PhysX Fluidmark uses shaders not compliant with OpenGL 2.0

[Dex]

I gave a try to run Fluidmark on intel integrated graphics, which allegedly supports OpenGL 2.0 and what I realized was that it fails because shader implemented in Fluidmark uses simplicit type conversion which is forbidden in GLSL 1.10 (introduced in 1.20), while it does not declare #version 120. Probably nvidia/ati does not check for that, but using it this way looks likes violating the spec:

GLSL 1.10.59 spec says in section 5.9 Expressions
The arithmetic binary operators add (+), subtract (-), multiply (*), and divide (/), that operate on integer and floating-point typed expressions (including vectors and matrices). The two operands must be the same type, or one can be a scalar float and the other a float vector or matrix, or one can be a scalar integer and the other an integer vector.

The part of shader that is doing the implicit conversion:
#define NUM_BLUR_TAPS 12
...
float shadowColor = 0.0;
for (i=0; i<NUM_BLUR_TAPS; i++)
        shadowColor += ...
shadowColor /= NUM_BLUR_TAPS; // <<< error: float type divided by integer const
 

Can you confirm / fix / comment?

Thank you
Dexter

[JeGX]

yep you're right, this code is not really legal. But it works on Radeon/GeForce and maybe on S3 Chrome. You can fix it yourself with:

Code: [Select]

shadowColor /= float(NUM_BLUR_TAPS);
Try this fix and let me know.

[Dex]

Yep, it works! You could make the change in some next release...
Thanks