Perspective: A Sensible Approach to Speculative Automatic Parallelization
Perspective: A Sensible Approach to Speculative Automatic Parallelization
复制标题
观点：推测自动并行化的明智方法
DOI：
10.1145/3373376.3378458
复制
发表时间：
2020
期刊：
影响因子：
--
通讯作者：
August, David I.
中科院分区：
文献类型：
--
作者：
Apostolakis, Sotiris;Xu, Ziyang;Chan, Greg;Campanoni, Simone;August, David I.
自动并行化（将程序员从容易出错且耗时的高效使用并行处理资源的过程中解放出来）的承诺仍未实现。几十年来，内存分析的不精确性限制了非推测性自动并行化的适用性。推测性自动并行化的引入克服了这些适用性限制，但是，即使在没有错误推测的情况下，这些推测性技术在验证和提交方面也表现出较高的通信和记账成本。本文提出了Perspective，这是一个投机- doall并行化框架，它在保持投机技术的适用性的同时接近非投机技术的效率。与当前应用推测技术来克服内存分析的不精确性的方法不同，Perspective结合了一种新颖的推测感知内存分析器、新的高效推测私有化方法和一个规划阶段，以选择一组最小成本的并行化转换。通过以先前的并行化系统无法实现的方式减少推测性并行化开销，Perspective获得了比Privateer更高的总体程序加速（在28核共享内存商用机器上运行12个通用C/ c++程序的23.0倍），Privateer是先前具有最高适用性的自动DOALL并行化系统（11.5倍）。
The promise of automatic parallelization, freeing programmers from the error-prone and time-consuming process of making efficient use of parallel processing resources, remains unrealized. For decades, the imprecision of memory analysis limited the applicability of non-speculative automatic parallelization. The introduction of speculative automatic parallelization overcame these applicability limitations, but, even in the case of no misspeculation, these speculative techniques exhibit high communication and bookkeeping costs for validation and commit. This paper presents Perspective, a speculative-DOALL parallelization framework that maintains the applicability of speculative techniques while approaching the efficiency of non-speculative ones. Unlike current approaches which subsequently apply speculative techniques to overcome the imprecision of memory analysis, Perspective combines a novel speculation-aware memory analyzer, new efficient speculative privatization methods, and a planning phase to select a minimal-cost set of parallelization-enabling transforms. By reducing speculative parallelization overheads in ways not possible with prior parallelization systems, Perspective obtains higher overall program speedup (23.0x for 12 general-purpose C/C++ programs running on a 28-core shared-memory commodity machine) than Privateer (11.5x), the prior automatic DOALL parallelization system with the highest applicability.
登录
查看更多内容
集群的自动推测 DOALL
DOI：
--
发表时间：
2012
期刊：
IEEE/ACM International Symposium on Code Generation and Optimization
影响因子：
--
作者：
Hanjun Kim;Nick P. Johnson;Jae W. Lee;S. Mahlke;David I. August
通讯作者：
David I. August
DOI：
10.1109/cgo.2017.7863736
发表时间：
2017
期刊：
2017 IEEE/ACM International Symposium on Code Generation and Optimization (CGO)
影响因子：
--
作者：
Nick P. Johnson;Jordan Fix;S. Beard;Taewook Oh;T. Jablin;David I. August
通讯作者：
David I. August
DOI：
10.1145/3173162.3173181
发表时间：
2018
期刊：
Proceedings of the Twenty-Third International Conference on Architectural Support for Programming Languages and Operating Systems
影响因子：
--
作者：
E. A. Deiana;Vincent St;P. Dinda;Nikolaos Hardavellas;Simone Campanoni
通讯作者：
Simone Campanoni
乐观混合分析：通过预测静态分析加速动态分析
DOI：
10.1145/3173162.3177153
发表时间：
2018
期刊：
Architectural Support for Programming Languages and Operating Systems
影响因子：
--
作者：
Devecsery, David;Chen, Peter M.;Flinn, Jason;Narayanasamy, Satish
通讯作者：
Narayanasamy, Satish
DOI：
--
发表时间：
2018
期刊：
International Conference on Architectural Support for Programming Languages and Operating Systems
影响因子：
--
作者：
Jordan Fix;N. P. Nagendra;Sotiris Apostolakis;Hansen Zhang;Sophie Qiu;David I. August
通讯作者：
David I. August