Ah the wild, wild world of hardware manufacturing! Intel has decided to bump forward its NDA on Penryn’s performance to before its launch. Hey! Why not? NVIDIA has already managed the same maneuver to be sure to precede the new Radeon HD 3000 release! Hence comes the second major review published today, after months of wandering in the wilderness, waiting for those new products to deem it time to show up, with (of course) the availability of last minute of test samples and drivers. It’s not like we were closing on a specific date or even a particular season, one traditionally good for manufacturer’s sales or anything like that…

The behavior of the two graphic cards giants is all the more unusual (and reprehensible) after the launch of their new Direct X 10 architectures. Generally, high end cards quickly give way to mid-range cards with more interesting performance-to-price ratios. This time it hasn’t worked, in part because of the cost of transistors the new architectures brings (due to new API support and unified architecture), making the arrival of new (cheap-to-produce) GPUs harder, but efficient using a process that has been mastered couple of months ago. NVIDIA was first with its disappointing GeForce 8600 GTS and GT, and yet hardly measured up to by the Radeon HD 2600 that arrived later on. Never since, perhaps, the first GeForce 3 had the gap between high end and mid-range, been so important. This gap pleased manufacturers (especially NVIDIA), since gamers logically abandoned those cards in favor of the less out of reach high end models, with the GeForce 8800 GTS 320 MB in the spotlight. The card was introduced at the beginning of the year for no less than $300.

Finally, manufacturers decided it was time to offer genuine mid-range items, reasonably close to the high end ones in terms of performance. They’ve also gotten great help, from the availability of new processes in foundries (TSMC leading the way).

Is the GeForce 8800 GT the ultimate card for broke gamers?