Parallel and concurrent search for fast AND/OR tree search on multicore processors

Fumiyo Takano, Yoshitaka Maekawa, Hironori Kasahara

    Research output: Chapter in Book/Report/Conference proceedingConference contribution

    1 Citation (Scopus)

    Abstract

    This paper proposes a fast AND/OR tree search algorithm using a multiple paths parallel and concurrent search scheme for embedded multicore processors. Currently, not only PCs or supercomputers but also information appliances such as game consoles, mobile devices and car navigation systems are equipped with multicore processors for better cost performance and lower power consumption. However, the number of processor cores and the amount of memories in embedded multicore processors are limited for lowering power consumption and chip costs. Development of parallel application programs on embedded multicore processors requires exploitation of parallelism and effective utilization of small memories. The proposed algorithm allows us to search in parallel many paths including lowly evaluated nodes and paths including highly evaluated nodes. The algorithm also uses depth-first search, working on small memories. The proposed algorithm is applied for a tsume-shogi (Japanese chess problem) solver as a typical AND/OR tree search problem on an embedded multicore processor system, NEC Electronics NaviEngine with four ARM processor cores. Evaluation results for 100 problems show that the proposed algorithm executed on two processor cores is 2.36 times faster, and executed on four processor cores is 4.17 times faster than the sequential algorithm on the average.

    Original languageEnglish
    Title of host publicationProceedings of the IASTED International Conference on Parallel and Distributed Computing and Networks, PDCN 2009
    Pages139-144
    Number of pages6
    Publication statusPublished - 2009
    EventIASTED International Conference on Parallel and Distributed Computing and Networks, PDCN 2009 - Innsbruck
    Duration: 2009 Feb 162009 Feb 18

    Other

    OtherIASTED International Conference on Parallel and Distributed Computing and Networks, PDCN 2009
    CityInnsbruck
    Period09/2/1609/2/18

    Fingerprint

    Data storage equipment
    Electric power utilization
    ARM processors
    Supercomputers
    Navigation systems
    Application programs
    Mobile devices
    Costs
    Electronic equipment
    Railroad cars

    Keywords

    • Embedded systems
    • Multicore processors
    • Parallel AND/OR tree search algorithms
    • Parallel processing

    ASJC Scopus subject areas

    • Computational Theory and Mathematics
    • Computer Networks and Communications
    • Software

    Cite this

    Takano, F., Maekawa, Y., & Kasahara, H. (2009). Parallel and concurrent search for fast AND/OR tree search on multicore processors. In Proceedings of the IASTED International Conference on Parallel and Distributed Computing and Networks, PDCN 2009 (pp. 139-144)

    Parallel and concurrent search for fast AND/OR tree search on multicore processors. / Takano, Fumiyo; Maekawa, Yoshitaka; Kasahara, Hironori.

    Proceedings of the IASTED International Conference on Parallel and Distributed Computing and Networks, PDCN 2009. 2009. p. 139-144.

    Research output: Chapter in Book/Report/Conference proceedingConference contribution

