A flexible and dynamic page migration infrastructure based on hardware counters

详细信息    查看全文

• 作者：Juan A. Lorenzo-Castillo (1)
  Juan C. Pichel (1)
  Francisco F. Rivera (1)
  Tomás F. Pena (1)
  José C. Cabaleiro (1)
  
• 关键词：Hardware counters ; Page migration ; NUMA
• 刊名：The Journal of Supercomputing
• 出版年：2013
• 出版时间：August 2013
• 年：2013
• 卷：65
• 期：2
• 页码：930-948
• 全文大小：1281KB
• 参考文献：1. Bolosky WJ, Scott ML, Fitzgerald RP, Fowler RJ, Cox AL (1991) NUMA policies and their relation to memory architecture. In: Int conf on architectural support for programming languages and operating systems, pp 212-21
  2. Bull JM, Johnson C (2002) Data distribution, migration and replication on a ccNUMA architecture. In: Proceedings of the fourth European workshop on OpenMP
  3. Eranian S (2005) The Perfmon2 interface specification. Technical report HPL-2004-200R1, HP Labs
  4. Galicia supercomputing centre (CESGA): http://www.cesga.es
  5. Goglin B, Furmento N (2009) Enabling high-performance memory migration for multithreaded applications on Linux. In: Proc of the IEEE int symposium on parallel & distributed processing, pp 1-
  6. Hewlett Packard (2006) Dual-core update to the Intel Itanium 2 processor reference manual. Technical paper
  7. Jin H, Jin H, Frumkin M, Frumkin M, Yan J, Yan J (1999) The OpenMP implementation of NAS parallel benchmarks and its performance. Technical report
  8. Larowe RP Jr, Schlatter Ellis C (1991) Experimental comparison of memory management policies for NUMA multiprocessors. ACM Trans Comput Syst 9(4):319-63 CrossRef
  9. Majo Z, Gross TR (2012) Matching memory access patterns and data placement for NUMA systems. In: Proc of the tenth international symposium on code generation and optimization, CGO-2, New York, NY, USA, pp 230-41 CrossRef
  10. Marathe J, Mueller F (2006) Hardware profile-guided automatic page placement for ccNUMA systems. In: Proc of the ACM SIGPLAN symposium on principles and practice of parallel programming, pp 90-9
  11. move_pages manual: http://linux.die.net/man/2/move_pages
  12. Nikolopoulos DS, Papatheodorou TS, Polychronopoulos CD, Labarta J, Ayguadé E (2000) A case for user-level dynamic page migration. In: Proceedings of the int conf on supercomputing, pp 119-30
  13. Nikolopoulos DS, Papatheodorou TS, Polychronopoulos CD, Labarta J, Ayguadé E (2000) User-level dynamic page migration for multiprogrammed shared-memory multiprocessors. In: Proc of the int conf on parallel processing, p 95
  14. Nikolopoulos DS, Polychronopoulos CD, Papatheodorou TS, Labarta J, Ayguadé E (2002) Scheduler-activated dynamic page migration for multiprogrammed DSM multiprocessors. J Parallel Distrib Comput 62(6):1069-103 CrossRef
  15. OpenMP: Simple, portable, scalable SMP programming. http://openmp.org
  16. Perfmon2 monitoring interface and Pfmon monitoring tool: http://perfmon2.sourceforge.net
  17. Tao J, Schulz M, Karl W (2002) Improving data locality using dynamic page migration based on memory access histograms. In: Proc of the international conference on computational science—Part II, pp 933-42
  18. Thakkar V (2008) Dynamic page migration on ccNUMA platforms guided by hardware tracing. Master’s thesis, Graduate Faculty of North Carolina State University
  19. Tikir MM, Hollingsworth JK (2004) Using hardware counters to automatically improve memory performance. In: Proc of the ACM/IEEE conference on supercomputing, SC-4, p 46 CrossRef
  20. Tikir MM, Hollingsworth JK (2008) Hardware monitors for dynamic page migration. J Parallel Distrib Comput 68:1186-200 CrossRef
  21. Wang X, Wen X, Li Y, Luo Y, Li X, Wang Z (2012) A dynamic cache partitioning mechanism under virtualization environment. In: Proc of the 11th international conf on trust, security and privacy in computing and communications (TrustCom), pp 1907-911
  22. Wilson KM, Aglietti BB (2001) Dynamic page placement to improve locality in CC-NUMA multiprocessors for TPC-C. In: Proceedings of the ACM/IEEE conference on supercomputing, pp 98-07
  
• 作者单位：Juan A. Lorenzo-Castillo (1)
  Juan C. Pichel (1)
  Francisco F. Rivera (1)
  Tomás F. Pena (1)
  José C. Cabaleiro (1)
  
  1. Centro de Investigación en Tecnoloxías da Información (CITIUS), University of Santiago de Compostela, Santiago de Compostela, Spain
  

文摘

Performance counters, also known as hardware counters, are a powerful monitoring mechanism included in the Performance Monitoring Unit (PMU) of most of the modern microprocessors. Their use is gaining popularity as an analysis and validation tool for profiling, since their impact is virtually imperceptible and their precision has noticeably increased thanks to the new Precise Event-Based Sampling (PEBS) features. In this paper, we present and evaluate a novel user-level tool, based on hardware counters, for monitoring and migrating pages dynamically. This tool supports different migration strategies, being able to attach and monitor a target application without need to modify it whatsoever. The page migration process is performed timely and its overhead is overcome by the benefit of the data locality achieved. As a case study, an access-based migration algorithm was implemented and integrated into our tool. Performance results on a NUMA system show a noticeable reduction of remote accesses and execution time, achieving speedups of up to ?1?% in a multiprogrammed environment.