• Login
    View Item 
    •   MINDS@UW Home
    • MINDS@UW Madison
    • College of Letters and Science, University of Wisconsin–Madison
    • Department of Computer Sciences, UW-Madison
    • CS Technical Reports
    • View Item
    •   MINDS@UW Home
    • MINDS@UW Madison
    • College of Letters and Science, University of Wisconsin–Madison
    • Department of Computer Sciences, UW-Madison
    • CS Technical Reports
    • View Item
    JavaScript is disabled for your browser. Some features of this site may not work without it.

    Region-level Tracking for Scalable Directory Cache

    Thumbnail
    File(s)
    tech report (1.007Mb)
    Date
    2015-04-19
    Author
    Yoon, Hongil
    Sohi, Gurindar S.
    Metadata
    Show full item record
    Abstract
    Traditional coherence directories track sharing information at a cache-line granularity. In practice, however, as data sharing occurs at a coarser granularity in a large region of memory, common sharing patterns tend to be observed across multiple proximate lines. Hence, the directory entries for the lines replicate the same sharing information, resulting in inefficient use of space, power, and energy. In this paper, we empirically demonstrate "region-level sharing pattern locality", that is, a small number of distinct sharing patterns are observed across proximate lines within a large region of memory, e.g., a page unit. We leverage this phenomenon to propose a new representation of sharing information, called Region-level Sharing information Tracking (RST), that dynamically maintains common sharing information in a space-efficient manner at a region-level. Our experimental results based on conventional parallel and server workloads show that RST reduces over 75% of the area (and hence energy) compared to conventional directory caches, with almost negligible performance overhead.
    Subject
    directory caches
    Directory-based coherence protocols
    cache coherence
    region-level sharing pattern locality
    Region-level Sharing information Tracking
    RST
    multicore
    Permanent Link
    http://digital.library.wisc.edu/1793/73603
    Citation
    TR1823
    Part of
    • CS Technical Reports

    Contact Us | Send Feedback
     

     

    Browse

    All of MINDS@UWCommunities & CollectionsBy Issue DateAuthorsTitlesSubjectsThis CollectionBy Issue DateAuthorsTitlesSubjects

    My Account

    LoginRegister

    Contact Us | Send Feedback