There are times when this industry that we're all so interested in is set alight. Those that choose the path of the true exegete call them inflection points, and run with that faster than Usain Bolt. Others call them the times when cans of whoop ass are opened (nota bene, the whooping doesn't always hit its intended target!).
In general, these are the times when one of the competitors slips, or hits an absolute home-run, or some linear combination of the above two possibilities occurs. These are the times when the glorious stories of saviours and cloak-and-dagger development of features are published, much to the excitement of crowds everywhere. Interestingly, it's never an engineer that saves a design, or makes the right calls; intriguingly, and somewhat conveniently, the heroes of these stories come from departments that are not only allowed, but also encouraged to interact and interface with the public -- a consequence of the move towards the knowledge-based economy, no doubt!
Which brings us to today's topic of discussion, NVIDIA's Fermi (Beyond3D codename: Slimer). Troubled by delays, and faced with fierce competition in the form of the earlier-launching and quite excellent ATI Cypress, nobody could say that NVIDIA's latest offspring had an easy life -- then again, no pain, no gain!
We warn you a priori that we don't have any heroic tales to share, nor do we bring news of impending doom of one IHV or another. Our work here at Beyond3D is far less glamorous, and deals mainly with figuring out how these increasingly complex things work. We sincerely hope that once you've reached the end you'll have a better understanding of how Fermi practices its magic. Intro over, we can move on, but not without seeing the single most important picture in this article: the so-called money shot: