The calculation time is directly proportional to the number of points. There is rarely any point in using 500 points unless there are very detailed features in the transmission spectrum for which you wish to have a very fine resolution. However, you don't know if you have such features until you have studied the spectrum
So the best idea is to use fewer points (100 is usually enough) first and then "zoom in" if needed.
The interesting region for transport is a window around the Fermi energy (E=0 eV) corresponding roughly to the bias you intend to apply. Such biases are typically in the range 1-2 V, thus this is the usual range for plotting the transmission spectrum. Sometimes you can extend it to, say, +-5 eV to get an overview. States far below and above the Fermi energy do not conduct anyway, since they are either fully occupied or empty.
Finally, the calculation time also decreases basically linearly with the number of parallel nodes, at least if the nodes are "true" individual MPI nodes (separate machines in a cluster). Thus by using, say, 10 MPI nodes, and going down to 100 points, you can reduce the calculation by a factor of 50 or so, compared to the serial calculation with 500 points.
