Darwinian Data Structure Selection

Data structure selection and tuning is laborious but can vastly improve application performance and memory footprint. We introduce ARTEMIS a multiobjective, cloud-based optimisation framework that automatically finds optimal, tuned data structures and rewrites applications to use them. ARTEMIS achieves substantial performance improvements for every project in a set of 29 Java programs uniformly sampled from GitHub. For execution time, CPU usage, and memory consumption, ARTEMIS finds at least one solution for each project that improves all measures. The median improvement across all these best solutions is 8.38% for execution time, 24.27% for memory consumption and 11.61% for CPU usage. In detail, ARTEMIS improved the memory consumption of JUnit4, a ubiquitous Java testing framework, by 45.42% memory, while also improving its execution time 2.29% at the cost a 1.25% increase in CPU usage. LinkedIn relies on the Cleo project as their autocompletion engine for search. ARTEMIS improves its execution time by 12.17%, its CPU usage by 4.32% and its memory consumption by 23.91%.