Implementing an Efficient Java Interpreter
Implementing an Efficient Java Interpreter
复制标题
实现高效的 Java 解释器
DOI：
--
复制
发表时间：
2001
期刊：
影响因子：
--
通讯作者：
A. Krall
中科院分区：
文献类型：
--
作者：
David Gregg;M. Ertl;A. Krall
Java虚拟机（JVM）通常使用解释器或即时（JIT）编译器实现。JIT编译器提供了最好的性能，但必须针对它们移植到的每个体系结构进行大量重写。解释器更容易开发和维护，并且几乎不需要任何更改就可以移植到新的体系结构中。解释器的缺点是它们比JIT编译器慢得多。本文描述了一个高效Java解释器的开发过程。我们描述了使解释器高效的主要特性。我们最初的实验结果表明，对于一些重要的应用程序，基于解释器的JVM可能只比基于编译器的JVM慢1.9倍。
The Java virtual machine (JVM) is usually implemented with an interpreter or just-in-time (JIT) compiler. JIT compilers provide the best performance, but must be substantially rewritten for each architecture they are ported to. Interpreters are easier to develop and maintain, and can be ported to new architectures with almost no changes. The weakness of interpreters is that they are much slower than JIT compilers. This paper describes work in progress on a highly efficient Java interpreter. We describe the main features that make our interpreter efficient. Our initial experimental results show that an interpreter-based JVM may be only 1.9 times slower than a compiler-based JVM for some important applications.