Advanced Compilers: The Self-Guided Online Course
摘要
这是 CS 6120（Advanced Compilers）的自学在线课程页面，由 Cornell 的 Adrian Sampson 设计，面向编译器与编程语言实现的进阶学习。课程以线性 lesson 组织，每一节结合视频、阅读论文和开放式编程任务，使用 LLVM 和专门设计的教学 IR（Bril）进行实践。 内容覆盖从基础的程序表示、局部/全局数据流分析、SSA，到 LLVM pass 编写、循环优化、跨过程分析、别名分析与内存管理，再到动态编译器、JIT、并行与并发优化等研究型主题，并穿插多篇经典论文阅读。 自学版本与正式课程的区别包括没有截止日期、无法参与讨论区，以及用“改变世界的编译器项目”替代期末项目。整体结构强调通过读论文 + 动手实现优化与分析来理解编译器核心技术。
荐读理由
把课程里的 IR 设计与 LLVM pass 思路拆出来，可以直接用于自己在 AI 工程里设计代码生成或执行链路时的优化与改写机制，比如用数据流分析和 SSA 思想改造中间表示以支持后续优化与转换
原文
CS 6120: Advanced Compilers: The Self-Guided Online Course
CS 6120 is a PhD-level Cornell CS course by Adrian Sampson on programming language implementation. It covers universal compilers topics like intermediate representations, data flow, and “classic” optimizations as well as more research-flavored topics such as parallelization, just-in-time compilation, and garbage collection. The work consists of reading papers and open-source hacking tasks, which use LLVM and an educational IR invented just for this class.
This page lists the curriculum for following this course at the university of your imagination, for four imagination credits (ungraded). There’s a linear timeline of lessons interspersed with papers to read. Each lesson has videos and written notes, and some have implementation tasks for you to complete. Tasks are all open-ended, to one degree or another, and are meant to solidify your understanding of the abstract concepts by turning them into real code. The order represents a suggested interleaving of video-watching and paper-reading.
Some differences with the “real” CS 6120 are that you can ignore the task deadlines and you can’t participate in our discussion threads on Zulip. Real 6120 also has an end-of-semester course project—in the self-guided version, your end-of-semester assignment is to change the world through the magic of compilers.
The instructor is a video production neophyte, so please excuse the production values, especially in the early lessons. CS 6120 is open source and on GitHub, so please file bugs if you find problems.
When you finish the course, please fill out this feedback form.
Lesson 1: Welcome & Overview
Producing Wrong Data Without Doing Anything Obviously Wrong! Todd Mytkowicz, Amer Diwan, Matthias Hauswirth, and Peter F. Sweeney. ASPLOS 2009.
Lesson 2: Representing Programs
Lesson 3: Local Analysis & Optimization
Lesson 4: Data Flow
Lesson 5: Global Analysis
Efficient Path Profiling Thomas Ball and James R. Larus. MICRO 1996.
Lesson 6: Static Single Assignment
Provably Correct Peephole Optimizations with Alive Nuno P. Lopes, David Menendez, Santosh Nagarakatte, and John Regehr. PLDI 2015.
Lesson 7: LLVM
Lesson 8: Loop Optimization
Lesson 9: Interprocedural Analysis
Type-Based Alias Analysis Amer Diwan, Kathryn S. McKinley, and J. Eliot B. Moss.
Lesson 10: Alias Analysis
A Unified Theory of Garbage Collection David F. Bacon, Perry Cheng, and V. T. Rajan. OOPSLA 2004.
Fast Conservative Garbage Collection Rifat Shahriyar, Stephen M. Blackburn, and Kathryn S. McKinley. OOPSLA 2014.
Lesson 11: Memory Management
An Efficient Implementation of SELF, a Dynamically-Typed Object-Oriented Language Based on Prototypes C. Chambers, D. Ungar, and E. Lee. OOPSLA 1989.
Trace-Based Just-in-Time Type Specialization for Dynamic Languages Andreas Gal, Brendan Eich, Mike Shaver, David Anderson, David Mandelin, Mohammad R. Haghighat, Blake Kaplan, Graydon Hoare, Boris Zbarsky, Jason Orendorff, Jesse Ruderman, Edwin W. Smith, Rick Reitmaier, Michael Bebenita, Mason Chang, and Michael Franz. PLDI 2009.
Lesson 12: Dynamic Compilers
Superoptimizer: A Look at the Smallest Program Alexia Massalin. ASPLOS 1987.
Chlorophyll: Synthesis-Aided Compiler for Low-Power Spatial Architectures Phitchaya Mangpo Phothilimthana, Tikhon Jelvis, Rohin Shah, Nishant Totla, Sarah Chasins, and Rastislav Bodik. PLDI 2014.
Lesson 13: Concurrency & Parallelism
Threads Cannot Be Implemented as a Library Hans-J. Boehm. PLDI 2005.
Exploiting Superword Level Parallelism with Multimedia Instruction Sets Samuel Larsen and Saman Amarasinghe. PLDI 2000.
A Type and Effect System for Deterministic Parallel Java Robert L. Bocchino, Vikram S. Adve, Danny Dig, Sarita V. Adve, Stephen Heumann, Rakesh Komuravelli, Jeffrey Overbey, Patrick Simmons, Hyojin Sung, and Mohsen Vakilian. OOPSLA 2009.
Formal Verification of a Realistic Compiler Xavier Leroy. CACM in 2009.
Lesson 14: Fast Compilers
这条对你有帮助吗？