3DMark2001SE is in my opinion still a good DX8 benchmark and overall tool to gauge system performance. At default 3GHz speeds there scores between the RX600 and 5900 are relatively close. However, as CPU speeds increase, the PCX5900 picks up steam, while the RX600 seems to be barely gaining anything. The limitation in fill rate and memory bandwidth is really not improved by added CPU power. The Radeon X700Pro just opens up a can of whoopa** on both of these cards and takes no prisoners with a stranglehold of over 4,000+ points over it's next competitor in the PCX5900 that doesn't decrease as CPU/System speeds increase.
3DMark03 was very controversial when first introduced, and to a lesser extent today as well. However, as a video card benchmark, it does provide a good gauge as to how effective your video card can render complex scenes.
DX9 is used and visuals were quite impressive at introduction and continue to be so even today. Performance of these cards begins to show the disparity among them. The RX600 is simply overmatched by the complex shader and intense polygons of 3DMark03. The Radeon X700Pro really shows a good lead over the PCX5900 as it should given the added Vertex Shaders (2 versus 6).
3DMark05 can be frustrating for most. Too see my high end FX55@3GHz with a tweaked out 6800Ultra card scoring 4,000-5,000 points is simply humbling. However, in hindsight, this has been the story ever since 3DMark 2000 was introduced to replace the previous 99 variant. Scores here are simply horrible for all cards. In comparison the Radeon X700 Pro wins, however the whole benchmark is basically a slide show, so I would go around bragging about these scores to anyone. The PCX 5900 runs poorly on this benchmark and these results were repeatable.