COMPILERS: Next Generation Garbage Collection

编译器：下一代垃圾收集

• 批准号: 0429859
• 负责人: Kathryn McKinley
• 金额: --
• 依托单位: University of Texas at Austin
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2004
• 资助国家: 美国
• 起止时间: 2004-12-15 至 2009-11-30
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=0429859&HistoricalAwards=false
• 关键词: COMPILERS; Next; Generation; Garbage; Collection

A key component of their performance is automatic memory management (garbage collection) which frees programmers from explicit memory management, easing the programming burden and eliminating a source of many errors.The growing need for high performance collectors has sparked a resurgence in garbage collection research.Garbage collection provides a number of software engineering benefits such as preventing twocommon programmer errors which are among the most difficult to diagnose and fix: (1) prematurely freeing memory that the program later references, and (2) systematically never freeing which can causememory consumption to grow without bound.Intellectual MeritThis project will explore (1) dynamic optimizations and policy tuning applicable to a variety of collectors;(2) new basic mechanisms for attaining locality and flexible heap organizations, and (3) adaptive algorithmsto match applications to policies.Broader Impact. The research impact from this project will be high performance garbage collectors that play to the strengths of modern architectures and that adapt to the phases of modern applications. High performance will in turn enable higher productivity programming, i.e., programmers will be able to solve new, more difficult, and larger problems. The publicly available research artifacts, MMTk and architecture simulators, will add to the national research infrastructure for exploring memory management algorithms and performance, and the influences on total application performance for modern languages. The educational impact will include training graduate and undergraduate students in this key language system component. Outreach to under-represented groups will include summer research experiences for undergraduate women.

其性能的一个关键组成部分是自动内存管理（垃圾收集），将程序员从显式内存管理中解放出来，减轻编程负担并消除许多错误的根源。对高性能收集器的日益增长的需求引发了垃圾收集研究的复兴。垃圾收集提供了许多软件工程方面的好处，例如防止两种最常见的程序员错误诊断和修复：（1）过早地释放程序以后引用的内存，（2）系统地从不释放，这会导致内存消耗无限制地增长。智力优点这个项目将探索（1）适用于各种收集器的动态优化和策略调整;（2）获得局部性和灵活堆组织的新基本机制，以及（3）使应用程序与策略匹配的自适应算法。该项目的研究影响将是高性能的垃圾收集器，发挥现代体系结构的优势，并适应现代应用程序的阶段。高性能将反过来实现更高的生产力编程，即，程序员将能够解决新的、更困难的和更大的问题。公开的研究工件，MMTk和架构模拟器，将增加国家研究基础设施，探索内存管理算法和性能，以及对现代语言的总体应用程序性能的影响。教育方面的影响将包括在这一关键的语言系统组成部分培训研究生和本科生。对代表性不足群体的外联活动将包括为本科女生提供暑期研究经验。