Some hardware supports texture filtering on floating point textures, some doesn’t. For those cases, it can be emulated within GLSL.
What is the best way to test at run time if texture filtering on floating point textures is supported by the hardware your application is running on?
Any extension which could be indicative?
I dont think its legal to support float textures but not filtering for them.
ARB_texture_float doesnt special case them with regards to filtering, so you should always have filtering available.
You can create a 2-pixels 1-channel 1D texture and call glGenerateMipmaps for it. Then read LOD level 1 and see if it is an average or one of the values.
Kyle_: I agree it is somewhat odd, but sadly that’s the way it is… Depending on the graphics board you use, you can perfectly encounter situations where you can create and use a FP texture (32-bit shadow map in this case) but will be unable to use linear filtering on it. There is some explanation here but it is not definitive: http://www.opengl.org/wiki/Floating_point_and_mipmapping_and_filtering
DmitryM: Thanks, would have liked to avoid it that does seem the most reliable way 