Taking a Deep Look at NVIDIA's Fermi Architecture

[Rob Williams]

It has taken a bit longer than we all had hoped, but last week, NVIDIA gathered members of press together to dive deep into its upcoming Fermi architecture, where it divulged all the features that are set to give AMD a run for its money. In addition, the company also discussed PhysX, GPU Compute, developer relations and a lot more.

You can read our full look at all that's happening with NVIDIA here and discuss it here!

 

[Doomsday]

Very nice read. Very informative. The developer stuff might come in handy once i try and get into that field. I'll be most probably heading to UK this September to start my BSc (Hon) in Games Designing!

 

[Rob Williams]

Yes, the developer stuff was -amazing-. I wish I could have talked at greater depth about it all, but I didn't have as much backup information as I wanted (including screenshots), so I didn't want to talk about it too much. I do plan to tackle it a bit more in the future, though, because even as a non-developer, I found it all to be quite exciting.

 

[Kougar]

The perk of having the game devs work closerly with you and putting your brand's logo on their game comes with the latest (and sometimes future unreleased hardware) to develop said game on and make sure it works and performs at reasonable levels. Valve has frequently used unreleased future flagship cards from ATI to develop their games just to name another.

I will mention that a number of people/sites are worried that Fermi is a huge gamble for NVIDIA because of it's unprecedented ~3Billion transistors and massive die size, and all the problems inherent that causes during the manufacturing of said silicon chips. At risk of sounding like an NVIDIA fanboy several sites have done the complex math to extrapolate Fermi's die size, and have indicated it is actually smaller than G80's die in total size.