Efficient Hybrid Transactional Memory Scheme using Near-optimal Retry Computation and Sophisticated Memory Management in Multi-core Environment
1
조회
0
다운로드
1
인용
0
참고문헌
키워드
초록
Recently, hybrid transactional memory (HyTM) has gained much interest from researchers because itcombines the advantages of hardware transactional memory (HTM) and software transactional memory(STM). To provide the concurrency control of transactions, the existing HyTM-based studies use a bloomfilter. However, they fail to overcome the typical false positive errors of a bloom filter. Though the existingstudies use a global lock, the efficiency of global lock-based memory allocation is significantly low in multicoreenvironment. In this paper, we propose an efficient hybrid transactional memory scheme using nearoptimalretry computation and sophisticated memory management in order to efficiently process transactionsin multi-core environment. First, we propose a near-optimal retry computation algorithm that provides anefficient HTM configuration using machine learning algorithms, according to the characteristic of a givenworkload. Second, we provide an efficient concurrency control for transactions in different environments byusing a sophisticated bloom filter. Third, we propose a memory management scheme being optimized for theCPU cache line, in order to provide a fast transaction processing. Finally, it is shown from our performanceevaluation that our HyTM scheme achieves up to 2.5 times better performance by using the Stanfordtransactional applications for multi-processing (STAMP) benchmarks than the state-of-the-art algorithms.
Abstract
Recently, hybrid transactional memory (HyTM) has gained much interest from researchers because itcombines the advantages of hardware transactional memory (HTM) and software transactional memory(STM). To provide the concurrency control of transactions, the existing HyTM-based studies use a bloomfilter. However, they fail to overcome the typical false positive errors of a bloom filter. Though the existingstudies use a global lock, the efficiency of global lock-based memory allocation is significantly low in multicoreenvironment. In this paper, we propose an efficient hybrid transactional memory scheme using nearoptimalretry computation and sophisticated memory management in order to efficiently process transactionsin multi-core environment. First, we propose a near-optimal retry computation algorithm that provides anefficient HTM configuration using machine learning algorithms, according to the characteristic of a givenworkload. Second, we provide an efficient concurrency control for transactions in different environments byusing a sophisticated bloom filter. Third, we propose a memory management scheme being optimized for theCPU cache line, in order to provide a fast transaction processing. Finally, it is shown from our performanceevaluation that our HyTM scheme achieves up to 2.5 times better performance by using the Stanfordtransactional applications for multi-processing (STAMP) benchmarks than the state-of-the-art algorithms.
참고문헌 (11)
- M. Herlihy, "Transactional memory: architectural support for lock-free data structures", Proceedings of the 20th International Symposium on Computer Architecture, pp.289~300 (1993)
- V. Pankratius, "A study of transactional memory vs. locks in practice", Proceedings of the 23rd Symposium on Parallelism in Algorithms and Architectures, pp.43~52 (2011)
- L. Dalessandro, "NOrec: streamlining STM by abolishing ownership records", Proceedings of the 15th ACM SIGPLAN Symposium on Principles and Practice of Parallel Programming, pp.67~78 (2010)
- N. Diegues, "Self-tuning intel transactional synchronization extensions", Proceedings of the 11th International Conference on Autonomic Computing, pp.209~219 (2014)
- J. Vermorel, "Multi-armed bandit algorithms and empirical evaluation", European Conference on Machine Learning, pp.437~448 (2005)
- L. C. Baird, "Gradient descent for general reinforcement learning", Advances in Neural Information Processing Systems, 11, pp.968~974 (1999)
- L. Dalessandro, "Hybrid norec: a case study in the effectiveness of best effort hardware transactional memory", ACM SIGARCH Computer Architecture News, 39(1), pp.39~52 (2011)[DOI]
- R. N. Zare, "Angular Momentum: Understanding Spatial Aspects in Chemistry and Physics" (2011), Wiley
- H. Yun, "PALLOC: DRAM bank-aware memory allocator for performance isolation on multicore platforms", Proceedings of the 2014 IEEE 20th Real-Time and Embedded Technology and Applications Symposium, pp.155~166 (2014)
- "TCMalloc: thread-caching Malloc"[URL]
- C. C. Minh, "STAMP: Stanford transactional applications for multi-processing", Proceedings of IEEE International Symposium on Workload Characterization, pp.35~46 (2008)