In my attempt to configure a well-optimized VM (OS X host, Linux secondary), I was wondering if it is a good idea to use a virtualized Linux swap file, or if I should just run Linux without swap and rely on the OS X host to provide the swap. e.g. allocating 1 GB RAM to the Linux VM instead of allocating 512 MB to the VM and set a 512 MB swap partition on the VM hard disk file. Instinctively, I'd say the former should be the best approach. since it removes an emulation layer. I've read some forum postings and FAQ's, but wasn't able to make an informed decision. Has someone performed performance tests in boths scenarios? What if the official word (if any) from Parallels? Or does it just not matter?
