Reproduce “An Empirical Study of Issues in GPU Kernel Libraries for LLM Systems” on OpenResearch

Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.

Universal app for Apple silicon and Intel. Requires macOS 11 or later.

About this paper

Yanjie Gao, Jiayi Bi, Haoxiang Lin, Ling Feng

The authors analyze 300 GitHub issues from seven open-source GPU kernel libraries, mapping symptoms, causes, fixes, and testing practices. They classify causes across kernel-library, compiler-and-build-system, and runtime-environment dimensions. Compiler/build and runtime-environment causes account for 60.33% of the sample; this applies only to selected issue reports, while kernel-library causes account for 39.67%.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.