The post above displays timedemo results at 640x480/low quality.
This was the only setting that would run on all the cards at roughly 10 FPS or higher. I didn't have the patience to run older cards at settings that I knew would generate under 10 FPS. Other Doom3 settings for the same cards are detailed in the spreadsheet.
All testing was using 71.89 drivers; I'm not certain I said that in the spreadsheet. FWIW my NV5 engineering sample (Riva TNT?) was not recognized by 71.89 and therefore was not testable. (So much for the universal driver architecture NVIDIA likes to claim! Did that just disappear without telling us?
All cards except the NV35GL were clocked at the speeds of consumer boards.
- NV30 (FX 5800 Ultra) was 500 GPU/1000 RAMs (500 DDR).
- NV28GL (Quadro 980 XGL or GF4 Ti4600) was 300/600 (300 DDR).
I don't recall the rest off the top of my head.
My 5800U gets around 48.7 FPS on 1024×768 at Medium, so your 5800U is running like mine. 10×7@M is how I play Doom3 when I'm not testing.
Speaking of playing Doom3 with a 5800U. The stock dustbuster fan was too loud to play after the kids went to bed, so I took off the plastic FlowFX from one of my 5800U cards. I replaced the thermal pads with paste, and put a fan on each side: a 80mm LED fan on the GPU copper, and a 60mm CPU fan on the backside copper. The 5800U temperature now stays under 60°C silently, and a soft blue LED glow reminds me when the card goes into 3D modes.