Google对比实验：C++、Java、Go与Scala编程语言性能深度解析

需积分: 40 110 浏览量更新于2024-09-12 收藏 311KB PDF 举报

在本篇报告中，我们深入探讨了Google公司对四种主流编程语言——C++、Java、Go和Scala的运行效率进行了全面的对比研究。这份实验报告由Robert Hundt撰写，他在Google的1600 Amphitheatre Parkway地址工作，电子邮件为rhundt@google.com。报告的核心内容是通过编码一个精心设计的基准测试，该测试包括语言的典型特性展示，如容器类、循环结构以及内存管理和对象分配策略。作者并未专注于挖掘特定语言和运行时特性来达到极致性能，而是选择了一种相对公平的方法来比较语言特性、代码复杂度、编译时间和二进制大小、执行时间以及内存占用。这样做的目的是为了揭示语言本身的性能表现，而不受高级特性如多线程和类型系统的影响。基准测试本身设计得简洁而紧凑，但也包含了丰富的语言特性。例如，它涉及到了高级数据结构，如列表、映射（map）、集合（set）以及嵌套列表的使用；算法方面，如并查集（union-find）、深度优先搜索（DFS）和基于Tarjan的循环识别技术。此外，还考察了迭代操作在集合类型上的应用，以及一些面向对象特性，展示了不同的内存分配模式。值得注意的是，尽管没有探索多线程或更高级的类型系统，但这份报告仍为开发者提供了宝贵的信息，帮助他们了解在实际项目中选择不同语言时可能面临的性能差异。通过阅读这份报告，开发者能够更好地理解如何根据项目需求和团队技能来权衡各种语言的优缺点，从而做出更明智的决策。

Loop Recognition in C++/Java/Go/Scala

Robert Hundt

Google

1600 Amphitheatre Parkway

Mountain View, CA, 94043

rhundt@google.com

Abstract—In this experience report we encode a well speciﬁed,

compact benchmark in four programming languages, namely

C++, Java, Go, and Scala. The implementations each use the

languages’ idiomatic container classes, looping constructs, and

memory/object allocation schemes. It does not attempt to exploit

speciﬁc language and run-time features to achieve maximum

performance. This approach allows an almost fair comparison

of language features, code complexity, compilers and compile

time, binary sizes, run-times, and memory footprint.

While the benchmark itself is simple and compact, it em-

ploys many language features, in particular, higher-level data

structures (lists, maps, lists and arrays of sets and lists), a few

algorithms (union/ﬁnd, dfs / deep recursion, and loop recognition

based on Tarjan), iterations over collection types, some object

oriented features, and interesting memory allocation patterns.

We do not explore any aspects of multi-threading, or higher level

type mechanisms, which vary greatly between the languages.

The benchmark points to very large differences in all examined

dimensions of the language implementations. After publication of

the benchmark internally at Google, several engineers produced

highly optimized versions of the benchmark. We describe many

of the performed optimizations, which were mostly targeting run-

time performance and code complexity. While this effort is an

anecdotal comparison only, the benchmark, and the subsequent

tuning efforts, are indicative of typical performance pain points

in the respective languages.

I. INTRODUCTION

Disagreements about the utility of programming languages

are as old as programming itself. Today, these “language wars”

become increasingly heated, and less meaningful, as more

people are working with more languages on more platforms

in settings of greater variety, e.g., from mobile to datacenters.

In this paper, we contribute to the discussion by imple-

menting a well deﬁned algorithm in four different languages,

C++, Java, Go, and Scala. In all implementations, we use the

default, idiomatic data structures in each language, as well as

default type systems, memory allocation schemes, and default

iteration constructs. All four implementations stay very close

to the formal speciﬁcation of the algorithm and do not attempt

any form of language speciﬁc optimization or adaption.

The benchmark itself is simple and compact. Each imple-

mentation contains some scaffold code, needed to construct

test cases allowing to benchmark the algorithm, and the

implementation of the algorithm itself.

The algorithm employs many language features, in partic-

ular, higher-level data structures (lists, maps, lists and arrays

of sets and lists), a few algorithms (union/ﬁnd, dfs / deep

recursion, and loop recognition based on Tarjan), iterations

over collection types, some object oriented features, and

interesting memory allocation patterns. We do not explore any

aspects of multi-threading, or higher level type mechanisms,

which vary greatly between the languages. We also do not

perform heavy numerical computation, as this omission allows

ampliﬁcation of core characteristics of the language implemen-

tations, speciﬁcally, memory utilization patterns.

We believe that this approach highlights features and charac-

teristics of the languages and allows an almost fair comparison

along the dimensions of source code complexity, compilers

and default libraries, compile time, binary sizes, run-times, and

memory footprint. The differences along these dimensions are

surprisingly large.

After publication of the benchmark internally at Google,

several engineers produced highly optimized versions of the

benchmark. We describe many of the performed optimizations,

which were mostly targeting run-time performance and code

complexity. While this evaluation is an anecdotal comparison

only, the benchmark itself, as well as the subsequent tuning

efforts, point to typical performance pain points in the respec-

tive languages.

The rest of this paper is organized as follows. We brieﬂy

introduce the four languages in section II. We introduce the

algorithm and provide instructions on how to ﬁnd, build, and

run it, in section III. We highlight core language properties

in section IV, as they are needed to understand the imple-

mentation and the performance properties. We describe the

benchmark and methodology in section V, which also contains

the performance evaluation. We discuss subsequent language

speciﬁc tuning efforts in section VI, before we conclude.

II. THE CONTENDERS

We describe the four languages by providing links to the

the corresponding wikipedia entries and cite the respective

ﬁrst paragraphs from wikipedia. Readers familiar with the

languages can skip to the next section.

C++ [7] is a statically typed, free-form, multi-paradigm,

compiled, general-purpose programming language. It is re-

garded as a ”middle-level” language, as it comprises a com-

bination of both high-level and low-level language features.

It was developed by Bjarne Stroustrup starting in 1979 at

Bell Labs as an enhancement to the C language and originally

named C with Classes. It was renamed C++ in 1983.

Java [9] is a programming language originally developed

by James Gosling at Sun Microsystems (which is now a sub-

下载后可阅读完整内容，剩余9页未读，立即下载

tiangtangcao

粉丝: 9
资源: 7

Google对比实验：C++、Java、Go与Scala编程语言性能深度解析

几种WEB开发语言比较

asp、php、asp.net、jsp等主流网站开发语言比较

Java与主流WEB开发语言对比分析

JavaScript与其他Web开发语言对比解析

JavaWeb入门：历史、平台与开发语言对比

专家系统开发语言比较与选择建议

WEB开发语言对比：PHP、Java、ASP、JSP

ASP、JSP与PHP：动态网页开发语言对比详解

主流网站开发语言对比：ASP、PHP、ASP.NET、JSP分析

主流网站开发语言对比：ASP、PHP、ASP.NET与JSP的优劣分析

最新资源