I recently had occasion to re-implement the R-tree structure [1] which I based on the original code by Guttman. There, each set of child nodes were stored in a

I recently had occasion to re-implement the R-tree structure [1] which I based on the original code by Guttman. There, each set of child nodes were stored in a (hard-coded 1k) page of memory, but there was a macro which allowed n sets of child-nodes in a page, but that value was set to 1. I assumed that this was an experiment which turned out to not work, hence the hard-coded 1 to disable it. As the code was nearing completion I experimented with values other than 1, and found a 3-fold increase in speed for a value of n=8 (IIRC) against a 4k memory page. It's always worth testing your assumptions :-) [1] http://soliton.vm.bytemark.co.uk/pub/jjg/en/code/librtree/ http://soliton.vm.bytemark.co.uk/pub/jjg/en/code/librtree/