NVIDIA details the evolution of GPGPU computing architectures towards
From 4Gamer comes a very interesting article detailing the evolution of graphics architectures focused on GPU accelerated computing (GPGPU) from Nvidia, and their plans for a future where the GPU accelerated computing will increasingly greater role in the games.
Tesla (not to be confused with the Tesla product line), was the first graphics architecture from Nvidia GPGPU computing focused, and that was in force in five product lines GeForce: 8000 Series, 9000 Series, 100 Series, 200 Series and 300 Series (although there are GeForce 400 Series low-end based on Tesla as the GeForce 405), becoming the longest running Nvidia architectures, far exceeding its predecessor: CineFX architecture, which was used in three product lines GeForce (5000 Series, 6000 Series and 7000 Series).
Tesla was replaced by Fermi graphics architecture used in two GeForce product line (400 Series and 500 Series, but is also present in some low-end 600 Series) and in turn was replaced by the current Kepler architecture , which is used in GeForce 600 Series GPUs, and we know that future GeForce 700 Series GPUs make use of a slightly evolved version of this same architecture.
GPGPU computing in games
Perhaps the most interesting (for gamers) This lengthy article is the vision of GPGPU computing Nvidia, which will be present in more and more games, for Nvidia believes that traditional graphic APIs such as directx and OpenGL have stalled, which opens the way to new techniques and post-processing shader graph running on GPGPU APIs like CUDA, DirectCompute (part of DirectX 11 and above) and OpenCL as deferred shading techniques, global and interactive ilumination ilumination indirect (octres and voxels) used in the game Battlefield 3 and Unreal Engine 3.
It is noteworthy that there are several techniques used to process DirectCompute APIs as effects and filters, among which we can mention MLAA, FXAA, or deferred rendering technique: Forward + (based on DirectCompute) used in the game Dirt Showdown. So as mentioned by Nvidia, silently GPGPU computing is here to stay in games.Tags: API, architecture, CUDA, deferred shading, DirectCompute, directx, DirectX 11, directx 11.1, evolution, Fermi, games, GPGPU, kepler, Nvidia, octree, Tesla, voxel