Cubenet
makes a list of jobs to expand at smaller point.
looks at mode switch for:
0 - standard local expansion
1 - run server
2 - run client (on system number of threads)
its a bit slapped together but is correct to my knowledge as long as worker nodes are well behaved
has some issues giving out jobs multiple times before updating when the job was last given due to race condition on locks I think but need to look into the issue