[Custom Pipeline] Passing uniform arrays through code

[panthavma]

Hello !
I'm trying to pass values to my shader, stored as an array. I've found a way for now by setting the uniform array directly (array[id]), but I think this is pretty inefficient performance-wise. Is there a better way to do this easily ?

// Shader code
uniform int myArray[128];

# Custom Pipeline code
MaterialShader.uniforms['myArray[0]'].set_value(16)
MaterialShader.uniforms['myArray[1]'].set_value(64)
MaterialShader.uniforms['myArray[2]'].set_value(48)
# versus something like MaterialShader.uniforms['myArray'].set_value([16,64,48])

[pragma37]

I think there's no way to set array uniforms like that in OpenGL.

In those cases I would use a Uniform Buffer Object:
https://github.com/bnpr/Malt/blob/Development/Malt/Render/Common.py

Malt/Malt/Pipeline.py

Line 310 in a987506

models_UBO = UBO()

Malt/Malt/Pipeline.py

Line 381 in a987506

batch['BATCH_MODELS'].bind(shader.uniform_blocks['BATCH_MODELS'])

[panthavma]

Thanks !
I managed to make it work, but I have a second question now when packing ints, why aren't they packed by default when using ctypes (as in, why do I need the *4 and why did you use ivec4 when faced with a similar problem) ?

Here's my code that works for passing an array for future reference:

# Python side
dataBuffer = (ctypes.c_int32 * (4 * BUFFER_SIZE))()
dataBuffer[10 * 4] = 314
dataUBO = UBO()
dataUBO.load_data(dataBuffer)
dataUBO.bind(TestShader.uniform_blocks["BLOCKNAME"])

// GLSL Side
layout(std140) uniform BLOCKNAME
{
	uniform int variableName[BUFFER_SIZE];
};

// variableName[10] == 314

[pragma37]

Because the std140 layout requires each element to have a 16 bytes alignment:
https://www.khronos.org/opengl/wiki/Interface_Block_(GLSL)#Memory_layout