Gaming Performance
3DMark06 is used in graphics cards and CPU reviews because it’s simulated gaming engine looks at how a processor affects gaming performance. Because only a small part of the overall 3DMark06 score is driven the CPU results, the scores are pretty close; only 27% separates the top score from the bottom. In our individual CPU tests though, it is obvious that the quad-core parts have a distinct advantage when a game engine can utilize the hardware.
Valve helped out reviewers with the release of their Source Engine tests a while back that CPU performance in both a game simulation of particle physics (that could be used in future game engines) and a developer simulation of the process of building a map in the Source engine. The particle test produces a score based on internal metrics and we can see that the quad-core parts are clearly dominant. The QX9650 has a 138% performance advantage over the top dual-core part and is able to beat out the QX6850 by 7.1%. The map compilation test shows a 98% scaling from dual-cores to quad-cores but the differences between Yorkfield and the previous generation are minimal.
World in Conflict is a new, DX10 real-time strategy game that is heavily CPU-limited when using a high graphics card like the NV 8800 GTX we tested with. Though I ran the tests at 1024×768, this resolution isn’t as big of a move down the field as we see in some other CPU reviews as the lowest score on our graph is only getting 45 average FPS with a minimum of 17 FPS; that is close to be non-playable. Both AMD processors are left behind the Intel Core 2 line up, especially when looking at the minimum frame rates the two systems allowed. After that we see a near-linear progression up the line where the new QX9650 produces the best results. A 9.3% gain in frame rate over the QX6850 is actually a great source of excitement for anyone that has read our initial Crysis performance article.
