• DocumentCode
    3090994
  • Title

    JVM Virtual Method Invoking Optimization Based on CAM Table

  • Author

    Cai, Songsong ; Yang, Yongqiang ; Lin, Chuanwen ; Liu, Qi

  • fYear
    2011
  • fDate
    28-30 July 2011
  • Firstpage
    122
  • Lastpage
    129
  • Abstract
    In Java programs, it needs to use the information of the method type to resolve the virtual method dynamically, which restricts the performance greatly. Currently, the solution is mainly the technique of inline caching, which can be divided into two categories: monomorphic inline caching and polymorphic inline caching. Because of the simple implementation of monomorphic inline caching, it is more commonly used, but it cannot resolve the problem of frequently seeing different types of objects at one call site. Although polymorphic inline caching can solve this problem, it is costly. This paper presents the CAM hardware table, proposes a virtual method invoking mechanism based on software and hardware co-design. It solves the problem of frequently seeing different types of objects at one call site, and does not introduce additional overhead. The experimental evaluation shows that it improves the cached hit rate from 13.3% to 76.4% and improves 16.2% performance of the virtual test, it improves the performance of SPECjvm 98 by 6.4% on average.
  • Keywords
    Java; cache storage; content-addressable storage; hardware-software codesign; virtual machines; CAM hardware table; JVM virtual method invoking optimization; Java programs; Java virtual machine; content associated memory; monomorphic inline caching; polymorphic inline caching; software-and-hardware co-design; Computer aided manufacturing; Cryptography; Hardware; Indexes; Java; Software; Virtual machining; CAM table; JVM virtual method invocation; inline cache;
  • fLanguage
    English
  • Publisher
    ieee
  • Conference_Titel
    Networking, Architecture and Storage (NAS), 2011 6th IEEE International Conference on
  • Conference_Location
    Dalian, Liaoning
  • Print_ISBN
    978-1-4577-1172-5
  • Electronic_ISBN
    978-0-7695-4509-7
  • Type

    conf

  • DOI
    10.1109/NAS.2011.13
  • Filename
    6005431