Data Structures for Multi-Core

该项目旨在改进数据结构，以减少多个处理器内核造成的瓶颈效应。用于典型服务器平台的硬件具有越来越多的处理器核心。当访问这些服务器的共享内存中结构化的数据时，内核数量的增加会产生瓶颈效应。这些竞争的数据结构限制了服务器的性能，因此需要新的算法。该项目建议放宽传统的一致性标准，以提供高并发性，并利用乐观执行，同时进行，但可能会回滚，这取决于他们遇到的其他并发执行的冲突。并发数据结构将允许应用程序性能随着更多数量的硬件核心而扩展。

The project intends to improve data structures to reduce the bottleneck effect caused by multiple processor cores. The hardware used for a typical server platform has increasing numbers of processor cores. This growing number of cores creates a bottleneck effect when accessing the data that are structured in the shared memory of these servers. These contended data structures limit the server performance and new algorithms are necessary. The project proposes to relax traditional consistency criteria to provide high concurrency and to leverage optimistic executions that proceed concurrently but may roll back depending on the conflicts with other concurrent executions they encounter. The concurrent data structures would allow application performance to scale with higher numbers of hardware cores.