ABSTRACT
Programming multicore systems is currently considered very difficult. One reason is that those are mostly constructed from the hardware point of view. Many of the processor core design solutions in contemporary constructions emphasize execution speed of a single thread. Since the memory access delay is the real bottleneck, such techniques often aim at maximizing cache hits by programmer guided locality of memory references and prefetching memory locations, etc.
In this paper, we consider constructing processor core solutions that support easy-to-use programming approach based on the PRAM model. Specifically, we consider a processor core design of a multicore system, where the aim is to amortize the memory access delays by having multiple simultaneous executable software threads per each processor core. The core switches the executed extremely light-weight thread at each step, and thus the core can wait for pending memory requests to complete without any penalty (as long as it has non-blocked threads). Moreover, we consider the core to support moving threads paradigm instead of traditional moving data paradigm. We present an outline of such a processor core architecture, where we change the traditional pipelined execution model of RISC.
- P. Bach, M. Braun, A. Formella, J. Friedrich, T. Grun, C. Lichtenau. "Building the 4 processor SB-PRAM prototype". In Proc. of the 30th Hawaii International Conference on System Sciences: Advanced Technology Track - Vol. 5, 1997. Google ScholarDigital Library
- V. Chaudhary and H. Jiang, Techniques for Migrating Computations on the Grid, In Engineering the Grid: Status and Perspective, Editors: Beniamino Di Martino, Jack Dongarra, Adolfy Hoisie, Hans Zima, and Laurence T. Yang, American Scientific Publishers, January 2006, 399--415.Google Scholar
- M. Forsell. "A Scalable High-Performance Computing Solution for Network on Chips." IEEE Micro 22(5) (September--October 2002), pp. 46--55. Google ScholarDigital Library
- M. Forsell, V. Leppänen. "High-Bandwidth On-Chip Communication Architecture for General Purpose Computing." Proceedings of 9th World Multi-Conference on Systems, Cybernetics and Informatics, WMSCI'2005, pages 1--6, 2005.Google Scholar
- M. Forsell and V. Leppänen. "Moving Threads: A Non-Conventional Approach for Mapping Computation to MP-SOC." In Proceedings of the 2007 International Conference on Parallel and Distributed Processing Techniques and Applications (PDPTA'07), pages 232--238, Jun 2007.Google Scholar
- S. Fortune, J. Wyllie. "Parallelism in Random Access Machines". In Proc. 10th ACM Symposium on Theory of Computing, pp 114--118, 1978. Google ScholarDigital Library
- J. J'aj'a. An Introduction to Parallel Algorithms. Addison Wesley, 1992. Google ScholarDigital Library
- S. Jenks and J-L. Gaudiot, A Multithreaded Runtime System with Thread Migration for Distributed Memory Parallel Computing, In Proceedings of High Performance Computing Symposium, 2003, Advanced Simulation Technologies Conference, Orlando, FL, 2003.Google Scholar
- J. Keller, C. Kessler, and J. Träff. Practical PRAM Programming. Wiley, 2001. Google ScholarDigital Library
- V. Leppänen: Studies on the Realization of PRAM, PhD thesis, University of Turku, Department of Computer Science, TUCS Dissertation 3, November, 1996.Google Scholar
- V. Leppänen. "Balanced PRAM Simulations via Moving Threads and Hashing." Journal of Universal Computer Science, 4:8, 675--689, 1998.Google Scholar
- D. A. Patterson and J. L. Hennessy. Computer Organization and Design; The Hardware/Software Interface, third edition, Morgan Kaufmann, San Francisco, 2005 Google ScholarDigital Library
- K. A. Shaw and W. J. Dally, Migration in Single Chip Multiprocessors, Computer Architecture Letters, Vol. 1, No. 3, Nov. 2002, pp. 2--5. Google ScholarDigital Library
- X. Wen, U. Vishkin. "FPGA-based prototype of a PRAM-On-Chip processor." Computer Frontiers 2008, May 5--7, 2008. Google ScholarDigital Library
Index Terms
- Outline of RISC-based core for multiprocessor on chip architecture supporting moving threads
Recommendations
RISC-based moving threads multicore architecture
CompSysTech '11: Proceedings of the 12th International Conference on Computer Systems and TechnologiesIn this paper, we describe the architectural output of our `Moving threads realization study' (MOTH) project, which is a RISC-based multicore architecture framework. Each fraction of the memory can be accesses only via a certain core, via its cache ...
A moving threads processor architecture MTPA
Moving threads is a new kind of approach for multicore processor architectures. Traditionally, each thread stays in the core where it is created, and data is moved from the main memory via caches to each core and thread. In the moving threads approach, ...
Comments