Most transactional memory (TM) research focused on multi-core processors, and others investigated at the clusters, leaving the area of non-uniform memory access (NUMA) system unexplored. The existing TM implementation...Most transactional memory (TM) research focused on multi-core processors, and others investigated at the clusters, leaving the area of non-uniform memory access (NUMA) system unexplored. The existing TM implementations made significant performance degradation on NUMA system because they ignored the slower remote memory access. To solve this problem, a latency-based conflict detection and a forecasting-based conflict prevention method were proposed. Using these techniques, the NUMA aware TM system was presented. By reducing the remote memory access and the abort rate of transaction, the experiment results show that the NUMA aware strategies present good practical TM performance on NUMA system.展开更多
多核结构上采用由用户显式制导的并行程序设计模型,使用锁和同步变量来实现同步。事务存储模型能够解决由锁机制带来的一系列问题,提高程序的并发性。介绍了在文中提出的一种基于事务存储模型的多核结构(Transactional-Memory based Chi...多核结构上采用由用户显式制导的并行程序设计模型,使用锁和同步变量来实现同步。事务存储模型能够解决由锁机制带来的一系列问题,提高程序的并发性。介绍了在文中提出的一种基于事务存储模型的多核结构(Transactional-Memory based Chip Multiple-Superscaler,TMCMS)上的并行编程模型,以及针对循环程序的执行模型;以FFT程序为例具体介绍了循环结构的并行化方法和编译转换过程。在初步的实验中,将处理单元从1增加到16个时,在所设计的编程模型的支持下,IPC(Instruction PerCycle)有接近线性的增长,说明该并行编程模型能够充分发掘程序中潜在的细粒度线程级并行性,同时保持并行程序设计的简单性。展开更多
Transactional memory （TM） is a new promising concurrency-control mechanism that can avoid many of the pitfalls of the traditional lock-based techniques. TM systems handle data races between threads automatically so ...Transactional memory （TM） is a new promising concurrency-control mechanism that can avoid many of the pitfalls of the traditional lock-based techniques. TM systems handle data races between threads automatically so that programmers do not have to reason about the interaction of threads manually. TM provides a programming model that may make the development of multi-threaded programs easier. Much work has been done to explore the various implementation strategies of TM systems and to achieve better performance, but little has been done on how to formally reason about programs using TM and how to make sure that such reasoning is sound. In this paper, we focus on the semantics of transactional memory and present a proof-carrying code （PCC） system for reasoning about programs using TM . We formalize our reasoning with respect to the TM semantics, prove its soundness, and use examples to demonstrate its effectiveness.展开更多
基金Projects(61003075, 61170261) supported by the National Natural Science Foundation of China
文摘Most transactional memory (TM) research focused on multi-core processors, and others investigated at the clusters, leaving the area of non-uniform memory access (NUMA) system unexplored. The existing TM implementations made significant performance degradation on NUMA system because they ignored the slower remote memory access. To solve this problem, a latency-based conflict detection and a forecasting-based conflict prevention method were proposed. Using these techniques, the NUMA aware TM system was presented. By reducing the remote memory access and the abort rate of transaction, the experiment results show that the NUMA aware strategies present good practical TM performance on NUMA system.
文摘多核结构上采用由用户显式制导的并行程序设计模型,使用锁和同步变量来实现同步。事务存储模型能够解决由锁机制带来的一系列问题,提高程序的并发性。介绍了在文中提出的一种基于事务存储模型的多核结构(Transactional-Memory based Chip Multiple-Superscaler,TMCMS)上的并行编程模型,以及针对循环程序的执行模型;以FFT程序为例具体介绍了循环结构的并行化方法和编译转换过程。在初步的实验中,将处理单元从1增加到16个时,在所设计的编程模型的支持下,IPC(Instruction PerCycle)有接近线性的增长,说明该并行编程模型能够充分发掘程序中潜在的细粒度线程级并行性,同时保持并行程序设计的简单性。
基金Supported by the National Natural Science Foundation of China under Grant Nos.60673126 and 90718026, and Intel China Research Center.
文摘Transactional memory （TM） is a new promising concurrency-control mechanism that can avoid many of the pitfalls of the traditional lock-based techniques. TM systems handle data races between threads automatically so that programmers do not have to reason about the interaction of threads manually. TM provides a programming model that may make the development of multi-threaded programs easier. Much work has been done to explore the various implementation strategies of TM systems and to achieve better performance, but little has been done on how to formally reason about programs using TM and how to make sure that such reasoning is sound. In this paper, we focus on the semantics of transactional memory and present a proof-carrying code （PCC） system for reasoning about programs using TM . We formalize our reasoning with respect to the TM semantics, prove its soundness, and use examples to demonstrate its effectiveness.
