Artifact of the paper: Engineering Scalable Distributed List Ranking

The kascade benchmark suite is the artifact accompanying "Engineering Scalable Distributed List Ranking."It consists of a C++ list ranking library implementing all algorithm variants described in the paper and a corresponding benchmark runner.Experiments are driven by kaval, a Python-based experiment runner, using YAML suite descriptions that specify algorithm configurations, input sizes, and core counts.Inputs are generated synthetically on-the-fly by the bundled KaGen graph generator; no external datasets are required.A Julia-based evaluation pipeline parses the JSON output and generates the plots.Reproducibility scripts regenerate Figures 2–5 of the paper.The maximum core count can be adjusted to run on smaller shared-memory machines instead of large HPC systems.Correctness and relative ordering of algorithm variants can be verified at small scale, while the benefits of message indirection (Figures 4–5) and the full scalability range (Figure 3) require at least 1,000 cores on a distributed HPC system.See REPRODUCING.md in the archive for full setup and step-by-step instructions.