Robust Adaptation to Available Parallelism in Transactional Memory Applications

摘要

Applications using transactional memory may exhibit fluc-tuating (dynamic) available parallelism, i.e. the maximum number of transactions that can be committed concurrently may change over time. Executing large numbers of transactions concurrently in phases with low available parallelism will waste processor resources in aborted transac-tions, while executing few transactions concurrently in phases with high available parallelism will degrade execution time by not fully exploit-ing the available parallelism. Three questions come to mind: (1) Are there such transactional applications? (2) How can such behaviour be exploited? and (3) How can available parallelism be measured or calcu-lated efficiently? The contributions of this paper constitute the answers to these questions.rnThis paper presents a system, called transactional concurrency tun-ing, that adapts the number of transactions executing concurrently in response to dynamic available parallelism, in order to improve processor resource usage and execution time performance. Four algorithms, called controller models, that vary in response strength were presented in pre-vious work and shown to maintain execution time similar to the best case non-tuned execution time, but improve resource usage significantly in benchmarks that exhibit dynamic available parallelism.rnThis paper presents an analysis of the four controller models' response characteristics to changes in dynamic available parallelism, and identi-fies weaknesses that reduce their general applicability. These limitations lead to the design of a fifth controller model, called P-only transactional concurrency tuning (PoCC). Evaluation of PoCC shows it improves upon performance and response characteristics of the first four controller mod-els, making it a robust controller model suitable for general use.