A compression algorithm may be implemented with many different, yet reasonable, data structures (including binary tree, splay tree, trie, hash table, and list) and yield vastly different performance results [3]. The quality and applicability of the implementation is as important as the underlying algorithm. This section has presented implementations from each algorithmic family. By choosing a top representative in each family, the implementation playing field is leveled, making it easier to gain insight into the underlying algorithm and its influence on energy. Nevertheless, it is likely that each application can be optimized further (Section 4.1 shows the benefit of optimization) or use a more uniform style of I/O. Thus, evaluation must focus on inherent patterns rather than making a direct quantitative comparison.