2Q is clever, but certainly not the only improvement over LRU for page caching. Of interest, also, is LIRS, which uses the timing between the last two accesses

2Q is clever, but certainly not the only improvement over LRU for page caching. Of interest, also, is LIRS, which uses the timing between the last two accesses to decide on a caching strategy. There is a good comparison of page caching algorithms here: http://people.cs.vt.edu/~butta/docs/sigmetrics05_kernelPrefetch.pdf http://people.cs.vt.edu/~butta/docs/sigmetrics05_kernelPrefe... . However, the conclusion seems to be that for non-synthetic loads, the difference is fairly minimal.