Introduction to LLVM (2023)

LLVM is great software to learn and complicated software to install and get started with. It has hundreds of tools and options that can easily overwhelm a novice trying to get started with the system. This document aims to solve this problem and provide an easy way to get started with LLVM. We provide the minimum instructions needed to install LLVM, write your first pass, and run the pass. After following this document, you may be able to find more complicated documentation online.


Many people find installing LLVM difficult and have written excellent guides that you can find online. The two most important (and very descriptive) ones come from the LLVM website itself. Links to both are below. While these two documents provide a lot of detail about the installation, the problem is that they also provide a lot of detail, which the novice does not need. You can skip these documents for now and come back to them later. My recommendation is to finish the tutorial on the current website, do some research on LLVM by writing/modifying passes, and then read both docs.

  1. Introduction to the LLVM system
  2. Writing an LLVM pass

A note about the host environment

The instructions in this document have been testedKubuntu 18.04. While the installation steps remain basically the same for other platforms, the instructions for installing individual packages differ on MAC and Windows. Look for support online.

Step 0. Required software

  1. To make
    sudo apt installeer cmake
    [Tested with cmake version 3.10.2]
  2. ninja
    sudo apt install ninja-build
    [Tested with ninja version 1.8.2]
  3. Sound
    sudo apt install clang
    [Tested with clang version 6.0.0-1ubuntu2 (tags/RELEASE_600/final)]
  4. Eclipse - with support for C++ development
    Install from the official website.
    [Tested with Eclipse 2019-06 (4.12.0)]

Phase 1. Clone the LLVM repository

LLVM releases newer versions fairly regularly. The features in each major release differ greatly from each other and would break backward compatibility. To ensure that everyone is using the same version of LLVM and to avoid problems during evaluation, we use the latest stable version.

cd ~git klon llvmcd llvmgit odjava -b cs380c llvmorg-(versão)

The repository is approximately 1.27 GB in size. Make sure you have a good network connection before cloning. After the above commands you will entercs380cchildren's.

Step 2. Upgrade LLVM

LLVM VSTo maketo generate a build system. You can specify your preferred build system and the specified build system will later be used to build the native LLVM files. Some of the interesting construction systems are:

  • ninja- to generate Ninja build files. Most llvm developers use Ninja.
  • Makefiles Unix— to generate parallel make-compatible make files.
  • visual studio— for generating Visual Studio projects and solutions.
  • X-kod- for generating Xcode projects.
  • Overshadow

You can choose any build system you want on your machine. I prefer it on UbuntuEclipse with Ninjaas a construction system. The steps to build LLVM are: a) generate a build system usingTo make, b) is usedninjato build a huge LLVM code base. Use the following commands:

(Video) A Brief Introduction to LLVM

cd ~/llvm/mkdir buildcd buildcmake -G Eclipse\ CDT4\ -\ Ninja -DLLVM_TARGETS_TO_BUILD=host ../llvm/ninja

On my laptop (i7 CPU, 16GB RAM machine) it took a while to build28 minutes. If you're using a laptop, make sure you're connected and have a novel to read before you start building. After the initial build, incremental builds using ninja are very fast (usually within seconds).

Object files and binaries are saved in the~/llvm/buildmap. LLVM supports a wide variety of target architectures such as x86, ARM, Sparc, MIPS, etc. Building all the objectives will take a long time. Since you run only compiled programs on your machine, you only care about the architecture of your machine. Therefore, you can request that LLVM only support the architecture on your computer by installing it-DLLVM_TARGETS_TO_BUILD=hostflag. As a result, other target architectures were not built and the build time was drastically reduced.

It houses the binaries for the LLVM tools~/llvm/build/bin. These tools are used to perform LLVM passes. To make the tool easier to use, it's a good idea to add the binary path to bashPADvariable so you can call the tools from any directory. To do this, add the following line to the~/.bashrcfile at the end of the file.
izvoz PATH=$PATH:~/llvm/build/bin/
Restart the terminal for the change to take effect.

