Oh yeah ur right. :) The file system itself is written in the stripes and stripe boundaries don't have to correspond to cluster boundaries although I think this would be advantageous. 1 cluster -> 1 stripe would be the optimum speed configuration I think.