In engineering problems, $Re = 10^{5}$ is quite common and not considered very large. Even if the numerical scheme is efficient enough to require only 100 operations per grid point per time step, a computer performing $10^{8}$ operations per second would still require 3.6 years of computation. If $Re = 10^{6}$, the computation time would be about 1000 years. Here we have not even considered additional numerical issues such as the number of grid points required to resolve the smallest scales.
Although a computational speed of $10^{8}$ operations per second is already quite fast, the rate of increase in computational power is still far from meeting the demands of directly solving high-Reynolds-number turbulent flows. Therefore, it can be seen that direct numerical solution of the N-S equations is not yet practical for engineering applications and will likely remain impractical for the foreseeable future. The development and improvement of turbulence modeling thus remain the only viable way to solve turbulence problems.