arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2603.10085cs.LGcs.AIcs.MA
收藏

KernelSkill:一种用于GPU内核优化的多智能体框架

KernelClaw: A Multi-Agent Framework for GPU Kernel Optimization with Dual-Memory Mechanisms

Qitong Sun, Jun Han, Tianlin Li, Zhe Tang, Sheng Chen, Fei Yang, Aishan Liu, Xianglong Liu, Yang Liu

AI总结:

KernelSkill通过多智能体框架和双层记忆架构,实现高效GPU内核优化,取得显著速度提升。

摘要:

提高GPU内核效率对于推进AI系统至关重要。近期的研究探索了利用大语言模型(LLMs)进行GPU内核生成和优化。然而,现有的基于LLM的内核优化流程通常依赖于LLM内部隐含学习的启发式方法来确定优化策略,这导致了低效的试错过程和可解释性差的优化。我们的关键见解是用知识驱动且了解任务轨迹的专家优化技能取代隐含的启发式方法。具体来说,我们提出了KernelSkill,一种具有双层记忆架构的多智能体框架。KernelSkill通过协调具有可重用专家技能长期记忆的智能体和短期记忆来防止重复回溯。在KernelBench级别1-3上,KernelSkill实现了100%的成功率,平均速度提升分别为5.44倍、2.82倍和1.92倍,优于先前的基线方法。代码可在https://github.com/0satan0/KernelMem/上获得。

英文摘要:

Improving GPU kernel efficiency is crucial for efficient AI systems. Recent work has explored leveraging large language models (LLMs) for GPU kernel generation and optimization. However, existing LLM-based kernel optimization pipelines typically rely on opaque, implicitly learned heuristics within the LLMs to determine optimization strategies. This leads to inefficient trial-and-error and weakly interpretable optimizations. Our key insight is to replace implicit heuristics with expert optimization knowledge that is aware of task trajectories. Specifically, we present KernelClaw, a multi-agent framework with a dual-level memory architecture. KernelClaw operates by coordinating agents with long-term memory of reusable expert knowledge and short-term memory to prevent repetitive backtracking. On KernelBench Levels 1-3, KernelClaw achieves a 100% success rate and average speedups of 6.98x, 7.89x, and 1.94x over Torch Eager on Levels 1, 2, and 3, respectively, outperforming prior baselines. Code is available at https://github.com/0satan0/KernelMem/.

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。

↑