Abstract
There are two competing models for the on-chip memory in Chip Multiprocessor (CMP) systems: hardware-managed coherent caches and software-managed streaming memory. This paper performs a direct comparison of the two models under the same set of assumptions about technology, area, and computational capabilities. The goal is to quantify how and when they differ in terms of performance, energy consumption, bandwidth requirements, and latency tolerance for general-purpose CMPs. We demonstrate that for data-parallel applications on systems with up to 16 cores, the cache-based and streaming models perform and scale equally well. For certain applications with little data reuse, streaming scales better due to better bandwidth use and macroscopic software prefetching. However, the introduction of techniques such as hardware prefetching and nonallocating stores to the cache-based model eliminates the streaming advantage. Overall, our results indicate that there is not sufficient advantage in building streaming memory systems where all on-chip memory structures are explicitly managed. On the other hand, we show that streaming at the programming model level is particularly beneficial, even with the cache-based model, as it enhances locality and creates opportunities for bandwidth optimizations. Moreover, we observe that stream programming is actually easier with the cache-based model because the hardware guarantees correct, best-effort execution even when the programmer cannot fully regularize an application's code.
- Adve, S. V. and Gharachorloo, K. 1996. Shared memory consistency models: A tutorial. IEEE Computer 29, 12 (Dec.), 66--76. Google ScholarDigital Library
- Agarwal, V., Hrishikesh, M. S., Keckler, S. W., and Burger, D. 2000. Clock rate versus IPC: the end of the road for conventional microarchitectures. In Proceedings of the 27th International Symposium Computer Architecture. Google ScholarDigital Library
- Ahn, J. et al. 2004. Evaluating the imagine stream architecture. In Proceedings of the 31st International Symposium Computer Architecture. Google ScholarDigital Library
- Andrews, J. and Backer, N. 2005. Xbox360 system architecture. In Conference Record of Hot Chips 17. Stanford, CA.Google Scholar
- Barroso, L. A. et al. 2000. Piranha: A scalable architecture based on single-chip multiprocessing. In Proceedings of the 27th International Symposium on Computer Architecture. Google ScholarDigital Library
- Chen, Y.-K., Li, E. Q., Zhou, X., and Ge, S. 2006. Implementation of h.264 encoder and decoder on personal computers. J. Visual Communication and Image Representation 17, 2, 509--532.Google ScholarCross Ref
- Chiueh, T. 1993. A generational algorithm to multiprocessor cache coherence. In International Conference on Parallel Processing. 20--24. Google ScholarDigital Library
- Culler, D., Singh, J. P., and Gupta, A. 1999. Parallel Computer Architecture: A Hardware/Software Approach. st. Louis: Morgan Kauffman. Google ScholarDigital Library
- Dally, W. et al. 2003. Merrimac: Supercomputing with Streams. In Proceedings of the 2003 Conference on Supercomputing. Google ScholarDigital Library
- Davis, J. D., Laudon, J., and Olukotun, K. 2005. Maximizing CMP throughput with mediocre cores. In Proceedings of the 14th International Conference on Parallel Architectures and Compilation Techniques. Google ScholarDigital Library
- Drake, M., Hoffmann, H., Rabbah, R., and Amarasinghe, S. 2006. Mpeg-2 decoding in a stream programming language. In Proceedings of the 20th IEEE International Parallel & Distributed Processing Symposium, Rhodes Island (IPDPS). Google ScholarDigital Library
- Eatherton, W. 2005. The push of network processing to the top of the pyramid. Keynote presentation at the Symposium on Architectures for Networking and Communication Systems, Princeton, NJ.Google Scholar
- Erez, M., Ahn, J. H., Gummaraju, J., Rosenblum, M., and Dally, W. J. 2007. Executing irregular scientific applications on stream architectures. In Proceedings of the 21st Annual International Conference on Supercomputing. 93--104. Google ScholarDigital Library
- Fatahalian, K., Knight, T. J., Houston, M. et al. 2006. Sequoia: Programming the memory hierarchy. In Proceedings of the 2006 ACM/IEEE Conference on Supercomputing. Google ScholarDigital Library
