and, best of all, overclocking >=)
