rpertusio

Something of interest to note... I had an older A7V333 motherboard (via chipset) and replaced my current (newer A7N8X Dlx, nForce2 chipset) with it. My scores jumped up to WinXP scores! The problem appears to be: (1) The motherboard itself [A7n8x] or (2) The chipset drivers I vaguely recall benching 3dmark'03 with the older WHQL nForce drivers (2.41 or similar) and got horrible CPU scores. As you can see in my first posting, my CPU scores now nearly match my XP scores, possibly related to the installation of the v3.13 chipset drivers from nVidia. food for thought! - rp

I have noticed the same thing regarding performance in 2003 Serv (eval). I do not see a performance hit with the CPU, but rather with the graphics card. allanon, you have the same motherboard, and a similar graphics card. There may be significance to that. Just now I benchmarked with Aquamark3. See the following scores: Win2003 Graphics: 2,466 points Win2003 CPU: 6,036 points WinXP Pro Graphics: 4,692 points WinXP Pro CPU: 6,123 points I am running an overclocked AthlonXP 1700+ at 2.0Ghz, with a Radeon 9500 Pro overclocked to 355/303, on an Asus A7N8X Dlx (nForce2) motherboard. Yes, I have enabled/installed DirectX 9.0b Yes, I have enabled DirectDraw/Direct3d/AGP Texture Acceleration Yes, I have installed the latest Chipset drivers (nForce v3.13) Yes, I have installed the latest vid drivers (Catalyst 3.9, Omega) Yes, I have turned video acceleration to maximum Yes, I have changed Processor scheduling/Memory usage to "Programs" Even with the specs I have, I can play all games fine without noticable difference from XP Pro. However, down the road, unless I find a 'fix' to improve performance, I will revert to WinXP again. I realize Win2003 Serv wasn't intended for 'gaming' but the desire to find the right combination of settings or tweaks lives on. In reality, WinXP and Win2003 Serv aren't fundamentally different. They were built on the same code base, and with enough tweaking should be able to mimic the other. We'll see. If anyone finds a solution, add to this thread. - rp (Edit: updated benchmarks, as my original post was skewed)