For all benchmarks the board was tested at 166/166 (The board requires PC2700 or better to run newer Athlon XPs). For comparison it comes up against the Epox 8K5A+, featuring the KT333 chipset and its predecessor - the KT333 based Soyo Dragon. The stiffest competition is likely to come from the Asus Nforce 2 based A7N8X. Lets see how they shape up!
Sisoft Sandra 2002 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 KT333 Dragon at 266MHz being beaten up quite badly by the new 333MHz bus competing solutions. One thing that has surprised me is the amount the nforce 2’s synthetic memory performance has been over-hyped. As you can see, even with 128bit DDR, there is only a spread of 30MB/s covering the KT400 and the nforce 2.
PCMark is a benchmark from MadOnion (now 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 KT400 in this test, with the spread being around 12%. The CPU tests are more or less the same across the board, with the exception of the Dragon KT333 which is slow due to its XP2000 processor. The Nforce shows a commanding lead in the memory test.
What seems to be everybody's favourite benchmark, 3DMark 2001 comes up next. After you have seen the tests as many times as I have however it starts to loose its appeal ;). 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 two KT400 boards are closely matched, with the nforce board up by around 700 marks. As the new 3DMark 2003 is much more heavily dependant on the graphics card than the subsystem, and does not produce results that correlate with any current (or future) game engine, OcPrices will not be using it in our motherboard test suite
As you can see the nForce2 wins this test again although at modern gaming resolutions the difference between the two platforms is reduced to insignificance.
Quake III has been dropped from the OcPrices graphics benchmarking suite, but even after all these years is a very handy memory and CPU subsystem test.
The KT400 performs admirably in this test, but you can see at the lowest resolution, the nForce2 is once again triumphant. Once you increase the options there is nothing to choose between any of them.
Serious Sam 2 is a very useful OpenGL benchmark, as you can run special scripts to make sure the detail setting on both cards is identical. The options are also endlessly modifiable, and it returns a detailed report detailing average, peak, low and average without peaks results. The average without peaks is the scores shown below.
All the platforms are performing almost identically in Serious Sam 2, although you can see the nForce2 still holds the performance crown.
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 KT400 is noticeably quicker than the nForce2 and the KT333 platforms by a reproducible amount - I checked this twice to verify, due to it being against the trend of results, and both times we saw the same result.
