The graphs look similar but the differences are important – this one (the correct one), appears to confirm that Peter Denning‘s findings about the working set model versus LRU still hold good, at least in broad terms – though this still suggests LRU has better performance characteristics than might be expected.
But it’s late now and I am going to bed – perhaps more later.
Related articles
- The graph is wrong (cartesianproduct.wordpress.com)