Abstract: Parallel programmers are frequently tasked with modifying, enhancing, and extending parallel applications. To perform these tasks and maintain correctness, parallel programmers must understand existing code by forming mental representations. The comprehension of parallel code requires programmers to mentally execute multiple timelines that are occurring in parallel at the machine level. The goal of the proposed research is to develop a model for parallel program comprehension. The study will investigate the mental representations formed by expert parallel programmers during the comprehension of parallel programs. The task used to stimulate the comprehension process will be verifying the correctness of parallel programs by determining the presence of data races. Eye tracking data and questionnaires will be used to formulate a model.
PPIG 2019 - 30th Annual Workshop
Constructing a Model of Expert Parallel Programmers’ Mental Representations Formed During Parallel Program Comprehension