HeCBench
Emerging5papers using it
2024first seen
HeCBench is a dataset containing 577 CUDA kernels annotated with execution attributes and ground-truth profiles, used to evaluate the ability of Large Language Models (LLMs) to predict floating-point operation counts without executing the code.
Papers using HeCBench (5)
- Leveraging LLMs to Automate Energy-Aware Refactoring of Parallel Scientific CodesCan Large Language Models Predict Parallel Code Performance?ParaCodex: A Profiling-Guided Autonomous Coding Agent for Reliable Parallel Code Generation and TranslationCounting Without Running: Evaluating LLMs' Reasoning About Code ComplexityOMPar: Automatic Parallelization with AI-Driven Source-to-Source
Compilation