Running for smaller periods or with a smaller timestep for the purpose of adjusting the model should do it. You can also try changing the type of algorithms E+ will use on the different modules to do the calculations, there are some articles on how each type performs and also on how the detailing of your model impacts on performance.
If you wish to distribute on different computers (or different cores on your processor, since each E+ model only uses 1 core at a time at 100%) you can do it in 2 ways to decrease your waiting time:
1st, you may adjust different parts of your model in separate copies and run them simultaneously (I suggest you use no more than 2 if you have 2 cores and no more than 5 if you have 4 cores, not threads, cores... can help if you give the CPU model).
2nd you can split your model (constructions and schedules) in separate files with a smaller amount of objects, maintaining the logic of the whole, this way the time shall decrease greatly and the model adjust can be easier since your HVAC will manage less zones and so will the other energy equations. With the correct assumptions (like putting adiabatic walls on the parts that were in contact before the split, and avoiding double count on internal gains and other energy consumers) you can split the model in 2 or more parts (same suggestion of cores an CPU) and then simply joining the results from the tables at the end.
I am not sure this will help, but those can accelerate model calibration.
Leandro Biondo
Environment Analyst / Meteorologist
Brazilian Forest Services - Environment Ministry