experience with semi-empirical MD

 Several people told me, in a reply to my question, that it's possible
 to do semi-empirical molecular dynamics (AM1, PM3) with
 Hyperchem, Mopac and Spartan. Now I wonder if somebody can
 give me some practical information on systems one can treat on a
 state of the art linux or windows 2000 machine. I know this is not
 something fixed and depends on how long one is willing to wait for
 the simulation to finish but I just want to get some feeling. Any
 experience, especially with solution systems, is more then welcome.
 Martijn Zwijnenburg
 Lab. of Applied Organic Chemistry and Catalysis
 Delft University of Technology
 Julianalaan 136
 2628 BL Delft
 The Netherlands
 Tel: 0031-(0)152782691
 Fax: 0031-(0)152784700
 e-mail: M.A.Zwijnenburg -x- at -x- tnw.tudelft.nl
 web page: http://come.to/tock