Programming and animation

GLSL shaders tips: NVidia and AMD difference.

GLSL shaders notes

In the course of work with the GLSL shaders I have a few helpful notes on compatibility:
1) # version 120 – a string indicating the version of the shader should be the first in the code, otherwise ATI swears at compile time. I have this error was the result of the shader assembly of parts.

2) most of the old ATI and INTEL cards do not support glsl shaders, instead they have only enhance GL_ARB_VERTEX_PROGRAM, GL_ARB_FRAGMENT_PROGRAM. Out here there is very concise – download library CG from NVidia and recompile the shaders into the arb programs.

For example: cgc-glsl shader.glsl-profile arbvp1-o vertex.vp

In this case, profile arbvp1 for vertex programs and arbfp1 – for the fragment.

3) The emergence glsl shaders, command glGetString complemented with new possible argument GL_SHADING_LANGUAGE_VERSION, which returns the version of the glsl shader.

4) Remember that NVidia has a very useful utility ShaderPerf – analysis of the number of shader instructions, and the speed of execution, as well as sweep shader performance for different video cards and a family of graphics NVidia chipsets. There is also an opportunity to compare the performance of two different shaders to each other.

5) Do not specify real numbers with f at the end as 1.0f, simply write 1.0. Also, ATI is very picky to the writting a real number with no fractional part.

6) Useful utility OpenGL Extension Viewer with a database of support OpenGL extensions for most video cards. http://www.realtech-vr.com/glview/

GLSL error – pre-mature EOF

At a ATI video cards shaders refuse to compile reference to this error. The reason – comments in Russian.

GLSL: missing uniform location

I have faced with such a feature of the GLSL compiler. If the variable is not used in the shader, then find its location there would be no possibility, it is simply truncated from the shader.

The search function on behalf of the Uniform position.
int loc = glGetUniformLocationARB ( , );

Moreover, even if caught insidious, like the variable is used, but the result of further calculations is not, and that’s why, again, to find the ancestral variable is not possible.

GLSL error C7549

Compiling of my old glsl fragment shader was fall down with warning C7549: OpenGL does not allow C style initializers and warning 

In fragment shader I have float array

To make it work properly, I add #version 120 command to my fragment shader, and rewrite array like this: