SearchThreadIdx CUDA C++

SearchThreadIdx.cpp

//CUDA C++
//1D grid of 1D blocks
__device__ int getGlobalIdx_1D_1D()
{
    return blockIdx.x*blockDim.x + threadIdx.x;
}

//1D grid of 2D blocks
__device__ int getGlobalIdx_1D_2D()
{
    return blockIdx.x*blockDim.x*blockDim.y
            + threadIdx.y*blockDim.x + threadIdx.x;
}

//1D grid of 3D blocks
__device__ int getGlobalIdx_1D_3D()
{
    return blockIdx.x*blockDim.x*blockDim.y*blockDim.z
            + threadIdx.z*blockDim.y*blockDim.x
            + threadIdx.y*blockDim.x + threadIdx.x;
}

//2D grid of 1D blocks
__device__ int getGlobalIdx_2D_1D()
{
    int blockId = blockIdx.y*gridDim.x + blockIdx.x;
    return blockId*blockDim.x + threadIdx.x;
}

//2D grid of 2D blocks
__device__ int getGlobalIdx_2D_2D()
{
    int blockId = blockIdx.y*gridDim.x + blockIdx.x;
    return blockId*(blockDim.x*blockDim.y) + (threadIdx.y*blockDim.x) + threadIdx.x;
}

//2D grid of 3D blocks
__device__ int getGlobalIdx_2D_3D()
{
    int blockId = blockIdx.y*gridDim.x + blockIdx.x;
    return blockId*(blockDim.x*blockDim.y*blockDim.z) 
            + (threadIdx.z*(blockDim.x*blockDim.y))
            + (threadIdx.y*blockDim.x) + threadIdx.x;
}

//3D grid of 1D blocks
__device__ int getGlobalIdx_3D_1D()
{
    int blockId = blockIdx.y*gridDim.x + blockIdx.x
                    + gridDim.x*gridDim.y*gridDim.z;
    return blockId*blockDim.x + threadIdx.x;
}

//3D grid of 2D blocks
__device__ int getGlobalIdx_3D_2D()
{
    int blockId = blockIdx.y*gridDim.x + blockIdx.x
                    + gridDim.x*gridDim.y*gridDim.z;
    return blockId*(blockDim.x*blockDim.y)
            + (threadIdx.y*blockDim.x) + threadIdx.x;
}

//3D grid of 3D blocks
__device__ int getGlobalIdx_3D_3D()
{
    int blockId = blockIdx.y*gridDim.x + blockIdx.x
                    + gridDim.x*gridDim.y*gridDim.z;
    return blockId*(blockDim.x*blockDim.y*blockDim.z)
            + (threadIdx.z(blockDim.x*blockDim.y))
            + (threadIdx.y*blockDim.x) + threadIdx.x;
}