Step 3. (Optional) Use the Eclipse IDE for development

IDEs are very useful for navigating large codebases. When writing your pass, you should refer to the different classes and the different methods available in those classes. It would be useful to use an IDE for these tasks. Because we asked CMakeEclipse with NinjaThe build system, along with the ninja build files, also generated Eclipse project files that we can use to open the code base in Eclipse. To import an LLVM project into Eclipse, do the following:

  1. open solar eclipse
  2. File -> Import -> General -> Existing projects in Workspace
  3. Choose~/llvm/buildas the root directory.

Step 4. Write your first pass

IMPORTANT: To make sorting easier, your LLVM pass should be named withYOUR EIDUppercase. In this document I will useSG12345such as Eid. Be careful not to capitalize names other than your EID.

To write a new LLVM pass you need to create source files and tell CMake to create your source files while building LLVM. If done correctly, a shared object file (.so file) will be created for your pass. Follow the steps below.

  1. Create a directory to house your LLVM pass
  2. Update the CMake file in the Transforms folder
  3. Create a CMake file for your LLVM pass
  4. CPP file for your LLVM pass
  5. Build your first passage

Step 4.1: Create a folder to house your LLVM pass

(Video) 2019 LLVM Developers’ Meeting: E. Christopher & J. Doerfert “Introduction to LLVM”

Most of the LLVM transformation passes are present in the~/llvm/llvm/lib/Transforms/. Let's go through the first time here. Create a directory to house our source files.
mkdir ~/llvm/llvm/lib/Transforms/SG12345

Step 4.2 Update the CMake file in the Transforms folder

We need to tell CMake that we have created a new pass in a new directory. To do this, add the following line below~/llvm/llvm/lib/Transforms/CMakeLists.txtfile at the end of the file.


Step 4.3 Create a CMake file for your LLVM pass

Now we need to tell CMake how to build our pass and what it should be called. First, create the CMake file.

cd ~/llvm/llvm/lib/Transforms/SG12345touch CMakeLists.txt

Add the following toCMakeLists.txtduration

add_llvm_library(LLVMSG12345 MODULE SG12345.cpp PLUGIN_TOOL opt)

The command above says that only one source file is referenced in the passSG12345.cpp. If you add more source files, add the filenames here. The command also says that the file name must be .soLLVMSG12345. Other details are irrelevant.

(Video) LLVM in 100 Seconds

Step 4.4 CPP file for your LLVM pass

tik op SG12345.cpp
You can use our referenceSG12345.cppdurationhere. Order execution will start from this file.

Step 4.5 Build your first pass

Once you've implemented your pass, it's time to compile.

cd ~/llvm/buildninja

A .so file for your card will be created and placed into it~/llvm/build/lib/map. Your pass is now ready to run on some test programs

Step 5. Perform your first pass on the test input

You will need some test programs to run your LLVM pass. LLVM passes work on intermediate view (IR). Therefore, test programs must be converted from a high-level language to LLVM IR. Your pass can then be performed on the LLVM IR test program.

Step 5.1 Create a test program

(Video) Programming Language with LLVM [1/20] Introduction to LLVM IR and tools

cd ~/llvm/mkdir testcasescd testcasestouch test1.c

Use the following simple program liketest1.c.

#Including int to do() {printf("hello world\n");wind back 0;}arm Hello() {wind back;}

Step 5.2 Generate LLVM IR

clang -emit-llvm -S test1.c
From LLVM IR totest1.cis rejectedtest1.ll. The representation used to encode the IR is called LLVM bytecode. The .ll file is the LLVM bytecode for thetest1.c.

Step 5.3 Request your LLVM Pass

OchooseThe program is used to perform its LLVM pass on LLVM bytecode. The following command loads the LLVM traversal object file and calls the appropriate traversal. Using our benchmark LLVM pass, this command prints the function names in the test program to standard output.
opt -load ../build/lib/ -SG12345 < test1.ll > /dev/null

And that. Now you can go back and change your pass, test more complicated programs, read more documentation, or just ditch the compiler altogether. The choice is yours, and the world is your playground! :)


Introduction to LLVM? ›

LLVM can provide the middle layers of a complete compiler system, taking intermediate representation (IR) code from a compiler and emitting an optimized IR. This new IR can then be converted and linked into machine-dependent assembly language code for a target platform.

