> Interesting how if anything there's a slight decrease in
performance.

This is not all that surprising, since there's a bunch of TLB
invalidation overhead needed in the single-process-mfs case which
isn't needed in the current scheme..

> The full CPU utilitisation is obviously because there's no context
> switch overhead at all...  

it's not the context switch overhead, it's merely improper
accounting.. a lot of the work you're causing happens in the mount_mfs
process, but bonnie doesn't know that..

					- Bill