A survey of interpretation and compilation in virtual machines

Abstract

Bytecode execution by virtual machine has several advantages over traditional native code execution. These are portability, safety, ease of compiling and debugging. However, using virtual machine gives an additional overhead. Two ways of bytecode execution are described in this paper. These are interpretation and compilation. Bottlenecks and overheads of each of these methods are discussed. A survey of major optimizations of these methods which allows achieving execution speed comparable with native code execution speed is given.