Language implementations that use a uniform pointer representation for generic datatypes typically apply boxing operations to convert non-uniform objects into the required form. The cost of this boxing process impacts upon the performance of programs that make heavy use of genericity with non-uniform data such as integers and other primitive value types. We show that the overhead of boxing objects into heap storage can be significantly reduced by taking a lazy approach: allowing pointers to stack-allocated objects that are only copied to the heap when necessary. Delaying the boxing of objects avoids unnecessary heap allocation, and results in speedups of around 25% for a range of test programs. © Springer-Verlag 2004.
CITATION STYLE
Owen, T., & Watson, D. (2004). Reducing the cost of object boxing. Lecture Notes in Computer Science (Including Subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), 2985, 202–216. https://doi.org/10.1007/978-3-540-24723-4_14
Mendeley helps you to discover research relevant for your work.