template <int BLOCK_SIZE> __global__ void
matrixMulCUDA_4(float* C, float* A, float* B, int wA, int wB)
{
	// Block index
	int bx = blockIdx.x;
	int by = blockIdx.y;

	// Thread index
	int tx = threadIdx.x;
	int ty = threadIdx.y;

	// Index of the first sub-matrix of A processed by the block
	int aBegin = wA * BLOCK_SIZE * by;

	// Index of the last sub-matrix of A processed by the block
	int aEnd = aBegin + wA - 1;

	// Step size used to iterate through the sub-matrices of A
	int aStep = BLOCK_SIZE;

	// Index of the first sub-matrix of B processed by the block
	int bBegin = BLOCK_SIZE * bx;

	// Step size used to iterate through the sub-matrices of B
	int bStep = BLOCK_SIZE * wB;

	// Csub is used to store the element of the block sub-matrix
	// that is computed by the thread
	float Csub = 0;

	// Declaration of the shared memory array Aa used to
	// store the sub-matrix of A
	__shared__ float Aa[BLOCK_SIZE][BLOCK_SIZE];

	// Declaration of the shared memory array Ba used to
	// store the sub-matrix of B
	__shared__ float Ba[BLOCK_SIZE][BLOCK_SIZE];

	// Declaration of the shared memory array Aa used to
	// store the sub-matrix of A
	__shared__ float Ab[BLOCK_SIZE][BLOCK_SIZE];

	// Declaration of the shared memory array Ba used to
	// store the sub-matrix of B
	__shared__ float Bb[BLOCK_SIZE][BLOCK_SIZE];

	// Load the matrices from device memory
	// to shared memory; each thread loads
	// one element of each matrix
	Aa[ty][tx] = A[aBegin + wA * ty + tx];
	Ba[ty][tx] = B[bBegin + wB * ty + tx];

	// Loop over all the sub-matrices of A and B
	// required to compute the block sub-matrix
	for (int a = aBegin, b = bBegin;
		a <= aEnd;
		a += aStep, b += bStep)
	{
		// Load the matrices from sub-matrix A to sub-matrix B
		Ab[ty][tx] = Aa[ty][tx];
		Bb[ty][tx] = Ba[ty][tx];

		// Load the matrices from device memory
		// to shared memory A; 
		// In last iteration skips this part
		if (a + aStep < aEnd) {
			Aa[ty][tx] = A[(a + aStep) + wA * ty + tx];
			Ba[ty][tx] = B[(b + bStep) + wB * ty + tx];
		}

		// Synchronize to make sure the matrices are loaded
		__syncthreads();

		// Multiply the two matrices together;
		// each thread computes one element
		// of the block sub-matrix
#pragma unroll

		for (int k = 0; k < BLOCK_SIZE; ++k)
		{
			Csub += Ab[ty][k] * Bb[k][tx];
		}

		// Synchronize to make sure that the preceding
		// computation is done before loading two new
		// sub-matrices of A and B in the next iteration
		__syncthreads();
	}

	// Write the block sub-matrix to device memory;
	// each thread writes one element
	int c = wB * BLOCK_SIZE * by + BLOCK_SIZE * bx;
	C[c + wB * ty + tx] = Csub;
}