Aware conflict detection of non-uniform memory access system and prevention for transactional memory

Abstract

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.

Keywords

transactional memory / non-uniform memory access (NUMA) / conflict detection / conflict prevention

References

[1] MudgeT.. Multicore architectures [C]. Proceedings of the 2007 International Conference on Compilers, Architecture, and Synthesis for Embedded Systems, 2007Salzburg, AustriaACM208

[2] KongetiraP., AingaranK., OlukotunK.. Niagara: A 32-way multithreaded SPARC processor [J]. IEEE MICRO, 2005, 25(2): 21-29

[3] HerlihyM.. The art of multiprocessor programming [C]. Proceedings of the Twenty-Fifth Annual ACM Symposium on Principles of Distributed Computing, 2006Denver, CO, USAACM1-2

[4] SmaragdakisY., KayA., BehrendsR., YoungM.. General and efficient locking without blocking [C]. Proceedings of the 2008 ACM SIGPLAN workshop on Memory Systems Performance and Correctness, 2008Seattle, WA, USAACM1-5

[5] HerlihyM., MossJ.. Transactional memory: Architectural support for lock-free data structures [C]. Proceedings of the 20th Annual International Symposium on Computer Architecture, 1993San Diego, CAIEEE Computer Society Press289-300

[6] ShavitN., TouitouD.. Software transactional memory [C]. Proceedings of the 14th ACM Symposium on Principles of Distributed Computing, 1995Ottawa, Ontario, CanadaACM204-213

[7] HarrisT., CristalA., UnsalO., AyguadeE., GagliardiF., SmithB., ValeroM.. Transactional memory: An overview [J]. IEEE MICRO, 2007, 27(3): 8-29

[8] McdonaldA., CarlstromB., ChungJ., MinhC., ChafiH., KozyrakisC., OlukotunK.. Transactional memory: The hardware-software interface [J]. IEEE MICRO, 2007, 27(1): 67-76

[9] HerlihyM.. The transactional manifesto: Software engineering and non-blocking synchronization [C]. Proceedings of the 2005 ACM SIGPLAN Conference on Programming Language Design and Implementation, 2005Chicago, IL, USAACM280

[10] WangR.-b., LuK., LuX.-cheng.. Investigating transactional memory performance on ccNUMA machines [C]. Proceedings of the 18th ACM International Symposium on High Performance Distributed Computing, 2009Garching, GermanyACM67-68

[11] ManassievK., MihailescuM., AmzaC.. Exploiting distributed version concurrency in a transactional memory cluster [C]. Proceedings of the 11th ACM SIGPLAN Symposium on Principles and Practice of Parallel Programming, 2006New York, USAACM198-208

[12] AmzaC., CoxA., ZwaenepoelW.. Conflict-aware scheduling for dynamic content applications [C]. 4th USENIX Symposium on Internet Technologies and Systems, 2003Seattle, Washington, USAUSENIX71-85

[13] BocchinoR., AdveV., ChamberlainB.. Software transactional memory for large scale clusters [C]. Proceedings of the 13th ACM SIGPLAN Symposium on Principles and Practice of Parallel Programming, 2008Salt Lake City, UT, USAACM247-258

[14] KotselidisC., AnsariM., JarvisK., LujanM., KirkhamC., WatsonI.. DiSTM: A software transactional memory framework for clusters [C]. Proceedings of the 37th International Conference on Parallel Processing, 2008Portland, Oregon, USAIEEE Computer Society51-58

[15] BonacheaD.GASNet specification, v1.1 [R], 2002BerkeleyCS Division, EECS Department, University of California

[16] BaduelL., BaudeF., CaromelD., ContesA., HuetF., MorelM., QuiliciR.Grid computing: Software environments and tools [M], 20061st EditionLondonSpringer205-229

[17] MaratheV., SpearM., HeriotC., AcharyaA., EisenstatD., SchererW., ScottM.Lowering the overhead of nonblocking software transactional memory [R], 2006RochesterDepartment of Computer Science, University of Rochester

[18] FelberP., FetzerC., RiegelT.. Dynamic performance tuning of word-based software transactional memory [C]. Proceedings of the 13th ACM SIGPLAN Symposium on Principles and Practice of Parallel Programming, 2008Salt Lake City, UT, USAACM237-246

[19] DiceD., ShalevO., ShavitN.. Transactional locking II [C]. Proceedings of the 20th International Symposium on Distributed Computing, 2006Stockholm, SwedenSpringer-Verlag194-208

[20] GuerraouiR., KapalkaM., VitekJ.. STMBench7: A benchmark for software transactional memory [C]. Proceedings of the Second European Systems Conference, 2007Lisbon, PortugalACM315-324

[21] HarrisT., FraserK.. Language support for lightweight transactions [C]. Proceedings of the 2003 ACM SIGPLAN Conference on Object-Oriented Programming Systems, Languages and Applications, 2003Anaheim, CA, USAACM388-402

[22] HammondL., CarlstromB., WongV., ChenM., KozyrakisC., OlukotunK.. Transactional coherence and consistency: Simplifying parallel hardware and software [J]. IEEE Micro, 2004, 24(6): 92-103