Reproduce “Empowering Student Debugging in Parallel Programming with Execution Traces and Large Language Models” 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
God'salvation F. Oguibe, Vinodh Kumaran Jayakumar, Tongping Liu, Andrew Lan, Wei Wang
This paper introduces ParaView, a tool that records and visualizes execution traces for C/C++ parallel programs, enabling students to better understand and debug concurrency issues like deadlocks and race conditions. Integrating these traces with Large Language Models further enhances bug diagnosis and fix suggestions, leading to improved student success in complex parallel programming tasks.
Key insights
- Transparent execution tracing and visualization significantly enhance students' ability to identify, understand, and correct concurrency bugs.
- Providing LLMs with execution traces, in addition to source code, is critical for them to accurately diagnose and explain complex concurrency issues.
- While LLMs excel at explaining concurrency bugs and proposing basic fixes, their ability to generate optimal, context-aware, and constrained solutions for intricate synchronization patterns is currently limited.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.