Parconnect: Results from the student cluster competition at SC16

In this paper, we present results of our reproducibility study conducted as part of the SC16 Student Cluster Competition. The challenge application and its scalability study were described in a SC15 conference paper by Flick et al. [1]. The goal of the challenge was to validate the results obtained by the authors after performing the same algorithms on different data sets on our machine. We identify nine qualitative conclusions made in the performance analysis of the original work and use them to validate the results and judge reproducibility. We discuss the four tasks upon which our reproducibility study was evaluated at the competition and show that our work supports eight conclusions while one is refuted. Detailed analysis is provided in this paper.