System setup and notes
Here's a quick rundown of the test system should you wish to compare benchmark results with your own.- AMD XP2700+ S462 CPU (13 x 166MHz)
- Shuttle MN31N nForce2 micro-ATX motherboard in both single and dual channel DDR333 modes. Run with a dedicated Radeon 9800 Pro and on-board GeForce4 MX
- MSI K7N2-L nForce2 run in dual channel DDR333 mode
Other components
- ATi Radeon 9800 Pro 128MB 8x AGP
- ATi Radeon 9000 64MB 8x AGP
- 512MB ( 2 x 256MB) Mushkin PC3500 run at 2-6-2-2 (single and dual channel)
- 61.5GB IBM 120GXP Hard Drive.
- Liteon 16x DVD
- Samcheer 420w PSU
- Samsung 181T TFT monitor
- Akasa Silver Mountain heatsink
- Sony 16x DVD
- Lian-Li PC60 aluminium case
Software
- Windows XP Professional Build 2600.xpclient.010817-1148
- DirectX9
- NVIDIA nForce2 2.03 drivers
- NVIDIA 43.45 drivers for integrated GeForce4 MX
- ATI CATALYST 3.2 drivers and control panel (6307s) for 9800 Pro
- ATI CATALYST 3.1 drivers and control panel (6292s) for 9000
- Pifast v41 to 10m places
- Lame v3.91 MP3 encoding with Razor-Lame 1.15 front-end using U2's Pop album
- Virtual Dub 1.5.1 DVD encoding, DivX 5.03 Pro CODEC
- Hexus SETI benchmark
- 3DMark 2001SE
- UT2003 Demo (Build 2206)
- Comanche 4 benchmark
- Serious Sam 2 Demo
- Quake 3 v1.30
Notes
A number of different combinations available here. To compare the MN31N's performance to a standard ATX nForce2 motherboard, I'll run it in direct comparison to an MSI K7N2-L nForce2 motherboard, with both boards running dual channel DDR333 and a dedicated Radeon 9800 Pro graphics card. To test the performance of the on-board GeForce4 MX, it'll be run on the Shuttle in both single and dual DDR333 modes, with a framebuffer of 64MB. As integrated graphics cards' share system bandwidth, running single channel DDR333 should incur a performance hit in both gaming and non-gaming applications.
Dual channel DDR333 offers around 1.5GB/s greater bandwidth than the XP2700 can use at 166FSB, therefore the performance hit should be minimised when running the on-board graphics. The comparison MSI will also run a dedicated Radeon 9000 graphics card to show how on-board performance compares to a low-cost AGP card.
A quick word on the integrated graphics.
The integrated GeForce4 MX is linked to system memory via a dedicated 100MHz bus. It features two pixel pipelines, a 200MHz core frequency, and two texture units per pipe, giving a single-texture fill-rate of 400MPixels/s and a multi-texture fill-rate of 800 MTexels/s. Extremely similar to the discrete AGP GeForce4 MX offerings. The IGP even features up to 2x FSAA and 4XS anisotropic filtering, although those settings would cause any modern game to chug along at 1024x768x32 and above. Some may see its DirectX7-only compliance as something of a drawback, but that should be more than acceptable in the environment that the MN31N is destined for.
Overclocking
With no voltage adjustment and no mention of a locked AGP divider in BIOS, overclocking is not the MN31N's forte. That's not its purpose. Our test XP2700 seemed fine up to 175FSB. After that point, general system instability set in. That's more to do with the CPU's limit than any other factor.
Stability
Whilst overclocking prowess isn't important, stability most definitely is. Filling all three DIMM slots (2 x Mushkin PC3500 and 1 x OCZ PC3500 @ 166MHz, 2-7-3-3 timings) and running the motherboard through a series of Prime95 and 3DMark 2001SE looping tests caused no lockups, faults, or anomalies over a 12-hour period. That holds true with both a dedicated Radeon 9800 Pro and the on-board graphics. Installation of both the motherboard and Windows XP Professional went smoothly, as expected, and devices installed without issue.
