I think you should take a more detailed look to the performance comparisons (particularly the ones with numbers and resolutions and all): HD58
50 is beating GTX285 in nearly all games (including STALKER CS and Crysis Warhead), some by more, some by less, so no way that the HD5870 is not noticeably faster than GTX285.
On the other hand, in the comparison between HD5870 and GTX295, you can see that STALKER CS runs 38.4 vs 40.5 without AA and 24.2 vs 22.3 with 4xAA (so no 20% lower, even higher with AA) and Crysis warhead is 51.5 vs 53.6 and 43.6 vs 48.5 respectively (~3% and ~11% performance difference respectively) so...
In general, the difference between HD5850 and GTX285 (favorable to the Radeon) is equal or greater than the difference between GTX295 and HD5870 (favorable to the GeForce, if we don't count 8xMSAA results). So we could say that GTX285 is trading blows with HD5850 (favoring the ATi card) and GTX295 with HD5870 (favoring the NVIDIA dual card).
You can take a look to this post if you want to examine better the leaked benchmarks:
http://www.xtremesystems.org/forums/...&postcount=892
Of course, you should note that maybe a change from GTX295 to HD5870 is not so greatest performance wise (probably consumption, heat and price will be another thing), the same way that a change from HD3870 to HD4650 is not. If you want to see if changing from a dual high end chip configuration to the new generation is a gain, you should wait for HD5870 X2, or make an idea with a HD5870 CrossFire (wich
here you can see how it scales, destroying a GTX295 SLI -as it's to expect given that the NVIDIA configuration is a 2x2=4 chips scaling, so much harder to scale).
Bookmarks