Some of the new features have fairly obvious targets. There can be very little doubt who is being targeted by the improved driver detection, for instance.
One of the more curious additions is a new option that "forces PixelShader 1.1 code path in game tests 2 and 3." Could it be there be big differences in performance for different versions of PixelShader on some chips? If some chips suddenly get a performance jump it would certainly leave a question mark over this feature.
The good news for anyone using 3DMark03 as a reference is that the scores from the new version should be exactly the same. The new version is available as both as a patch and a full install.
For all the brouhaha that surrounded the launch of 3DMark03, the benchmark seems to be gaining ground. Although it's certain that some benchmarkers will never use it, many consider it the only way to find out about possible DirectX 9 performance opf graphics chips. ยต