ECCO:基于证据的编译器优化因果推理
ECCO: Evidence-Driven Causal Reasoning for Compiler Optimization
AI总结:
ECCO通过结合可解释推理与组合搜索,提升编译器优化性能,实现24.44%的平均周期减少。
摘要:
编译器自动调优面临着传统黑盒搜索方法缺乏语义指导与近期大型语言模型(LLM)方法常面临表面模式匹配和因果不透明的二元对立。在本文中,我们介绍ECCO,一个连接可解释推理与组合搜索的框架。我们首先提出一种逆向工程方法来构建一个思维链数据集,明确地将静态代码特征映射到可验证的性能证据。这使模型能够学习支配优化决策的因果逻辑,而不是仅仅模仿序列。利用这种可解释的先验知识,我们设计了一种协作推理机制,其中LLM作为战略家,定义优化意图,动态引导遗传算法的突变操作。在七个数据集上的实验结果表明,ECCO显著优于LLVM opt -O3基线,平均减少24.44%的周期数。
英文摘要:
Compiler auto-tuning faces a dichotomy between traditional black-box search methods, which lack semantic guidance, and recent Large Language Model (LLM) approaches, which often suffer from superficial pattern matching and causal opacity. In this paper, we introduce ECCO, a framework that bridges interpretable reasoning with combinatorial search. We first propose a reverse engineering methodology to construct a Chain-of-Thought dataset, explicitly mapping static code features to verifiable performance evidence. This enables the model to learn the causal logic governing optimization decisions rather than merely imitating sequences. Leveraging this interpretable prior, we design a collaborative inference mechanism where the LLM functions as a strategist, defining optimization intents that dynamically guide the mutation operations of a genetic algorithm. Experimental results on seven datasets demonstrate that ECCO significantly outperforms the LLVM opt -O3 baseline, achieving an average 24.44% reduction in cycles.
机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。