__global__ void fun_kernal(int *A, float *B, float *X, char *C, int steps){
	int i, j;
	__shared__ int shA [N];
	__shared__ float shB[N];
	__shared__ float shX[N];
	__shared__ char shC[N];

	int index = blockIdx.x * blockDim.x + threadIdx.x;

	int allthreads=gridDim.x*blockDim.x;

	for (int k = index; k < steps * allthreads; k += allthreads){
	//for(int k = steps*index; k < steps + steps * index; k++){
		shA[k]=A[k];
		shB[k]=B[k];
		shX[k]=X[k];
		shC[k]=C[k];
	}
	__syncthreads();

	index*=steps;
	for (int i = 0; i < steps; i++)
		//for (int j=0; j<M; j++)
			shX[index + i] = (float)shA[index + i] * shX[index + i] + (shB[index + i] - shX[index + i]) / shC[index + i];
	__syncthreads();

	for ( int k = index; k <steps + index; k++ )
		X[k] = shX[k];
}