Static and Dynamic Program Analysis: Synergies and Applications

Опубликовано 17 августа 2016, 0:58

Modern computing platforms pose unprecedented challenges to productively building reliable, scalable, and energy-efficient software. I will show how static and dynamic program analysis can be combined in novel ways to effectively address these challenges. My talk will focus on three challenges: seamlessly partitioning programs for rich mobile computing, automatically estimating program performance for predictable cloud computing, and scalably verifying programs for reliable parallel computing. I will demonstrate how we can significantly reduce energy consumption on mobile devices by offloading compute-intensive parts of rich apps to the cloud, how we can automatically predict the running time of general-purpose programs accurately and efficiently, and how we can scalably prove vast parts of real-world concurrent programs thread-safe while exposing harmful concurrency bugs in the remaining parts.