The LLVM Project Blog

LLVM Project News and Details from the Trenches

Interactive programming in C++ internships — LLVM Blog Post

Another program year is ending and our Compiler Research team is extremely happy to share the hard work and the results of our interns contributors!

The Compiler Research team includes researchers located at Princeton University and CERN. Our primary goal is research into foundational software tools helping scientists to program for speed, interoperability, interactivity, flexibility, and reproducibility. We work in various fields of science such as high energy physics, where research is fundamentally connected to software at exabyte scale. We develop computational methods and research software for scientific exploration and discovery. Our current research focuses on three main topics: interpretative C/C++/CUDA , automatic differentiation tools, and C++ language interoperability with Python and D.

This year, we had six participants who worked on seven projects, covering a wide range of topics, from LLVM’s new JIT linker (JITLink) to Clang-Repl, an interactive C++ interpreter for incremental compilation integrated in Clang. All projects rise toward a common, ambitious goal: to establish a proficient workflow in LLVM, where interactive development in C++ is possible, and exploratory C++ becomes an accessible experience to a wider audience.

Below you can find the list of projects from our interns, an overview of the objectives and results. We invite you to follow the links to access a more detailed description of each project.

Developer: Sunho Kim (Computer Science, De Anza College, Cupertino, California)

Mentors: Stefan Gränitz (Freelance Compiler Developer, Berlin, Deutschland), Lang Hames (Apple), Vassil Vassilev (Princeton University/CERN)

Funding: Google Summer of Code 2022

Suhno developed a JITLink specialization that extends JITLink’s generic linker algorithm, allowing JITLink to support ELF/aarch64 target and provides full support of all advanced PE/COFF object file features. By supporting the ELF/aarch64 target, JITLink can now be used in Julia, while COFF/X86_64 enables Microsoft Visual C++ (MSCV) target in Clang-Repl. Here you can find Sunho’s GSoC final report.

Hurray! This project has been accepted as a Tutorial at the 2022 LLVM Developers’ Meeting! The conference took place in San Jose, California, from 7th to 10th November 2022. Here you can find the video and the slides of Sunho’s presentation.

The project was also presented at the Compiler As a Service (CaaS) monthly meeting. Here you can find the video and the slides of Sunho’s presentation.

Extend Clang to resugar template specialization accesses

Developer: Matheus Izvekov

Mentors: Richard Smith (Google), Vassil Vassilev (Princeton University/CERN)

Funding: Google Summer of Code 2022

Clang’s type system was optimized by pushing type-syntactic-sugar on the arguments of a template specialization into those member accesses. This was achieved by: 1. creating a new type node into the Abstract Syntax Tree (AST) that represents the sugar of a member access in a template specialization; and 2. implementing single step desugaring logic which will perform the substitution of template parameter sugar into the corresponding specialization argument sugar. As a result, we improved Clang’s diagnostic system by enabling other constructs to preserve type sugar, allowing both for their representation when present in the specialization argument. Here you can find Matheus' GSoC final report.

Hurray! This project has been accepted as a Lightning Talk at the 2022 LLVM Developers’ Meeting! The conference took place in San Jose, California, from 7th to 10th November 2022. Here you can find the video and the slides of Matheus’ presentation.

Recovering from Errors in Clang-Repl and Code Undo

Developers: Jun Zhang (Software Engineering, Anhui Normal University, WuHu, China) and Purva Chaudhari (California State University Northridge, Northridge CA, USA)

Mentors: Vassil Vassilev (Princeton University/CERN)

Clang-Repl is an interactive C++ interpreter integrated in Clang, which enables incremental compilation. It supports interactive programming for C++ in a Read-Eval_Print Loop (REPL) style, compiling the code just-in-time with a JIT approach that reduces the compile-run cycles. We added the “undo” functionality to Clang-REPL, and we improved error-recovery by adding the possibility to recover the low-level execution infrastructure. As a result, Clang-REPL now supports reversal execution and recovery actions on behalf of a user in an automatic and convenient way.

Hurray! This project has been accepted as a Lightning Talk at the 2022 LLVM Developers’ Meeting! The conference took place in San Jose, California, from 7th to 10th November 2022. Here you can find the video and the slides of Purva and Jun’s presentation.

Developer: Anubhab Ghosh (Computer Science and Engineering, Indian Institute of Information Technology, Kalyani, India)

Mentors: Stefan Gränitz (Freelance Compiler Developer, Berlin, Deutschland), Lang Hames (Apple), Vassil Vassilev (Princeton University/CERN)

Funding: Google Summer of Code 2022

Anubhab introduced a new JITLinkMemoryManager based on a MemoryMapper abstraction that is capable of allocating JIT code (and data) using shared memory. The following advantages should arise from the implemented strategy: 1. a faster transport (and access) for code and data when running JIT’d code in a separate process on the same machine, and 2. the guarantee that all allocations are close together in memory and meet the constraints of the default code model allowing the use of outputs from regular compilers. Here you can find Anubhab’s GSoC final report.

Hurray! This project has been accepted as a Lightning Talk at the 2022 LLVM Developers’ Meeting! The conference took place in San Jose, California, from 7th to 10th November 2022. Here you can find the video and the slides of Anubhab’s presentation.

Optimize ROOT use of modules for large codebases

Developer: Jun Zhang (Software Engineering, Anhui Normal University, WuHu, China)

Mentors: David Lange (Princeton University), Alexander Penev (University of Plovdiv Paisii Hilendarski, Bulgaria), Vassil Vassilev (Princeton University/CERN)

Funding: Google Summer of Code 2022

The current performance of the modules usage in ROOT was evaluated, and a strategy for optimizing the memory footprint was developed. This strategy includes reducing unnecessary symbol-lookups and module-loading, and is especially useful for very large codebases like CMSSW, where Jun reduced the amount of memory by half (from 1.1GB to 600MB) for simple workflows like hsimple.C, and the number of loaded from 180 to 52. Here you can find Jun’s GSoC final report.

Add Initial Integration of Clad with Enzyme

Developer: Manish Kausik H (.Tech and M.Tech in Computer Science and Engineering(Dual Degree), Indian Institute of Technology Bhubaneswar)

Mentors: David Lange (Princeton University), William Moses (Massachusetts Institute of Technology), Vassil Vassilev (Princeton University/CERN)

Funding: Google Summer of Code 2022

Clad is an open source plugin to the Clang compiler that enables Automatic Differentiation for C++. Clad receives an Abstract Syntax Tree (AST) from the underlying
compiler platform (Clang), decides whether a derivative is requested and produces it, and modifies the AST to insert the generated code. Enzyme AD is an LLVM based AD plugin. It works by taking existing code as LLVM IR and computing the derivative (and gradient) of that function. In this project, Manish integrated Enzyme within Clad, giving a Clad user the option of selecting Enzyme for Automatic Differentiation on demand. The integration of Clad and Enzyme t results in a new tool that offers an optimized and flexible implementation of automatic differentiation. Here you can find Manish’s GSoC final report.

Thank you, developers!

We hope our interns contributors enjoyed our community. Our best reward is to know that we supported your early steps into the world of open-source software development and compiler construction. We hope that this experience motivated you to continue to be involved with the broad LLVM ecosystem. We express our gratitude to the Google Summer of Code Program and to the Institute for Research and Innovation in Software for High Energy Physics (IRIS-HEP) for supporting our research and providing six young developers this wonderful opportunity. A special thanks goes to the LLVM community for being supportive and for sharing their knowledge and introducing our community to the new contributors. Thank you so much! Thanks for choosing Compiler-Research for your internship! We were lucky to have you!