How do I get started with LLVM? ›

Follow the steps below.
  1. Create a directory to host your LLVM pass.
  2. Update the CMake file in the Transforms directory.
  3. Create a CMake file for your LLVM pass.
  4. CPP file for your LLVM pass.
  5. Compiling your first pass.

What is the overview of LLVM? ›

LLVM can provide the middle layers of a complete compiler system, taking intermediate representation (IR) code from a compiler and emitting an optimized IR. This new IR can then be converted and linked into machine-dependent assembly language code for a target platform.

What does LLVM stand for? ›

LLVM is an acronym that stands for low level virtual machine. It also refers to a compiling technology called the LLVM project, which is a collection of modular and reusable compiler and toolchain technologies.

Is LLVM better than GCC? ›

Advantages of GCC

GCC supports more traditional languages than Clang and LLVM, such as Ada, Fortran, and Go. GCC supports more less-popular architectures, and supported RISC-V earlier than Clang and LLVM. GCC supports more language extensions and more assembly language features than Clang and LLVM.

Is LLVM replacing GCC? ›

Long answer: Depends, it's a matter of adoption as well as other factors. For example, Apple uses LLVM (and Clang) for pretty much everything including building the kernel (previously built using GCC), bootloader and the userspace.

Which language is best for LLVM? ›

The typical way to work with LLVM is via code in a language you're comfortable with (and that has support for LLVM's libraries, of course). Two common language choices are C and C++. Many LLVM developers default to one of those two for several good reasons: LLVM itself is written in C++.

How many lines of code is LLVM? ›

In most GNU tools, the regression test suite is included with the main source. However for LLVM, the regression tests are a separate code base of 500 thousand lines.

Why is LLVM so popular? ›

Each library supports a particular component in a typical compiler pipeline (lexing, parsing, optimizations of a particular type, machine code generation for a particular architecture, etc.). What makes it so popular is that its modular design allows its functionality to be adapted and reused very easily.

Is LLVM like JVM? ›

The biggest difference between JVM bytecode and and LLVM bitcode is that JVM instructions are stack-oriented, whereas LLVM bitcode is not. This means that rather than loading values into registers, JVM bytecode loads values onto a stack and computes values from there.

Is LLVM a cross compiler? ›

On the other hand, Clang/LLVM is natively a cross-compiler, meaning that one set of programs can compile to all targets by setting the -target option.

Does Visual Studio use LLVM? ›

Clang/LLVM support for both CMake and MSBuild projects is available in Visual Studio 2019 and Visual Studio 2022.

Is LLVM slower than GCC? ›

I write some C code by myself and I compile them by LLVM or gcc. Nowadays, I compiled two different executable from two compiler (LLVM and gcc), and ran 100,000 times on both of them. I found that the performance of executable from gcc always better than LLVM one.

Does Nvidia use LLVM? ›

NVIDIA's CUDA Compiler (NVCC) is based on the widely used LLVM open source compiler infrastructure.

What are the different types in LLVM? ›

types in LLVM go into three groups: primitive types (integers, floats, etc.), derived types (structs, arrays, pointers, etc.), forward-declared types (opaque structs)

What is the competitor of LLVM? ›

If you're looking for LLVM alternative as a backend for your language, then there are several platforms - JVM, CLR, and also GCC's libgccjit.

Why did GCC drop Java? ›

Back in 2016, the GCC compiler dropped the Java/GCJ support over being unmaintained. The GCC Java support was removed over lack of maintenance and not seeing much attention given that there have been numerous free software Java implementations.

What is the difference between Clang and LLVM? ›

There are 2 different things here. LLVM is a backend compiler meant to build compilers on top of it. It deals with optimizations and production of code adapted to the target architecture. CLang is a front end which parses C, C++ and Objective C code and translates it into a representation suitable for LLVM.

Why use Clang instead of GCC? ›

Clang is much faster and uses far less memory than GCC. Clang aims to provide extremely clear and concise diagnostics (error and warning messages), and includes support for expressive diagnostics. GCC's warnings are sometimes acceptable, but are often confusing and it does not support expressive diagnostics.

Does Odin use LLVM? ›

Odin also relies on LLVM (for code generation) and an external linker.

Is LLVM stack based or register based? ›

LLVM is designed to support code generation for register machines, and this hin- ders the wide use of LLVM for stack machines.

