Hey guys, I just replaced my x1950 Pro with the 3850 AGP. Here’s what I have to report. I’m running all stock speeds and all stock cooling. The performance of the 3850 took me up 25-35% in 3d games over the x1950 Pro. In 3dMark06 I got a 8226. I was expecting a little bit more. I’m guessing that I am now CPU bottlenecked, which means there’s no more upgrades for this machine. Performance in ATiTool (the fuzzy cube benchmark) went from 570 fps with the x1950 Pro, to 825 fps with the 3850. That’s quite an increase, and kind of what I was expecting overall, but I don’t think ATiTool stresses the CPU like 3dMark06 and only gives GPU performance. This shows the true difference in performance of the two GPU’s without the CPU dependency of the other benchmarks. I was worried about power when I first installed the 3850 AGP because I’m using the stock 250w PSU, but the reduction in process size to 55nm did a lot to reduce the power consumption. The 3850 was nearly identical to the x1950 Pro at idle (122 W total system draw), and at full load the 3850 was actually a few watts less than the x1950 (at 210 W total system draw for the 3850 and 221 W for the x1950 Pro). Measurements were taken with a Kill-A-Watt. Drivers were an issue. I could not get any of the standard ATi drivers to install and had to use the Sapphire 8.3 Hotfix drivers. Not sure if it was worth the $240 but oh well. Next upgrade will be an all new system and run Vista SP1.