Abstract
The performance of a NUMA architecture depends on the efficient use of local memory. Therefore, software-level techniques that improve memory locality (in addition to parallelism) are extremely important to extract the best performance from these architectures. The proposed solutions so far include OS-based automatic data migrations and compiler-based static/dynamic data distributions.
This paper proposes and evaluates a hybrid strategy for optimizing memory locality in NUMA architectures. In this strategy, we employ both compiler-directed data distribution and OS-directed dynamic page migration. More specifically, a given program code is first divided into segments, and then each segment is optimized either using compiler-based data distributions (at compile-time) or using dynamic migration (at runtime). In selecting the optimization strategy to use for a program segment, we use a criterion based on the number of compile-time analyzable references in loops.
To test the effectiveness of our strategy in optimizing memory locality of applications, we implemented it and compared its performance with that of several other techniques such as compiler-directed data distribution and OS-directed dynamic page migration. Our experimental results obtained through simulation indicate that our hybrid strategy outperforms other strategies and achieves the best performance for a set of codes with regular, irregular, and mixed (regular + irregular) access patterns.
This is a preview of subscription content, log in via an institution to check access.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Preview
Unable to display preview. Download preview PDF.
References
Amarasinghe, S., Anderson, J., Lam, M., Tseng, C.-W.: The SUIF compiler for scalable parallel machines. In: Proc. the Seventh SIAM Conference on Parallel Processing for Scientific Computing (February 1995)
Anderson, J., Amarasinghe, S., Lam, M.: Data and computation transformations for multiprocessors. In: Proc. the 5th ACM SIGPLAN Symposium on Principles & Practice of Parallel Programming, Santa Barbara, CA, July 1995, pp. 166–178 (1995)
Chandra, R., Chen, D., Cox, R., Maydan, D., Nedeljkovic, N., Anderson, J.: Data-distribution support on distributed-shared memory multi-processors. In: Proc. the ACM SIGPLAN Conference on Programming Language Design and Implementation, Las Vegas, NV (1997)
Chandra, R., Devine, S., Verghese, B., Gupta, A., Rosenblum, M.: Scheduling and page migration for multiprocessor compute servers. In: Proc. ASPLOS V, October 1994, pp. 12–24 (1994)
Kandemir, M., Choudhary, A., Ramanujam, J., Banerjee, P.: Improving locality using loop and data transformations in an integrated framework. In: Proc. the International Symposium on Microarchitecture, Dallas, TX, December 1998, pp. 285–296 (1998)
LaRowe, R., Ellis, C.: Page placement policies for NUMA multiprocessors. Journal of Parallel and Distributed Computing 11(2) (February 1991)
Laudon, J., Lenoski, D.: The SGI Origin: A CC-NUMA highly scalable server. In: Proc. the 24th Annual International Symposium on Computer Architecture (May 1997)
Li, W.: Compiling for NUMA parallel machines. Ph.D. Thesis, Cornell University, Ithaca, New York (1993)
McKinley, K., Carr, S., Tseng, C.W.: Improving data locality with loop transformations. ACM Transactions on Programming Languages & Systems 18(4), 424–453 (1996)
Verghese, B., Devine, S., Gupta, A., Rosenblum, M.: Operating system support for improving data locality on CC-NUMA compute servers. In: Proc. ASPLOS VII, Cambridge, MA (1996)
Wolf, M., Lam, M.: A data locality optimizing algorithm. In: Proc. the ACMSIGPLAN Conference on Programming Language Design and Implementation, June 1991, pp. 30–44 (1991)
Wolf, M., Maydan, D., Chen, D.: Combining loop transformations considering caches and scheduling. In: Proc. the International Symposium on Microarchitecture, Paris, France, December 1996, pp. 274–286 (1996)
Wolfe, M.: High Performance Compilers for Parallel Computing. Addison-Wesley Publishing Company, Reading (1996)
Author information
Authors and Affiliations
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2005 Springer-Verlag Berlin Heidelberg
About this paper
Cite this paper
Kadayif, I., Kandemir, M., Choudhary, A. (2005). A Hybrid Strategy Based on Data Distribution and Migration for Optimizing Memory Locality. In: Pugh, B., Tseng, CW. (eds) Languages and Compilers for Parallel Computing. LCPC 2002. Lecture Notes in Computer Science, vol 2481. Springer, Berlin, Heidelberg. https://doi.org/10.1007/11596110_8
Download citation
DOI: https://doi.org/10.1007/11596110_8
Publisher Name: Springer, Berlin, Heidelberg
Print ISBN: 978-3-540-30781-5
Online ISBN: 978-3-540-31612-1
eBook Packages: Computer ScienceComputer Science (R0)