Launching in late 2011 is not impossible for NVIDIA
The other day we wrote about four codename from AMD’s South Island, found in their drivers are suggesting that a launch is not far away. But while AMD will have encountered the same problem as NVIDIA and it is still unclear when their new GPUs are ready to show itself. Earlier rumors said that it should happen sometime in the fourth quarter, but it is very likely that AMD should have the same problem as NVIDIA then TSMC is the manufacture of circuit boards of the two.
When TSMC rolled up their production on 40nm happened to be out of contamination by dirt, something that interfered with production and took several months to correct. With luck, the lessons learned from past mistakes and move to 28nm should be much smoother than the turbulent transition to 40nm.
It is question of Kepler, scheduled for next year
Kal-El, code-named of Tegra3, the next SOC Nvidia depth, should be available in a few weeks in late summer. Indeed, now we talk of an exit to the end of September on the chip that will include no fewer than four hearts Cortex A9 while the graphics can count on the presence of twelve cores. A more powerful than the Tegra 2, offering a variety of new features for application developers and – especially – mobile game, as this example will manage the anti-aliasing to avoid aliasing 3D models often noticeable even on small screens of our smartphones. A recent video released by Nvidia demo showed the potential of Kal-El .
Nvidia following a presentation held at the ISC (International Supercomputing Conference) as leaks some details that Kepler will be out in 2012. Remember that due to this new architecture works Nvidia GPU could clearly be delayed by production problems at TSMC. It is therefore reassuring to see Nvidia communicate a roadmap in which Kepler is noted for release next year. This will have two years left at the Fermi architecture, while the architecture replacement for Kepler, named Maxwell, is already planned for 2014.
AMD finished designing its GPUs Southern Islands three months ago and revealed details of its architecture a few days ago and Nvidia may disclose details of its new architecture, perhaps by August. GPUs are expected to have Kepler based on consumption similar to current Fermi, but with a considerably higher yield and squeezing more transistors due to its lower manufacturing process.