ily of file systems are layered on B+ trees) and several production databases. If you're interested in explor-
7 / 22
Cache-Oblivious Binary Search Trees
To get maximal efficiency from a B-tree, it's necessary to know something about the size of the disk pages
or cache lines in the machine. What if we could build a BST that minimized the number of cache misses
during core operations without having any advance knowledge of the underlying cache size? Such a BST
is called a cache-oblivious binary search tree and, amazingly enough, we know how to design them by
adapting techniques from van Emde Boas trees.
Do'stlaringiz bilan baham: