The Balanced Cube
Sequential computers spend a large fraction of their time manipulating ordered sets of data. For these operations to be performed efficiently on a concurrent computer, a new data structure for ordered sets is required. Conventional ordered set data structures such as heaps, balanced trees, and B-trees  have a single root node through which all operations must pass. This root bottleneck limits the potential concurrency of tree structures, making them unable to take advantage of the power of concurrent computers. Their maximum throughput is O(1). This chapter presents a new data structure for implementing ordered sets, the balanced cube , which offers significantly improved concurrency.
KeywordsAssure Bors Rebalance
Unable to display preview. Download preview PDF.