Fast sparse matrix multiplication