Current RepositoryQuery.php implementation, https://github.com/Islandora/tuque/blob/1.x/RepositoryQuery.php#L37-L69
loads the whole result document into memory using a simplexml element and iterates over it's children using foreach in two nested loops. This behaviour works well for small-medium results or when the answer is limited to a few 1000+- responses. But on large queries, this triggers quick memory exhaustion inside the foreach structure.
Since php 5.1.x XMLReader() is enabled by default. It enables huge xml files to be parsed element by element with minimum memory usage.