Hi Gregory, > Linux uses a copy-on-write for the memory image of forked processes. Thus, > you may also get significant memory savings by launching a single R process, > loading your large data object, and then using fork::fork() to split off the > other worker process.
True. The problem is each and every instance has its own dataset (but they are only similar). For example, QA environment might have about 90% of the rows similar to Showcase environment. Hence we need to have multiple instances instantiated with its own nearly similar dataset. This in turn makes it necessary to have some kind of de-duplication running on the entire memory image. - Varadharajan ______________________________________________ R-devel@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-devel
