NVIDIA has recently unveiled some spicy info on their latest G92 developments. This is the successor of the current G80 GPU and according to Michael Hara, vice president of the "green" company, the chip will be ready by Christmas 2007. This means that NVIDIA will try to update their high-end graphics card sector every fourth semester in the years to come. This release cycle was adopted with the G80 GPU last year, leaving the mid-range and low-end products for a spring release.

While ATI postponed their high-end R600 GPU series in order to be released with the mainstream and entry level models (this did not work according to plan, as the HD2600 and HD 2400 cards have been postponed for a June release), NVIDIA doesn't see the benefit in releasing all three ranges of cards together. This is quite awkward for NVIDIA, considering that the mainstream and entry-level products yield the greatest income for the company.

However, this accelerated release cycle puts AMD/ATI in a delicate position. Although AMD is readying the Barcelona quad-cores for a Q3/Q4 2007 release, ATI is supposed to release the R650 GPUs in Q3 this year. This gives NVIDIA a one year advantage over ATI, because, as I mentioned before, the GeForce 9800 series should hit the market in Q4 2007.

NVIDIA also claims that the upcoming G92 is set to break the 1 Teraflops barrier. The 8800 is capable of about 330 Gflops, which means the green team is suggesting that the 9800 could be three times more powerful. Keeping in mind that ATI's R600 can deliver 450 GigaFlops, this would be an outstanding generation leap for NVIDIA, as it usually makes its updated generations two times faster than the previous ones. ATI has some similar performance rates with their upcoming FireStream architecture, but what exactly are we going to do with all this graphics power? When we think of games such as Crysis and Unreal Tournament 3 cranked up to 2560X1600 and full eye candy activated, 1 Teraflops may seem just enough for a smooth gameplay.