Best Paper Nomination at Euro-Par 2025

I’m thrilled to share that our recent paper, led by my PhD student Jeffrey Spaan (co-supervised with Ana-Lucia Varbanescu), has received a Best Paper Nomination at European Conference on Parallel Processing (Euro-Par 2025)! 🏆

Paper Title: Wedge-Parallel Triangle Counting for GPUs
Authors: Jeffrey Spaan, Kuan-Hsun Chen, David A. Bader and Ana-Lucia Varbanescu

🔍 What’s the paper about?

Triangle counting is a fundamental task in graph analytics, used in community detection, link prediction, and more. However, making it efficient on GPUs—despite their massive parallel processing power—is far from trivial due to the irregular nature of real-world graphs.

In this paper, we present WeTriC, a novel wedge-parallel triangle counting algorithm designed specifically for GPU architectures. Unlike traditional vertex- or edge-parallel strategies, WeTriC maps wedges—two-hop paths that may form triangles—directly to GPU threads using a lightweight static strategy that drastically improves load balancing and parallel efficiency.

⚙️ Why is it special?

• 📌 Fine-grained parallelism for better GPU utilization

• 🧠 Theoretical analysis of different parallelization strategies

• ⚡ Optimizations that reduce memory overhead and improve cache locality

• 🚀 Substantial speedups over existing methods:

  • 5.63× faster than vertex-parallel baseline
  • 4.69× faster than edge-parallel binary search method
  • 2.86× faster than Trust
  • 2.32× faster than GroupTC

🏅 Why it matters?

WeTriC pushes the boundary of efficient GPU-based graph analytics, especially for large-scale graph datasets. The work highlights how algorithmic design and hardware-aware optimizations go hand-in-hand to achieve state-of-the-art performance.

I am incredibly proud of Jeffrey Spaan and the team for this outstanding recognition! 🎉

Stay tuned for more details once the paper is presented at Euro-Par 2025.