Memory allocation and higher-order functions

O. Danvy
{"title":"Memory allocation and higher-order functions","authors":"O. Danvy","doi":"10.1145/29650.29676","DOIUrl":null,"url":null,"abstract":"This paper presents a constant-time marking-collecting algorithm to efficiently implement recursion with a general heap memory rather than with a vectorial stack, in a context of frequent captures of continuations. It has been seen to reduce the 80% garbage collection overhead to less than 5% on average.The algorithm has been built into a virtual machine to efficiently implement at the assembly level the Actor language PLASMA, an Actor-oriented version of PROLOG and a variant of SCHEME, currently in use on 8086, 68000 and Vax.The rationale to use the heap memory is that continuations are available via a single pointer in a unified memory and can be shared optimally when recurrently captured, which is simply impossible using a strategy based on stack recopy. Further, non-captured continuations can be incrementally garbage collected on the fly.Part I describes the elementary recursive instructions of the virtual machine. Part II presents and proves the marking-collecting strategy. Part III safely generalizes the transformation \"call + return = branch\" in a way compatible with the possible capture of the current continuation. An appendix relates its integration in the Virtual Scheme Machine supporting Scheme 84.","PeriodicalId":414056,"journal":{"name":"SIGPLAN Conferences and Workshops","volume":"212 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"1987-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"20","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"SIGPLAN Conferences and Workshops","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1145/29650.29676","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 20

Abstract

This paper presents a constant-time marking-collecting algorithm to efficiently implement recursion with a general heap memory rather than with a vectorial stack, in a context of frequent captures of continuations. It has been seen to reduce the 80% garbage collection overhead to less than 5% on average.The algorithm has been built into a virtual machine to efficiently implement at the assembly level the Actor language PLASMA, an Actor-oriented version of PROLOG and a variant of SCHEME, currently in use on 8086, 68000 and Vax.The rationale to use the heap memory is that continuations are available via a single pointer in a unified memory and can be shared optimally when recurrently captured, which is simply impossible using a strategy based on stack recopy. Further, non-captured continuations can be incrementally garbage collected on the fly.Part I describes the elementary recursive instructions of the virtual machine. Part II presents and proves the marking-collecting strategy. Part III safely generalizes the transformation "call + return = branch" in a way compatible with the possible capture of the current continuation. An appendix relates its integration in the Virtual Scheme Machine supporting Scheme 84.
内存分配和高阶函数
本文提出了一种恒时标记收集算法,在连续序列频繁捕获的情况下,利用一般堆内存而不是向量堆栈有效地实现递归。它可以将80%的垃圾收集开销平均减少到5%以下。该算法已被构建到一个虚拟机中,以便在汇编级有效地实现Actor语言PLASMA,一个面向Actor的PROLOG版本和SCHEME的变体,目前在8086、68000和Vax上使用。使用堆内存的基本原理是,可以通过统一内存中的单个指针获得延续,并且可以在反复捕获时以最佳方式共享,而使用基于堆栈重新复制的策略根本不可能实现这一点。此外,可以动态地对未捕获的延续进行增量垃圾收集。第一部分描述了虚拟机的基本递归指令。第二部分提出并论证了营销收集策略。第三部分以一种与当前延续的可能捕获兼容的方式安全地概括了“调用+返回=分支”转换。附录介绍了它在支持方案84的虚拟方案机中的集成。
本文章由计算机程序翻译,如有差异,请以英文原文为准。
求助全文
约1分钟内获得全文 求助全文
来源期刊
自引率
0.00%
发文量
0
×
引用
GB/T 7714-2015
复制
MLA
复制
APA
复制
导出至
BibTeX EndNote RefMan NoteFirst NoteExpress
×
提示
您的信息不完整,为了账户安全,请先补充。
现在去补充
×
提示
您因"违规操作"
具体请查看互助需知
我知道了
×
提示
确定
请完成安全验证×
copy
已复制链接
快去分享给好友吧!
我知道了
右上角分享
点击右上角分享
0
联系我们:info@booksci.cn Book学术提供免费学术资源搜索服务,方便国内外学者检索中英文文献。致力于提供最便捷和优质的服务体验。 Copyright © 2023 布克学术 All rights reserved.
京ICP备2023020795号-1
ghs 京公网安备 11010802042870号
Book学术文献互助
Book学术文献互助群
群 号:481959085
Book学术官方微信