makora-ai/OptimizationCache
收藏Hugging Face2025-12-28 更新2026-01-03 收录
下载链接:
https://hf-mirror.com/datasets/makora-ai/OptimizationCache
下载链接
链接失效反馈官方服务:
资源简介:
该数据集是一个精选的GPU内核优化技术集合,用于与Lenny(GPU内核优化的AI助手)一起使用。数据集包含优化条目,每个条目包括描述、模式、指令、代码片段和文档。优化条目分为几个类别,如内存(内存合并、向量化访问、共享内存、银行冲突)、计算(张量核心、warp原语、循环展开)、并行性(占用调优、线程配置)和架构(PTX指令、双缓冲)。
A curated collection of GPU kernel optimization techniques for use with Lenny, the AI assistant for GPU kernel optimization. This dataset contains optimization entries that help guide LLM-based kernel generation. Each entry includes: Description, Patterns, Instructions, Code Snippets, and Documentation. Categories include: Memory (Memory coalescing, vectorized access, shared memory, bank conflicts), Compute (Tensor cores, warp primitives, loop unrolling), Parallelism (Occupancy tuning, thread configuration), and Architecture (PTX instructions, double buffering).
提供机构:
makora-ai



