Wade Holst

Multicodes: optimizing virtual machines using bytecode sequences

A virtual machine optimization technique that makes use of bytecode sequences is introduced. The process of determining candidate sequences is discussed and performance gains achieved when applied to a Java interpreter are presented. The suitability of this optimization for JVMs that perform just-in-time compilation is also discussed.

Meta: extending and unifying languages

<i>Meta</i> is an ambitious research project whose overall purpose is to increase the utility and expressive power of a wide range of existing languages. <i>Meta</i> provides augmented versions of existing languages and guarantees support for aspects, components, language interoperability, visualization, reflection, various inheritance models, and many other extensions.

Advancements in multicode optimization

In previous work, we have shown that multicodes can be used to improve the performance of Java applications. We extend that work by both implementing more multicodes and considering multicodes of greater length. This has resulted in significantly larger performance gains than those that have been achieved previously for some benchmarks.

Meta: a universal meta-language for augmenting and unifying language families, featuring meta(oopl) for object-oriented programming languages

Meta: a collection of meta-languages that augment/unify language-families, an advanced programming environment, a language interoperability framework, an infrastructure for reflection over source documents, and a generalization of XML. XML provides an extensible mechanism for defining syntax, while Meta provides an extensible mechanism for defining syntax and semantics.

Multi-dispatch in the Java Virtual Machine (poster session): design and implementation

Mainstream object-oriented languages, such as C++ and Java, provide only a restricted form of polymorphic methods, namely single-receiver dispatch. In common programming situations, programmers must work-around this limitation. We detail how to extend the Java Virtual Machine to support multiple-dispatch and examine the complications that Java imposes on multiple-dispatch in practice. Our technique avoids changes to the Java programming language itself, maintains source-code and library compatibility, and isolates the performance penalty and semantic changes of multiple-dispatch to the program sections which use it. We have micro-benchmark and application-level performance results for a dynamic Most Specific Applicable (MSA) dispatcher, two table-based dispatchers (Multiple Row Displacement (MRD) and Single Receiver Projections (SRP)), and a tuned SRP dispatcher. Our general-purpose technique provides smaller dispatch latency than equivalent programmer-written double-dispatch code.

An evaluation of secialized Java bytecodes

Specialized Java bytecodes provide functionality that is easily replicated using other Java bytecodes. This study uses profiling to explore how the set of specialized bytecodes currently implemented by the Java Virtual Machine is utilized by comparing it to the other specialized bytecodes which could have been implemented.

A technique for utilizing optimization potential during multicode identification

Multicodes have been demonstrated to provide performance gains of up to 25 percent. Present multicode identification techniques rely on frequency of occurrence and sequence length alone. This research extends previous work by presenting a multicode identification algorithm based on the frequency of occurrence and the optimization potential of the sequence.