Characterizing Real-World Bugs in Tile Programs for Automated Bug Detection
A new study investigates code generation bugs in tile-based programming frameworks, which are increasingly used for high-performance GPU kernels. The research identifies 301 specific bugs and categorizes their root causes and symptoms. This work aims to enhance debugging and testing tools for tile-based compiler infrastructures.
- ▪Tile-based programming frameworks are popular for writing GPU kernels in deep learning and scientific computing.
- ▪The study curated 401 bug reports from GitHub, identifying 301 tile-program code generation bugs.
- ▪The research categorizes the bugs by root causes, symptoms, input patterns, and strategies for fixing them.
Opening excerpt (first ~120 words) tap to expand
Computer Science > Software Engineering arXiv:2605.19652 (cs) [Submitted on 19 May 2026] Title:Characterizing Real-World Bugs in Tile Programs for Automated Bug Detection Authors:Ravishka Rathnasuriya, Zihe Song, Nidhi Majoju, Aaryaa Moharir, Tingxi Li, Wei Yang, Tao Xie View a PDF of the paper titled Characterizing Real-World Bugs in Tile Programs for Automated Bug Detection, by Ravishka Rathnasuriya and 6 other authors View PDF HTML (experimental) Abstract:Tile-based programming frameworks are increasingly adopted to write high-performance GPU kernels in domains such as deep learning and scientific computing.
…
Excerpt limited to ~120 words for fair-use compliance. The full article is at arXiv.org.
Discussion
0 commentsMore from arXiv.org
