You switched from a 3.4GHz P4 EE to a 2GHz Athlon 64. That's not really going to improve the CPU limitation problem, and you would have been much better off with a single-core CPU for benchmarking purposes. In real life, I'd take the 3800 over any single-core CPU any day, but for video card benchmarking it's wasteful (and I can't see why on Earth AMD would provide you with a CPU for a video card benchmark). That second core is doing nothing. However, this is fairly irrelevent since neither processor should limit you much in any games, even with a 7800GTX. It seems to have in Far Cry, but I bet overclocking it would do 3FPS or less in Doom 3. Half-Life 2 is a different story...
Are you sure vsync wasn't on in HL2? those we suspiciously low framerates for that card. My 6800GT/[email protected]/768MB@400MHz DDR 2-3-3-7 system handles the game at 1280x960@85 with vsync on and all settings on the highest at the same framerates you were getting. You're results are also inconsistent with test results at other sites using inferior cards and sometimes inferior prcoessors, too. When you get 50FPS more in Doom 3, you know there's something wrong with the benchmark.
Speaking of Doom 3, wouldn't it make sense to max out AA and AF, on top of running it at Ultra quality, seeing as you're getting framerates well over 100?
Lastly, you really ought to stop using 3dMark entirely and start using Battlefield 2. The second-most played FPS out there (the original CS still holds #1) ought to be first on a list of games to be benchmarked, rather than a synthetic benchmark that has next to no relevance to any real-world gaming. There are certainly some other games that would be worth testing, too. RTS fans would enjoy Warhammer 40K Dawn of War and Rome Total War being tested, though one or both of them would likely be unstressed by SLId 7800GTXs. There are plenty of RPGs out there worth testing, as well. But really, BF2 is crucial and there's no reason you guys shouldn't be testing that.
Comment