Integrated graphics are the future.
But not for quite a while yet.
Future technologies such as eDRAM and the like will allow a graphics card to act autonomously on a single chip. The other operations that the north bridge has to do are tiny in comparison to graphics power, and so it just makes sense that one day, we may buy north bridge chip upgrades for a graphics upgrade instead of a new card.
On other topics:
nVidia will not use separate RAM for their integrated graphics on the Hammer platform. nVidia has stated that this defeats the entire purpose of integrated graphics today: price.
nVidia has already pioneered a socket-based system for their GeForce4 Go platforms, something which may eventually make its way to the PC.
The moniker NV31 does seem like a value product name (ex. NV11 = GeForce2 MX). This product will likely have four pixel pipelines, fewer vertex pipelines, and all the programmability of the NV31. nVidia has always used the first number in the codename to describe a parts' 3D capabilities. I don't see why they'd stop now. This is very exciting to me, as I think that it would be amazing if nVidia could have a value DX9 part out within six months! Such a thing could really bring DX9 to bear in a way we have yet to see with DX8.
But not for quite a while yet.
Future technologies such as eDRAM and the like will allow a graphics card to act autonomously on a single chip. The other operations that the north bridge has to do are tiny in comparison to graphics power, and so it just makes sense that one day, we may buy north bridge chip upgrades for a graphics upgrade instead of a new card.
On other topics:
nVidia will not use separate RAM for their integrated graphics on the Hammer platform. nVidia has stated that this defeats the entire purpose of integrated graphics today: price.
nVidia has already pioneered a socket-based system for their GeForce4 Go platforms, something which may eventually make its way to the PC.
The moniker NV31 does seem like a value product name (ex. NV11 = GeForce2 MX). This product will likely have four pixel pipelines, fewer vertex pipelines, and all the programmability of the NV31. nVidia has always used the first number in the codename to describe a parts' 3D capabilities. I don't see why they'd stop now. This is very exciting to me, as I think that it would be amazing if nVidia could have a value DX9 part out within six months! Such a thing could really bring DX9 to bear in a way we have yet to see with DX8.