I create 2+ simple UBO's with vec4 colorA's and colorB's like so. I only get a black screen using the UBO creation / binding process with std140. How do I specify which index while using glCreate and get this mess to work so I can choose colorA or colorB?
//APP
glCreateBuffers(1, &testUBO);
glNamedBufferData(testUBO, sizeof(glm::vec4), 0, GL_DYNAMIC_DRAW);
glGetNamedBufferParameterui64vNV(testUBO, GL_BUFFER_GPU_ADDRESS_NV, &uboScene_64);
glMakeNamedBufferResidentNV(testUBO, GL_READ_ONLY);
glm::vec4 myVec4 = glm::vec4(0.f, 1.f, 0.f, 1.f); //add to structs.h
glNamedBufferSubData(testUBO, 0, sizeof(glm::vec4), &myVec4
//SHARED HEADER
typedef glm::vec4 vec4;
layout(std140, binding = 0) uniform sceneBuffer
{
vec4 colorA;
};
layout(std140, binding = 1) uniform objectBuffer
{
vec4 colorB;
};
//SHADER PROGRAM
void main()
{
Ci = colorA;
Ci = colorB;
}
Given this shader:
//GLSL
layout(std140, binding = 0) uniform sceneBuffer
{
vec4 colorA;
};
layout(std140, binding = 1) uniform objectBuffer
{
vec4 colorB;
};
And this C++ buffer initialization code:
//Create scene buffer.
glCreateBuffers(1, &sceneUbo);
glNamedBufferStorage(sceneUbo, sizeof(glm::vec4), 0, GL_DYNAMIC_STORAGE_BIT);
glm::vec4 myVec4 = glm::vec4(0.f, 1.f, 0.f, 1.f);
glNamedBufferSubData(sceneUbo, 0, sizeof(glm::vec4), &myVec4);
//Create object buffer
glCreateBuffers(1, &objectUbo);
glNamedBufferStorage(objectUbo, sizeof(glm::vec4), 0, GL_DYNAMIC_STORAGE_BIT);
glm::vec4 myVec4 = glm::vec4(0.f, 1.f, 0.f, 1.f);
glNamedBufferSubData(objectUbo, 0, sizeof(glm::vec4), &myVec4);
Here is what the the NV_uniform_buffer_unified_memory
"bindless" code looks like:
//Get addresses
GLuint64 sceneUboAddr;
glGetNamedBufferParameterui64vNV(sceneUbo, GL_BUFFER_GPU_ADDRESS_NV, &sceneUboAddr);
glMakeNamedBufferResidentNV(sceneUbo, GL_READ_ONLY);
GLuint64 objectUboAddr;
glGetNamedBufferParameterui64vNV(objectUbo, GL_BUFFER_GPU_ADDRESS_NV, &objectUboAddr);
glMakeNamedBufferResidentNV(objectUbo, GL_READ_ONLY);
//You have to call this to turn on bindless buffers.
glEnableClientState(UNIFORM_BUFFER_UNIFIED_NV);
//0 represents the scene UBO's `binding` from GLSL:
glBufferAddressRangeNV(UNIFORM_BUFFER_ADDRESS_NV, 0, sceneUboAddr, sizeof(glm::vec4));
//1 represents the object UBO's `binding` from GLSL:
glBufferAddressRangeNV(UNIFORM_BUFFER_ADDRESS_NV, 1, objectUboAddr, sizeof(glm::vec4));
Note that this extension effectively requires glEnable/DisableClientState
, which is a function that was removed from the core profile. So you kind of need to use a compatibility profile to use it.
And just to prove that the non-bindless code is hardly "needlessly complex", here it is:
//The first 0 represents the scene UBO's `binding` from GLSL:
glBindBufferRange(GL_UNIFORM_BUFFER, 0, sceneUbo, 0, sizeof(glm::vec4));
//1 represents the object UBO's `binding` from GLSL:
glBindBufferRange(GL_UNIFORM_BUFFER, 1, objectUbo, 0, sizeof(glm::vec4));