- Foley, T. and Sugerman, J. 2005. KD-tree acceleration structures for a GPU raytracer. In Proceedings of the Graphics Hardware Conference Google ScholarDigital Library
- Gordon, M. I. et al. 2002. A stream compiler for communication-exposed architectures. In Proceedings of the 10th International Conference on Architectural Support for Programming Languages and Operating Systems. Google ScholarDigital Library
- Gschwind, M. et al. 2005. A novel SIMD architecture for the cell heterogeneous chip-multiprocessor. In Conference Record of Hot Chips 17.Google Scholar
- Gummaraju, J., Coburn, J., Turner, Y., and Rosenblum, M. 2008. Streamware: programming general-purpose multicore processors using streams. In Proceedings of the 13th International Conference on Architectural Support for Programming Languages and Operating Systems. 297--307. Google ScholarDigital Library
- Gummaraju, J., Erez, M., Coburn, J., Rosenblum, M., and Dally, W. J. 2007. Architectural support for the stream execution model on general-purpose processors. In Proceedings of the 16th International Conference on Parallel Architecture and Compilation Techniques. 3--12. Google ScholarDigital Library
- Gummaraju, J. and Rosenblum, M. 2005. Stream programming on general-purpose processors. In Proceedings of the 38th International Symposium on Microarchitecture. Google ScholarDigital Library
- Havran, V. 2002. Heuristic ray shooting algorithms. Ph.D. thesis, Czech Technical University in Prague.Google Scholar
- Heinlein, J., Gharachorloo, K., Dresser, S., and Gupta, A. 1994. Integration of message passing and shared memory in the stanford flash multiprocessor. SIGOPS Oper. Syst. Rev. 28, 5, 38--50.Google ScholarDigital Library
- Ho, R., Mai, K., and Horowitz, M. 2001. The Future of wires. Proceedings of the IEEE 89, 4 (Apr.).Google ScholarCross Ref
- Ho, R., Mai, K., and Horowitz, M. 2003. Efficient on-chip global interconnects. In Symposium on VLSI Circuits. 271--274.Google Scholar
- Horowitz, M. and Dally, W. 2004. How scaling will change processor architecture. In Proceedings of the International Solid-State Circuits Conference. 132--133.Google Scholar
- Independent JPEG Group. 1998. IJG's JPEG Software Release 6b.Google Scholar
- ITU-T Rec. H.264. 2002. ISO/IEC 144496-10 AVC. 2002.Google Scholar
- Jani, D., Ezer, G., and Kim, J. 2004. Long words and wide ports: Reinventing the Configurable Processor. In Proceedings of the Conference Record of Hot Chips 16. Stanford, CA.Google Scholar
- Jayasena, N. 2005. Memory hierarchy design for steram computing. Ph.D. thesis, Stanford University. Google ScholarDigital Library
- Khailany, B., Williams, T., Lin, J., Long, E., Rygh, M., Tovey, D., and Dally, W. 2008. A programmable 512 gops stream processor for signal, image, and video processing. IEEE Journal of Solid-State Circuits 43, 1, 202--213.Google ScholarCross Ref
- Klaiber, A. C. and Levy, H. M. 1994. A comparison of message passing and shared memory architectures for data parallel programs. In Proceedings of the 21th International Symposium on Computer Architecture. Google ScholarDigital Library
- Kongetira, P. 2004. A 32-way Multithreaded sparc processor. In Proceedings of the Conference Record of Hot Chips.Google Scholar
- Kranz, D., Johnson, K., Agarwal, A., Kubiatowicz, J., and Lim, B.-H. 1993. Integrating message-passing and shared-memory: early experience. In Proceedings of the 4th ACM SIGPLAN symposium on Principles and Practice of Parallel Programming. 54--63. Google ScholarDigital Library
- Kumar, R., Zyuban, V., and Tullsen, D. M. 2005. Interconnections in multi-core architectures: Understanding mechanisms, overheads and scaling. In Proceedings of the 32nd International Symposium on Computer Architecture. Google ScholarDigital Library
- Leverich, J., Arakida, H., Solomatnikov, A., Firoozshahian, A., Horowitz, M., and Kozyrakis, C. 2007. Comparing memory systems for chip multiprocessors. In Proceedings of the 34th Annual International Symposium on Computer Architecture. 358--368. Google ScholarDigital Library
- Lewis, B. and Berg, D. J. 1998. Multithreaded Programming with Pthreads. Upper saddle River. NJ: Prentice Hall. Google ScholarDigital Library
- Li, M. et al. 2005. ALP: efficient support for all levels of parallelism for complex Media applications. Tech. Rep. UIUCDCS-R-2005-2605, UIUC CS. July.Google Scholar
