Just-in-time compilers offer the biggest achievable payoff performance-wise, but their implementation is a non-trivial, time-consuming task affecting the interpreter’s maintenance for years to come, too. Recent research addresses this issue by providing ways of leveraging existing just-in-time compilation infrastructures.
Though there has been considerable research on improving the efficiency of just-in-time compilers, the area of optimizing interpreters has gotten less attention as if the implementation of a dynamic translation system was the “ultima ratio” for efficiently interpreting programming languages. We present optimization techniques for improving the efficiency of interpreters without requiring just-in-time compilation thereby maintaining the ease-of-implementation characteristic that brought many people to implementing an interpreter in the first place.