On the adaptive instantiation of type-specific collections
| Autor: | Bruno Sofiato, Ricardo Luis de Azevedo da Rocha |
|---|---|
| Rok vydání: | 2017 |
| Předmět: |
Object-oriented programming
Measure (data warehouse) Java Computer science Programming language Aspect-oriented programming 020207 software engineering 02 engineering and technology computer.software_genre Set (abstract data type) 020204 information systems 0202 electrical engineering electronic engineering information engineering General Earth and Planetary Sciences Compiler computer Implementation General Environmental Science computer.programming_language |
| Zdroj: | ANT/SEIT |
| ISSN: | 1877-0509 |
| DOI: | 10.1016/j.procs.2017.05.392 |
| Popis: | After an age of almost unlimited computing resources, the advent of resource-constrained computing devices such as cell phones, brought questions of performance back into the spotlight. Programming languages usually provide a set of general-purpose col- lections. Albeit highly optimized, they fall short of type-specific implementations in both processing time and memory usage. We describe a method for automatically deciding which implementation to use in runtime. Two complementary strategies for deter- mining whether a type-specific collection should be employed are described. To measure their gains, we conducted an experiment composed of eleven benchmarks, ranging from databased to compilers. Results show improvements of 13% in processing and of 3.5% in memory usage. As an unexpected side-effect, our technique reduced instantiations of collection in runtime. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|