- Lim, A. W., Liao, S.-W., and Lam, M. S. 2001. Blocking and array contraction across arbitrarily nested loops using affine partitioning. ACM SIG-PLAN Notices 36, 7, 103--112. Google ScholarDigital Library
- Lin, Y. 2004. A programmable Vector coprocessor architecture for wireless applications. In Proceedings of the 3rd Workshop on Application Specific Processors.Google Scholar
- Loghi, M. and Pncino, M. 2005. Exploring energy/performance tradeoffs in shared memory MPSoCs: Snoop-based cache coherence vs. software solutions. In Proceedings of the Design Automation and Test in Europe Conference Google ScholarDigital Library
- Machnicki, E. 2005. Ultra high performance scalable DSP family for multimedia. In Proceedings of the Conference Record of Hot Chips 17.Google ScholarCross Ref
- Mai, K. et al. 2000. Smart memories: A modular reconfigurable architecture. In Proceedings of the 27th International Symposium on Computer architecture. Google ScholarDigital Library
- MIPS32 2001. MIPS32 Architecture For Programmers Volume II: The MIPS32 Instruction Set. MIPS Technologies, Inc.Google Scholar
- Moshovos, A. 2005. Regionscout: Exploiting coarse grain sharing in snoop-based coherence. In Proceedings of the 32nd International Symposium on Computer Architecture. Google ScholarDigital Library
- MPEG Software Simulation Group. Mssg mpeg2 encoder and decoder. Available at: http://www.mpeg.org/MPEG/MSSG/.Google Scholar
- Sankaralingam, K. 2004. TRIPS: A polymorphous architecture for exploiting ILP, TLP, and DLP. ACM Trans. Archit. Code Optim. 1, 1, 62--93. Google ScholarDigital Library
- Suh, J. et al. 2003. A performance analysis of PIM, stream processing, and tiled processing on memory-intensive signal processing kernels. In Proceedings of the 30th International Symposium on Computer Architecture. Google ScholarDigital Library
- Tarjan, D., Thoziyoor, S., and Jouppi, N. P. 2006. CACTI 4.0. Tech. Rep. HPL-2006-86, HP Labs.Google Scholar
- Taylor, M. et al. 2004. Evaluation of the raw microprocessor: An exposed-wire-delay architecture for ILP and streams. In Proceedings of the 31st International Symposium on Computer Architecture. Google ScholarDigital Library
- Tensilica 2007. Tensilica Software Tools. http://www.tensilica.com/products/software.htm.Google Scholar
- VanderWiel, S. P. and Lilja, D. J. 2000. Data prefetch mechanisms. ACM Computing Surveys 32, 2, 174--199. Google ScholarDigital Library
- Wang, D. et al. 2005. DRAMsim: A memory-system simulator. SIGARCH Computer Architecture News 33, 4. Google ScholarDigital Library
- Wang, Z. et al. 2002. Using the compiler to improve cache replacement decisions. In Proceedings of the Conference on Parallel Architectures and Compilation Techniques. Google ScholarDigital Library
- Wang, Z. et al. 2003. Guided region prefetching: a cooperative hardware/software approach. In Proceedings of the 30th International Symposium on Computer Architecture. Google ScholarDigital Library
- Yeh, T.-Y. 2005. The low-power high-performance architecture of the PWRficient processor family. In Proceedings of the Conference Record of Hot Chips 17.Google Scholar
Index Terms
- Comparative evaluation of memory models for chip multiprocessors
Recommendations
Comparing memory systems for chip multiprocessors
ISCA '07: Proceedings of the 34th annual international symposium on Computer architectureThere are two basic models for the on-chip memory in CMP systems:hardware-managed coherent caches and software-managed streaming memory. This paper performs a direct comparison of the two modelsunder the same set of assumptions about technology, area, ...
Comparing memory systems for chip multiprocessors
There are two basic models for the on-chip memory in CMP systems:hardware-managed coherent caches and software-managed streaming memory. This paper performs a direct comparison of the two modelsunder the same set of assumptions about technology, area, ...
Improving support for locality and fine-grain sharing in chip multiprocessors
PACT '08: Proceedings of the 17th international conference on Parallel architectures and compilation techniquesBoth commercial and scientific workloads benefit from concurrency and exhibit data sharing across threads/processes. The resulting sharing patterns are often fine-grain, with the modified cache lines still residing in the writer's primary cache when ...
Comments