I'm not sure how we missed this in our initial testing of Creo 2 but the Fast HLR in drawings appeared to offer a genuine speed improvement versus waiting for all the views to re-render (through 1 CPU core!). However we didn't appear to read the fine print. Apparently this is actually how it works today:
"Fast HLR rely on multi-pass (3 for dim-hidden) rendering to generate HLR display. And will require these multiple passes for each view in the drawing. So if the drawing has 5 views there will be 15 draws to generate the HLR display."
"Non-Fast HLR on the other hand generates a flat representation of HLR data (a new object is created which contain only the HLR lines) so redisplaying it will be much faster. This could also be seen with even a single drawing view in the drawing."
The point is that this option provides benefit only when setting up the view styles the first time. It actually induces a disadvantage when tweaking drawings to conclusion (i.e. the last 20 percent that takes 80% of the effort/time). Our user base reports that working in large assembly drawings is a tedious affair with Creo Parametric versus other software of old irrespective of the various config options and so on.
My suggest for a product idea is "please fix drawing performance for larger assemblies". We spend millions of engineering dollars each year in this mode and no amount of MBD features are going to change that in the next few years in all liklihood.
