A moving threads processor architecture MTPA

Martti Forsell (Corresponding Author), Ville Leppänen

    Research output: Contribution to journalArticleScientificpeer-review

    1 Citation (Scopus)

    Abstract

    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, each core can access only a certain portion of the main memory via its local memory block, and thus extremely lightweight threads are moved between the cores. As a consequence, all kinds of cache coherence problems and need for read reply messages are eliminated. Also Lamport’s sequential consistency of shared memory multiprocessor systems is achieved for free. In this paper, we propose a processor architecture (MTPA) for the moving threads paradigm. We describe the overall structure, operation, instruction set, and thread management mechanism as well as evaluate the proposed architecture with different functional unit settings with simulations and give early silicon area and power consumption estimates.
    Original languageEnglish
    Pages (from-to)5-19
    Number of pages15
    JournalThe Journal of Supercomputing
    Volume57
    Issue number1
    DOIs
    Publication statusPublished - 2011
    MoE publication typeA1 Journal article-refereed

    Fingerprint

    Thread
    Data storage equipment
    Electric power utilization
    Cache Coherence
    Shared-memory multiprocessors
    Multi-core Processor
    Silicon
    Multiprocessor Systems
    Architecture
    Power Consumption
    Paradigm
    Unit
    Evaluate
    Estimate
    Simulation

    Keywords

    • Moving threads
    • Parallel computing
    • MP-SOC
    • Processor architecture
    • Computer architecture

    Cite this

    Forsell, Martti ; Leppänen, Ville. / A moving threads processor architecture MTPA. In: The Journal of Supercomputing. 2011 ; Vol. 57, No. 1. pp. 5-19.
    @article{54aa645f0c084e7b99fe424880c909df,
    title = "A moving threads processor architecture MTPA",
    abstract = "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, each core can access only a certain portion of the main memory via its local memory block, and thus extremely lightweight threads are moved between the cores. As a consequence, all kinds of cache coherence problems and need for read reply messages are eliminated. Also Lamport’s sequential consistency of shared memory multiprocessor systems is achieved for free. In this paper, we propose a processor architecture (MTPA) for the moving threads paradigm. We describe the overall structure, operation, instruction set, and thread management mechanism as well as evaluate the proposed architecture with different functional unit settings with simulations and give early silicon area and power consumption estimates.",
    keywords = "Moving threads, Parallel computing, MP-SOC, Processor architecture, Computer architecture",
    author = "Martti Forsell and Ville Lepp{\"a}nen",
    year = "2011",
    doi = "10.1007/s11227-011-0573-9",
    language = "English",
    volume = "57",
    pages = "5--19",
    journal = "The Journal of Supercomputing",
    issn = "0920-8542",
    publisher = "Springer",
    number = "1",

    }

    A moving threads processor architecture MTPA. / Forsell, Martti (Corresponding Author); Leppänen, Ville.

    In: The Journal of Supercomputing, Vol. 57, No. 1, 2011, p. 5-19.

    Research output: Contribution to journalArticleScientificpeer-review

    TY - JOUR

    T1 - A moving threads processor architecture MTPA

    AU - Forsell, Martti

    AU - Leppänen, Ville

    PY - 2011

    Y1 - 2011

    N2 - 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, each core can access only a certain portion of the main memory via its local memory block, and thus extremely lightweight threads are moved between the cores. As a consequence, all kinds of cache coherence problems and need for read reply messages are eliminated. Also Lamport’s sequential consistency of shared memory multiprocessor systems is achieved for free. In this paper, we propose a processor architecture (MTPA) for the moving threads paradigm. We describe the overall structure, operation, instruction set, and thread management mechanism as well as evaluate the proposed architecture with different functional unit settings with simulations and give early silicon area and power consumption estimates.

    AB - 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, each core can access only a certain portion of the main memory via its local memory block, and thus extremely lightweight threads are moved between the cores. As a consequence, all kinds of cache coherence problems and need for read reply messages are eliminated. Also Lamport’s sequential consistency of shared memory multiprocessor systems is achieved for free. In this paper, we propose a processor architecture (MTPA) for the moving threads paradigm. We describe the overall structure, operation, instruction set, and thread management mechanism as well as evaluate the proposed architecture with different functional unit settings with simulations and give early silicon area and power consumption estimates.

    KW - Moving threads

    KW - Parallel computing

    KW - MP-SOC

    KW - Processor architecture

    KW - Computer architecture

    U2 - 10.1007/s11227-011-0573-9

    DO - 10.1007/s11227-011-0573-9

    M3 - Article

    VL - 57

    SP - 5

    EP - 19

    JO - The Journal of Supercomputing

    JF - The Journal of Supercomputing

    SN - 0920-8542

    IS - 1

    ER -