The 7900GTX beats the X1900XTX in some games when showing raw rendering. Since the 7900GTX can't render the features that the X1900s can they had to cut the major advantage of the X1900 cards so that it would be fair comparison.
The 7900GTX is like a 400 horse power car and the X1900XTX is like a 370 horse power car with extra luxuries. The 7900GTX has a slightly more powerful engine the the X1900XTX but the X1900XTX has nice add-on features that makes the driving experience more enjoyable. Since the 7900GTX doesn't have those features a review company can only compare things that both cars can do, so in case of graphics benchmarks they can only compare the raw power of the engines. While the 7900GTX does have TR SSAA, it doesn't have the great effect that HQAF has. I like HQAF over TR SSAA.
People with 7900GTXs can go ahead and get an extra 7fps while I still get a very nice framerate with the pleasures of HQAF, AAA, and HDR+AA. Like 003 said, right now the frame rates are so high from these cards that if you compared two different rigs with similar setups you won't be able to tell the X1900XTX from the 7900GTX with no identifible settings (no HQAF, AAA, etc.)
You can't compare these cards just by the benchmarks that were nerfed down to make it fair. Is it fair to compare the 7900GTX with raw settings compared to the X1900XTX with HQAF on? No it's not. This is also why I don't look at [H]ardOCP benchmarks because they put on the extra features; like in BF2, it compares with the X1900XT giving HQAF which would obviously reduce frames, resulting in an unbalanced comparison.
I guess if you feel like having 100fps over 94fps with raw settings, go ahead and spend 50 more dollars and waiting for it to be instock to get the 7900GTX.