I understand that the Rosetta code is very complex and the code is in the hundred thousands of lines range but at least on my Mac Pro the excutable runtime size is running well over 300 meg as compared to about half that for the prior Rosetta C version. As I have 4 gig of memory and only four cores this is not an issue except I fear for the efficiency of the code. I hope there are efforts underway to decrease this footprint for the sake of increasing the science done. It seems the app is doing way too much in main memory and too little in the smaller but much faster cpu caches. I could be wrong and it could need this large dataset in which case main memory much than on the disk. I just hate bloat due to inefficient code. I know code optimization generally comes after correctness and feature completeness but I am sure the developers understand in this app efficiency is a critical requirement. This is due to the fact any inefficiency is magnified thousands of times over and not only increase the use of fossil fuels and pollution (from extra power required by computers) but also as well puts a heavier cost on voluteers with less benefits for mankind. But hey no pressure :).

For me, typical workunits take about 500 MB, and I've heard of a few that took up to 1.2 GB, possibly due to the memory leak in some of the recent versions of the software. Looks like Rosetta@home is going to some extra effort to give you workunits that require less memory than usual.