Does LLVM run on Windows? ›

LLVM fully supports the COFF object file format, which is compatible with all other existing Windows toolchains.

Why does Rust use LLVM? ›

LLVM supports many architectures, and so does Rust; the most common are x86 (Intel/AMD 32-bit), x64 (Intel/AMD 64-bit), and ARM64. Rust has its own frontend that compiles down to LLVM IR and then relies on LLVM to optimize this and compile it down to machine code.

How long does it take to build a LLVM? ›

LLVM does take a long time to compile, but four hours is an exaggeration. A mid-range computer can compile it in 30 minutes with -j4, and a good computer can compile it in 10 with -j8. Core duo netbook with 4GB RAM and 5400 RPM hard disk. I can assure you it really does take 4 hours.

Is Clang owned by Apple? ›

Neither is owned by Apple - they're open source projects with contributors from Microsoft, Google, ARM, Intel, AMD and many others. Apple has publicly referred to the C/C++/Objective-C compiler that ships with Xcode as the "Apple LLVM Compiler", apparently a marketing name for their variant of Clang.

What language is rust written in? ›

Rust is a statically-typed programming language designed for performance and safety, especially safe concurrency and memory management. Its syntax is similar to that of C++. It is an open-source project developed originally at Mozilla Research.

What is the difference between user and use in LLVM? ›

The difference is that a user of Value has the Value as one of its operands. When you are calling uses you get a list of all Use instances holding a reference from the Value to each of the users of the particular Value . Calling users gives you a list of User directly.

Is rust compiler based on LLVM? ›

rustc uses LLVM for code generation.

What version of C++ does LLVM use? ›

Unless otherwise documented, LLVM subprojects are written using standard C++17 code and avoid unnecessary vendor-specific extensions.

Does LLVM use JIT? ›

The LLVM JIT compiler is function-based because it is able to compile a single function at a time. This defines the granularity at which the compiler works, which is an important decision of a JIT system.

Does C++ compile to LLVM? ›

GraalVM can execute C/C++, Rust, and other languages that can be compiled to LLVM bitcode. As the first step, you have to compile a program to LLVM bitcode using some LLVM compiler front end, for example, clang for C and C++, rust for the Rust programing language, etc.

What linker does LLVM use? ›

LLD is a linker from the LLVM project that is a drop-in replacement for system linkers and runs much faster than them. It also provides features that are useful for toolchain developers. The linker supports ELF (Unix), PE/COFF (Windows), Mach-O (macOS) and WebAssembly in descending order of completeness.

Does LLVM generate assembly? ›

The llc command compiles LLVM input into assembly language for a specified architecture. If we do not mention any architecture as in the preceding command, the assembly will be generated for the host machine where the llc command is being used.

Does LLVM have garbage collection? ›

All LLVM code generators support garbage collection, including the C backend. We hope that the primitive support built into LLVM is sufficient to support a broad class of garbage collected languages, including Scheme, ML, scripting languages, Java, C#, etc.

Does LLVM have an assembler? ›

llvm-as is the LLVM assembler. It reads a file containing human-readable LLVM assembly language, translates it to LLVM bitcode, and writes the result into a file or to standard output.

Which C compiler produces fastest code? ›

The Zapcc compiler is the fastest compiler in this test, handily beating the nearest competitor by a factor of more than 1.6x. The PGI compiler is the slowest compiler in the test. According to the Portland Group website, they are working on an LLVM-based update to the PGI compiler, which may improve the compile time.

Does LLVM include Clang? ›

Clang uses the LLVM compiler as its back end and it has been included in the release of the LLVM since the LLVM 2.6. Clang is also built to be a drop-in replacement for GCC command. In its design, the Clang compiler has been constructed to work very similarly to GCC to ensure that portability is maximized.

What languages use LLVM? ›

LLVM is written in C++ and is designed for compile-time, link-time, run-time, and idle-time optimization. Popular languages with compilers that use LLVM include Fortran, Haskell, Julia, Kotlin, Lua, Objective-C, OpenGL, Rust, and Swift.