    Takano, F, Maekawa, Y & Kasahara, H 2009, Parallel and concurrent search for fast AND/OR tree search on multicore processors. in Proceedings of the IASTED International Conference on Parallel and Distributed Computing and Networks, PDCN 2009. pp. 139-144, IASTED International Conference on Parallel and Distributed Computing and Networks, PDCN 2009, Innsbruck, 09/2/16.
    Takano F, Maekawa Y, Kasahara H. Parallel and concurrent search for fast AND/OR tree search on multicore processors. In Proceedings of the IASTED International Conference on Parallel and Distributed Computing and Networks, PDCN 2009. 2009. p. 139-144
    Takano, Fumiyo ; Maekawa, Yoshitaka ; Kasahara, Hironori. / Parallel and concurrent search for fast AND/OR tree search on multicore processors. Proceedings of the IASTED International Conference on Parallel and Distributed Computing and Networks, PDCN 2009. 2009. pp. 139-144
    @inproceedings{5e557c40e3ca48a9bdce4f65c6deac1c,
    title = "Parallel and concurrent search for fast AND/OR tree search on multicore processors",
    abstract = "This paper proposes a fast AND/OR tree search algorithm using a multiple paths parallel and concurrent search scheme for embedded multicore processors. Currently, not only PCs or supercomputers but also information appliances such as game consoles, mobile devices and car navigation systems are equipped with multicore processors for better cost performance and lower power consumption. However, the number of processor cores and the amount of memories in embedded multicore processors are limited for lowering power consumption and chip costs. Development of parallel application programs on embedded multicore processors requires exploitation of parallelism and effective utilization of small memories. The proposed algorithm allows us to search in parallel many paths including lowly evaluated nodes and paths including highly evaluated nodes. The algorithm also uses depth-first search, working on small memories. The proposed algorithm is applied for a tsume-shogi (Japanese chess problem) solver as a typical AND/OR tree search problem on an embedded multicore processor system, NEC Electronics NaviEngine with four ARM processor cores. Evaluation results for 100 problems show that the proposed algorithm executed on two processor cores is 2.36 times faster, and executed on four processor cores is 4.17 times faster than the sequential algorithm on the average.",
    keywords = "Embedded systems, Multicore processors, Parallel AND/OR tree search algorithms, Parallel processing",
    author = "Fumiyo Takano and Yoshitaka Maekawa and Hironori Kasahara",
    year = "2009",
    language = "English",
    isbn = "9780889867840",
    pages = "139--144",
    booktitle = "Proceedings of the IASTED International Conference on Parallel and Distributed Computing and Networks, PDCN 2009",

    }

    TY - GEN

    T1 - Parallel and concurrent search for fast AND/OR tree search on multicore processors

    AU - Takano, Fumiyo

    AU - Maekawa, Yoshitaka

    AU - Kasahara, Hironori

    PY - 2009

    Y1 - 2009

    N2 - This paper proposes a fast AND/OR tree search algorithm using a multiple paths parallel and concurrent search scheme for embedded multicore processors. Currently, not only PCs or supercomputers but also information appliances such as game consoles, mobile devices and car navigation systems are equipped with multicore processors for better cost performance and lower power consumption. However, the number of processor cores and the amount of memories in embedded multicore processors are limited for lowering power consumption and chip costs. Development of parallel application programs on embedded multicore processors requires exploitation of parallelism and effective utilization of small memories. The proposed algorithm allows us to search in parallel many paths including lowly evaluated nodes and paths including highly evaluated nodes. The algorithm also uses depth-first search, working on small memories. The proposed algorithm is applied for a tsume-shogi (Japanese chess problem) solver as a typical AND/OR tree search problem on an embedded multicore processor system, NEC Electronics NaviEngine with four ARM processor cores. Evaluation results for 100 problems show that the proposed algorithm executed on two processor cores is 2.36 times faster, and executed on four processor cores is 4.17 times faster than the sequential algorithm on the average.

    AB - This paper proposes a fast AND/OR tree search algorithm using a multiple paths parallel and concurrent search scheme for embedded multicore processors. Currently, not only PCs or supercomputers but also information appliances such as game consoles, mobile devices and car navigation systems are equipped with multicore processors for better cost performance and lower power consumption. However, the number of processor cores and the amount of memories in embedded multicore processors are limited for lowering power consumption and chip costs. Development of parallel application programs on embedded multicore processors requires exploitation of parallelism and effective utilization of small memories. The proposed algorithm allows us to search in parallel many paths including lowly evaluated nodes and paths including highly evaluated nodes. The algorithm also uses depth-first search, working on small memories. The proposed algorithm is applied for a tsume-shogi (Japanese chess problem) solver as a typical AND/OR tree search problem on an embedded multicore processor system, NEC Electronics NaviEngine with four ARM processor cores. Evaluation results for 100 problems show that the proposed algorithm executed on two processor cores is 2.36 times faster, and executed on four processor cores is 4.17 times faster than the sequential algorithm on the average.

    KW - Embedded systems

    KW - Multicore processors

    KW - Parallel AND/OR tree search algorithms

    KW - Parallel processing

    UR - http://www.scopus.com/inward/record.url?scp=74549176954&partnerID=8YFLogxK

    UR - http://www.scopus.com/inward/citedby.url?scp=74549176954&partnerID=8YFLogxK

    M3 - Conference contribution

    AN - SCOPUS:74549176954

    SN - 9780889867840

    SP - 139

    EP - 144

    BT - Proceedings of the IASTED International Conference on Parallel and Distributed Computing and Networks, PDCN 2009

    ER -