Aug 16, 2011
Drawing effects that need to store multiple data elements per-pixel is not an easy task, when you consider an architecture as rigid as highly parallel GPU subsystem. It is even more so, when you want to put irregular number of data elements (e.g. fragments) per-pixel. Current methods often suffer from the low performance pitfalls of non-regular memory references. See how d-buffer storage scheme can help composing your multiple, irregular in terms of per-pixel count fragments into hardware friendly regular data structure. All with the help of GL_ARB_shader_image_load_store, GL_ARB_shader_atomic_counters or simply core OpenGL 4.2.
Category: Developers •
Posted on 08/16 at 02:08 PM
Leave a comment
Next entry: OpenGL Man pages updated to OpenGL 4.2
Previous entry: OpenGL and OpenCL SIGGRAPH BOF presentation slides now online