Having finally obtained a graphics card that can run the gamut of available shader types, and having the ability in 3DMark 2k5 Pro to choose which path to use for analysis, I decided to play with a bit of everything. Thus, I did the test on Proxycon with all the different paths in 1024x768 (no AA), then I redid the test for all paths with AA enabled, then I redid it in 1600x1200.
Basically, to avoid a lot of repeating, I took the results and drew the most significant results into seperate charts.
Here is the chart I made with the results for paths 2.0, 2_a and 3.0 (click the image to enlarge):
In this chart, I compare the basic shader paths to the 2_a path with anti-aliasing in order to see what impact there is (click image to enlarge) :