# AI Infra 学习社区

## Docs

- [第一章：介绍与 AI 系统概览](https://se7en.mintlify.app/books/ai-systems-performance-engineering/chapters/ch01.md)
- [第二章：AI 系统硬件概览](https://se7en.mintlify.app/books/ai-systems-performance-engineering/chapters/ch02.md)
- [第三章：GPU 环境下的 OS、Docker 与 Kubernetes 调优](https://se7en.mintlify.app/books/ai-systems-performance-engineering/chapters/ch03.md)
- [第四章：分布式网络通信调优](https://se7en.mintlify.app/books/ai-systems-performance-engineering/chapters/ch04.md)
- [第五章：基于 GPU 的存储 I/O 优化](https://se7en.mintlify.app/books/ai-systems-performance-engineering/chapters/ch05.md)
- [第六章：GPU 架构、CUDA 编程与最大化占用率](https://se7en.mintlify.app/books/ai-systems-performance-engineering/chapters/ch06.md)
- [第七章：GPU 内存访问模式的分析与调优](https://se7en.mintlify.app/books/ai-systems-performance-engineering/chapters/ch07.md)
- [第八章：占用率调优、Warp 效率与指令级并行](https://se7en.mintlify.app/books/ai-systems-performance-engineering/chapters/ch08.md)
- [第九章：提升 CUDA Kernel 效率与算术强度](https://se7en.mintlify.app/books/ai-systems-performance-engineering/chapters/ch09.md)
- [第十章：Kernel 内流水线、Warp 特化与协作线程块集群](https://se7en.mintlify.app/books/ai-systems-performance-engineering/chapters/ch10.md)
- [第十一章：Kernel 间流水线、同步与 CUDA 流有序内存分配](https://se7en.mintlify.app/books/ai-systems-performance-engineering/chapters/ch11.md)
- [第十二章：动态调度、CUDA Graphs 与设备发起的 Kernel 编排](https://se7en.mintlify.app/books/ai-systems-performance-engineering/chapters/ch12.md)
- [第十三章：PyTorch 的分析、调优与扩展](https://se7en.mintlify.app/books/ai-systems-performance-engineering/chapters/ch13.md)
- [第十四章：PyTorch 编译器、OpenAI Triton 与 XLA 后端](https://se7en.mintlify.app/books/ai-systems-performance-engineering/chapters/ch14.md)
- [第十五章：多节点推理、并行、解码与路由优化](https://se7en.mintlify.app/books/ai-systems-performance-engineering/chapters/ch15.md)
- [第十六章：大规模推理的分析、调试与调优](https://se7en.mintlify.app/books/ai-systems-performance-engineering/chapters/ch16.md)
- [第十七章：推理中分离式 Prefill 与 Decode 的扩展](https://se7en.mintlify.app/books/ai-systems-performance-engineering/chapters/ch17.md)
- [第十八章：高级 Prefill-Decode 与 KV Cache 调优](https://se7en.mintlify.app/books/ai-systems-performance-engineering/chapters/ch18.md)
- [第十九章：动态与自适应推理引擎优化](https://se7en.mintlify.app/books/ai-systems-performance-engineering/chapters/ch19.md)
- [第二十章：AI 辅助性能优化与百万级 GPU 集群扩展](https://se7en.mintlify.app/books/ai-systems-performance-engineering/chapters/ch20.md)
- [书籍概览](https://se7en.mintlify.app/books/ai-systems-performance-engineering/index.md)
- [阅读计划](https://se7en.mintlify.app/books/ai-systems-performance-engineering/schedule.md)
- [AI Infra 学习社区](https://se7en.mintlify.app/index.md): 在线分享、读书会、学习资源 — 一起深入 AI Infra 技术栈
- [博客](https://se7en.mintlify.app/resources/blogs.md): AI Infra 领域的推荐博客
- [书籍](https://se7en.mintlify.app/resources/books.md): AI Infra 领域的推荐书籍
- [合集](https://se7en.mintlify.app/resources/collections.md): AI Infra 领域的资源合集与 Awesome Lists
- [课程](https://se7en.mintlify.app/resources/courses.md): AI Infra 领域的系统性课程
- [学习资源](https://se7en.mintlify.app/resources/index.md): AI Infra 领域的高质量学习资源
- [论文](https://se7en.mintlify.app/resources/papers.md): AI Infra 领域的核心论文
- [视频](https://se7en.mintlify.app/resources/videos.md): AI Infra 领域的推荐视频资源
- [01 - vLLM 快速入门](https://se7en.mintlify.app/talks/01-vllm-quickstart.md): LLM 全景图介绍与 vLLM 快速入门
- [02 - PagedAttention](https://se7en.mintlify.app/talks/02-pagedattention.md): vLLM 核心技术 PagedAttention 原理详解
- [03 - Prefix Caching](https://se7en.mintlify.app/talks/03-prefix-caching.md): 实现 KV Cache 的跨请求高效复用
- [04 - Speculative Decoding](https://se7en.mintlify.app/talks/04-speculative-decoding.md): Speculative Decoding 推测解码方案详解
- [05 - Chunked prefills](https://se7en.mintlify.app/talks/05-chunked-prefills.md): Chunked-Prefills 分块预填充机制详解
- [06 - PD 分离](https://se7en.mintlify.app/talks/06-disaggregating-prefill-and-decoding.md): PD 分离推理架构详解
- [07 - 推理平台全景](https://se7en.mintlify.app/talks/07-inference-platform.md): 开源推理平台项目介绍
- [在线分享](https://se7en.mintlify.app/talks/index.md)