"RAM permitting" is the big question.
The new process/material is said to offer considerable performance improvements, which is always good, but RAM will still (always?) be an issue.
That said, I've been thinking more about other ways to approach a farm setup. Earlier today, I just about ripped out my very small portion of hair when I accidently clicked on a second open session in Logic, only to be faced with a painfully long VI sample reloading fiasco, from which there was no escape. It started me thinking that, actually, I don't think I want VIs on my sequencer machine at all anymore -- or any samples, for that matter. I know others have talked about the benefits of this philosophy (can't recall who, exactly), but it really seems like the way to go. You could easily have headless PCs all boot right into their templates, and you'd never have to mess with them at all!
Now, with that in mind, I started poking around on my PC. I noticed that the PC doesn't have the strange limitation that all plugin instance RAM gets loaded into a separate vsl-server process. Rather, it appears that the memory is all contained in the plugin host's process (Plogue Bidule, in my case). So, I'm wondering how much RAM one could put in a PC and address *if they used multiple hosts*. That is, if I used, say, Audiomulch and Bidule (or copies of Bidule), could I access around 3 GB "live", and without any 3 GB switch? Or is the "switch" a limitation of the system itself, regardless of how much a single process is requesting. I can't see why not, as there's no need for any single process to access more than it's 1.5-ish GBs, but I wonder if anybody knows for sure. If this is the case, one could do quite well with a couple of slave PCs, running 4+ GB of RAM, and two hosts on each machine.
Does that work in PC land? How much RAM can a PC physically address anyway?... If I could install 8 GB of RAM in a PC, could I access it all providing I accessed it using 3 or 4 different processes? I don't know how XP handles RAM, so sorry for the goofy questions.
J.