Chapel Graph Library (CGL)
summary
The project presents a hypergraph-backed graph library built on Chapel that exposes a global-view API and implements graphs (bipartite, directed, undirected, trees) on top of hypergraphs, plus an aggregation buffer (CAL) to batch fine-grained updates. Benchmarks compare CGL to C/C++ implementations, show comparable single-node speed and limited scalability at high thread counts, and identify hotspots fixed by adjacency caching and atomic edge-claiming. Useful specifics: Chapel Graph Library (CGL), Chapel Aggregation Library (CAL), hypergraph-backed graphs.
AI-written summary · about