In this benchmark we run a large set of context-sensitive points-to analyses on the DaCapo benchmark programs. All of Doop's features are enabled.
In this benchmark we compare Doop, running in Paddle compatible mode, to a slightly modified version of Paddle. The main purpose of this benchmark is to have both points-to analyses compute the exact same results and compare the performance. To achieve this, we have to modify the default analysis of Doop, as well as Paddle to work around some differences between the two.
In this benchmark we compare Doop, with all features enabled, to Paddle, also with all its features enabled. The two analyses differ in various ways. For example, Doop performs a very precise exception analysis and does a more sophisticated reflection analysis than Paddle, while Paddle analyses class initializers context-sensitively.
We also plan to add comparisons to other frameworks in the future, but it takes a lot of time to evaluate if the results are optimal and remotely comparable, so don't hold your breath for this.