What version of CUDA is LLVM? ›

CUDA is supported since llvm 3.9. Clang currently supports CUDA 7.0 through 12.1. If clang detects a newer CUDA version, it will issue a warning and will attempt to use detected CUDA SDK it as if it were CUDA 12.1. Before you build CUDA code, you'll need to have installed the CUDA SDK.

What language does NVIDIA use for AI? ›

NVIDIA AI Platform for Developers

GPU-accelerated deep learning frameworks offer flexibility to design and train custom deep neural networks and provide interfaces to commonly-used programming languages such as Python and C/C++.

How does LLVM backend work? ›

A LLVM backend compiles IR down to object or assembly code. Each backend targets a single architecture, but possibly multiple instruction sets. For example, LLVM has only one backend for the ARM architecture that can emit code for instruction sets like ARMv6 and ARMv7.

What is the difference between MLIR and LLVM? ›

An open ecosystem is the biggest difference from LLVM - in MLIR you can define your own operations and abstractions in the IR, suitable for the domain of problems you are trying to solve. It is more of a pure compiler *infrastructure* than LLVM is.

How to configure and build LLVM? ›

Quick start
  1. Download and install CMake. ...
  2. Open a shell. ...
  3. Create a build directory. ...
  4. Execute this command in the shell replacing path/to/llvm/source/root with the path to the root of your LLVM source tree: ...
  5. After CMake has finished running, proceed to use IDE project files, or start the build from the build directory:

How to setup LLVM on Windows? ›

How to set up Windows on Arm for LLVM development
  1. Adding to the PATH. ...
  2. Install Visual Studio Build Tools. ...
  3. Install MSVC Redist Libraries. ...
  4. Install the Latest LLVM for Windows on Arm. ...
  5. Install CMake. ...
  6. Install Python. ...
  7. Install Git. ...
  8. Build Ninja.
Feb 16, 2022

How do I generate C++ code in LLVM? ›

The easiest way to make LLVM generate C++ code is to go to and, under Output Options, choose LLVM C++ API code for the target.

How long does it take to build LLVM from source? ›

You can invoke CMake with -DCMAKE_BUILD_TYPE=Debug to produce a clang binary with assertions enabled. This takes only about 60 minutes to build, but you wouldn't want to use the resulting binary for anything heavy-duty because it's so slow.

Where does LLVM install Windows? ›

If you have built the LLVM in debug version all the executables (including llvm-config) have been placed in your build directory (containing Visual Studio project and solution files) in bin/Debug/ subdirectory.

Does LLVM support C? ›

Clang: a C language family frontend for LLVM. The Clang project provides a language front-end and tooling infrastructure for languages in the C language family (C, C++, Objective C/C++, OpenCL, CUDA, and RenderScript) for the LLVM project.

Does Visual C++ use LLVM? ›

Clang/LLVM support for both CMake and MSBuild projects is available in Visual Studio 2019 and Visual Studio 2022. You can use Visual Studio 2019 version 16.2 and later with Clang/LLVM to edit, build, and debug C++ Visual Studio projects (MSBuild) that target Windows or Linux.


1. Building domain-specific compilers quickly with MLIR compiler infrastructure | Chris Lattner
(Lex Clips)
2. Introduction to LLVM
(Jostein Kjønigsen)
3. Introduction to LLVM Building simple program analysis tools and instrumentation
4. 2019 LLVM Developers’ Meeting: J. Paquette & F. Hahn “Getting Started With LLVM: Basics”
5. How Compilers Work: Introduction to LLVM Passes - Andrzej Warzynski - C++ on Sea 2022
6. Tutorial1 Introduction to LLVM


Top Articles
Latest Posts
Article information

Author: Greg O'Connell

Last Updated: 30/07/2023

Views: 5911

Rating: 4.1 / 5 (62 voted)

Reviews: 85% of readers found this page helpful

Author information

Name: Greg O'Connell

Birthday: 1992-01-10

Address: Suite 517 2436 Jefferey Pass, Shanitaside, UT 27519

Phone: +2614651609714

Job: Education Developer

Hobby: Cooking, Gambling, Pottery, Shooting, Baseball, Singing, Snowboarding

Introduction: My name is Greg O'Connell, I am a delightful, colorful, talented, kind, lively, modern, tender person who loves writing and wants to share my knowledge and understanding with you.