So I did have a lot of trouble with shaders in the past because of some "different" implementations of ATI and Nvidia, or implementations deviating from the standard. Example: gl_ClipVertex, using LINE_SMOOTH causing fallback to software rendering, only specific buffer formats for FBOs are accelerated. I found some solutions by try and error, some in forums, some in documents around the web. But not all on the official vendor sites. So the question is if somebody has done a tabular overview of common mistakes or specializations for use with shaders. E.g. tips for special hardware acceleration features like ATIs HiZ, or states causing fallback to software mode and so on. Such a list would greatly improve shader performance and drecrease debugging time. A notation referring to affected hw would be nice either. Somebody knows of such an overview or has interest in contributing with his/her own experience to create one? [Edited by - bene81 on October 31, 2007 5:17:40 AM]

Seconded...

But does anyone know what the differences in implementations are?

– and are there any other graphics hardware manufacturers doing more odd things with what I thought was a clearly defined specification?

Some info can be found here:

http://ati.amd.com/developer/SDK/AMD_SDK_Samples_May2007/Documentations/ATI_OpenGL_Programming_and_Optimization_Guide.pdf
http://ati.amd.com/developer/SDK/AMD_SDK_Samples_May2007/Documentations/Depth_in-depth.pdf
http://download.nvidia.com/developer/GLSL/GLSL%20Release%20Notes%20for%20Release%2060.pdf

http://developer.download.nvidia.com/GPU_Programming_Guide/GPU_Programming_Guide.pdf