Question

我想要一个在输出缓冲区中写入1的计算着色器。我编译着色器并将其附加到程序没有问题，然后我调用 glDispatchCompute（）函数，我等到计算着色器结束。但是当我看到数组时，只有0个。

谁能告诉我哪里出错了？

这是我的计算着色器代码：

#version 430 core

layout  (local_size_x  =  2)  in;

layout(std430, binding=0) writeonly buffer Pos{
    float Position[];
};

void main(){
    Position[gl_GlobalInvocationID.x] = 1.0f;
}

这是我的主要内容：

GLuint program_compute = 0, SSBO = 0;

//...(Create, compile and link the program with the shader)...

vector<GLfloat> initPos;
int num_numeros = 12;

for (int i = 0; i < num_numeros; i++){
    initPos.push_back(0.0f);
}

glUseProgram(program_compute);

glGenBuffers(1, &SSBO);
glBindBufferBase(GL_SHADER_STORAGE_BUFFER, 0, SSBO);
glBufferData(GL_SHADER_STORAGE_BUFFER, num_numeros * sizeof(GLfloat), &initPos, GL_DYNAMIC_DRAW);

glDispatchCompute(num_numeros/2, 1, 1);
glMemoryBarrier(GL_SHADER_STORAGE_BARRIER_BIT);
glBindBufferBase(GL_SHADER_STORAGE_BUFFER, 0, 0);

for (int i = 0; i < num_numeros; i++){
    cout << "p" << i << ": " << initPos[i] <<  endl;
}
cout << endl;

编辑：最后它有效。谢谢BDL。

正如BDL所说，我忘了从GPU内存中读取缓冲区。现在它有效。这是新代码：

GLuint program_compute = 0, SSBO = 0;

// ...The same code as above

glDispatchCompute(num_numeros/2, 1, 1);
glMemoryBarrier(GL_SHADER_STORAGE_BARRIER_BIT);
glBindBufferBase(GL_SHADER_STORAGE_BUFFER, 0, 0);

glBindBuffer(GL_SHADER_STORAGE_BUFFER, SSBO);

GLfloat *ptr;
ptr = (GLfloat *) glMapBuffer(GL_SHADER_STORAGE_BUFFER, GL_WRITE_ONLY);
initPos.clear();

for (int i = 0; i < num_numeros; i++){
    initPos.push_back(ptr[i]);
}

glUnmapBuffer(GL_SHADER_STORAGE_BUFFER);

for (int i = 0; i < num_numeros; i++){
    cout << "p" << i << ": " << initPos[i] <<  endl;
}
cout << endl;

编辑：谢谢Andon M. Coleman。

我从只写缓冲区读取。这是固定的行：

ptr = (GLfloat *) glMapBuffer(GL_SHADER_STORAGE_BUFFER, GL_READ_ONLY);

Answer 1

glBufferData 将initPos的内容复制到SSBO中。然后着色器在缓冲区上运行，而不是在cpu内存阵列上运行。除非你在某处读取从GPU到CPU内存的缓冲区，否则initPos永远不会改变。

OpenGL Compute Shader SSBO

1 个答案: