The world of hardware has been resembling something out of a soap opera in recent times.

In the CPU arena, we have the traditional Intel and AMD rivalry. It's interesting to observe the current situation where Intel is caught behind in the performance stakes once again (The first was with the launch of the Pentium 4 processor). Strangely enough, AMD seems to lack pugnacity in capitalizing on a greater market share. The firm's Athlon 64 chip productions have been lackluster, to say the least.

Moving on to the GPU arena, it's open season once again. Both ATI and Nvidia are due to introduce their next-generation wares to the market next month. Despite the obvious animosity between the two manufacturers, deep down, there is also no lack of respect. Both firms have held their specifications tightly to their chest, acknowledging the other party's ability to do a one-up over them.

In the weeks before the actual announcements, both vendors have been handing--or leaking--out specs as if in a poker game. ATI only revealed details after Nvidia showed its hand. Most closely guarded was the number of pipelines in each architecture (or should we say Marchitecture?). For the uninitiated, pipelining in general increases the number of instructions that can be processed in a given amount of time. Think of it as the number of assembly lines in a factory--the more the merrier.

In any case, the top-of-the-line parts will both feature 16 pipelines, or double what we have in our ATI Radeon 9800 XTs today. Despite the obvious increase in speed that these extra pipelines will bring, the new parts do introduce other new features, too. Nvidia appears most notably to have learned from its past experience.

As we've mentioned before, Nvidia's GeForce FX part has not been performing well in standard DirectX9 games, causing the company to relinquish market share. This time, though, the firm's GeForce 6800 line appears to be designed with the standard API (Application Program Interface) in mind, and manages to be the first GPU to support Microsoft's Shader Model 3.0 feature--something that's noticeably missing from ATI's soon-to-be-released X800 line. The latter has maintained that version 3.0 will not look noticeably better over it's version 2.0 implementation.

This round, ATI seems to be marshalling its efforts towards the performance crown instead of competing on the features list. The company's X800 line looks to sport the same feature set that has served the firm's high-end offerings so well thus far. We hazard that all the extra transistors in that spanking new silicon are just devoted to doubling the number of pipelines. Anyway, combined with the higher clock speeds and the manufacturer's new compression technology (3Dc), we're expecting this new GPU core to perform like the old 9700 Pro, but on steroids. In fact, the grapevine has it that the 12-pipeline X800 Pro will beat the 16-pipeline GeForce 6800 Ultra!

However, there is a downside to all that speed. Both top-of-the-line cards require two power connectors to power them. Even more disturbing is the fact that Nvidia recommends the 6800 Ultra be used with a 480W power supply! Let's not forget most desktops these days typically feature only 300W power supplies. With PCI Express and Intel's new BTX (Balanced Technology eXtended) form factor just round the corner, this round's for people with money to burn.

Alas, what is rivalry without the mandatory finger pointing? Like last year's optimization debacle, this round looks no different. The first salvos have already been fired--despite the cards not having hit the retail stands yet!