#include "Thread.hlsl"

#define MAX_DIM_GROUPS 1024
#define MAX_DIM_THREADS (THREAD_SIZE_VAL * MAX_DIM_GROUPS)
#pragma kernel BitonicSort 
#pragma kernel InitKeys

struct MeshTriangle
{
    int3 triangleIndices;
    float distanceToCamera;
    int isVisible;
};

uint trianglesCount;

int block; 
int dim;
uint count;
RWStructuredBuffer<uint> Keys;
StructuredBuffer<MeshTriangle> Values;
StructuredBuffer<int> IntValues;

THREAD_SIZE
void BitonicSort(uint3 id : SV_DispatchThreadID)
{
    uint i = id.x + id.y * MAX_DIM_THREADS;
    uint j = i ^ block;

    if (j < i || i >= count)
        return;

    uint key_i = Keys[i];
    uint key_j = Keys[j];
    float value_i = 1000000.0;
    float value_j = 1000000.0;
    if (key_i < trianglesCount)
    {
        const MeshTriangle mt = Values[key_i];
        if (mt.isVisible > 0)
        {
            value_i = mt.distanceToCamera;
        }
    }
    if (key_j < trianglesCount)
    {
        const MeshTriangle mt = Values[key_j];
        if (mt.isVisible > 0)
        {
            value_j = mt.distanceToCamera;
        }
    }

    float diff = (value_i - value_j) * ((i & dim) == 0 ? 1 : -1);
    if (diff > 0)
    {
        Keys[i] = key_j;
        Keys[j] = key_i;
    }
}

THREAD_SIZE
void InitKeys(uint3 id : SV_DispatchThreadID)
{
    uint i = id.x + id.y * MAX_DIM_THREADS;
    if (i < count)
        Keys[i] = i;
}