Comparison of runs with different transverse size for M = 20 parallel shock (Run A is five times larger than Run B; see Table 1). Left panel: magnetic field profiles, as in the legend; B and max B correspond to the average over y and the maximum of B_tot(x, y), as a function of x. max B in Run B is not shown since it almost coincides with B. The averaged profiles are quite similar in the two runs, even if the spread from the mean value may locally be quite large in Run A. Right panel: time evolution of non-thermal spectra for Run A (dashed lines) and B (solid lines). Spectra agree very well until , after which the diffusion length at E_max becomes comparable with the box size in Run A.