Big releases for the end of this year?

[Liquiduk]

I haven't heard of anything for either Nvidia or ATI as for rolling in to next year other then ATI rumoured to be making a new graphic proccessor. Does anybody have any further knowledge on what we have to look forward too?

 

[Duff-Man]

Towards the end of the year we can expect a new revision from both manufacturers, based on a 28nm process. Given that 28nm allows roughly double the transistor density of 40nm, you can expect around twice the horsepower of current top-end models (~1024 cores for Nvidia, ~3200 SPs from ATI).

Beyond that there really isn't much information, though AMD are due to discuss their new GPU in June. As for when they appear, it will depend largely on when TSMC get their 28nm process sufficiently stable to create such large (5 Bn+ transistor) designs. That's anyone's guess, though it's unlikely to be any time before September. My guess would be a lot closer to Christmas.

 

[Marvin]

Towards the end of the year we can expect a new revision from both manufacturers, based on a 28nm process. Given that 28nm allows roughly double the transistor density of 40nm, you can expect around twice the horsepower of current top-end models (~1024 cores for Nvidia, ~3200 SPs from ATI).

Beyond that there really isn't much information, though AMD are due to discuss their new GPU in June. As for when they appear, it will depend largely on when TSMC get their 28nm process sufficiently stable to create such large (5 Bn+ transistor) designs. That's anyone's guess, though it's unlikely to be any time before September. My guess would be a lot closer to Christmas.

In theory twice the horsepower! In practice, it might be more of a compromise to optimise power usage.

 

[Duff-Man]

In theory twice the horsepower! In practice, it might be more of a compromise to optimise power usage.

Very, very true.

Power containment is only going to become more important with each successive generation.

 

[SeeKified]

I thought they were going to use GF for the 28nm jump? Or am I thinking of the next CPU node? *Confused*

 

[Duff-Man]

I thought they were going to use GF for the 28nm jump? Or am I thinking of the next CPU node? *Confused*

AMD are planning to use Global Foundaries for some 28nm designs, but the consensus seems to be that GF is unlikely to be ready for such large designs before TSMC. But we'll see. If there are big problems at TSMC (like there were with the 40nm process), and GF can manage to get "up to speed" very quickly, then AMD could have an advantage in terms of capacity.