Isn't the problem to do with the amount of number crunching involved (we have stability problems running LE in real time as it is) and that PT doesn't create the appropriate temp or preview files, unlike (I don't really know about other sound progs) say Premiere that will fill a temp folder full of rendered scraps - mixes, fades - as it goes along, then if you export your project the prog only needs to do the neccesary number crunch on any bits you haven't yet previewed. So one possible way in PT with minimal re-coding might be that whenever you listen to a track the whole way through a temp wav is generated (which overwrites any previous temp wav) so when you come to bounce to disk the work of rendering is already done, and the bounce can then be done super-quick. After all, who bounces to disk without first having one last listen all the way through?