WAMR — WebAssembly Micro Runtime for Embedded and IoT

Introduction

WAMR is a compact WebAssembly runtime developed under the Bytecode Alliance. It brings the portability and sandboxing of WebAssembly to environments where Wasmtime or V8 would be too large, including microcontrollers, RTOS-based systems, and edge computing platforms.

What WAMR Does

• Executes WebAssembly modules on devices with as little as 50KB RAM
• Offers three execution modes: interpreter, ahead-of-time (AOT) compilation, and JIT
• Provides WASI (WebAssembly System Interface) support for file and socket access
• Supports multi-threading with the WASM threads proposal
• Embeds into C/C++ host applications via a clean native API

Architecture Overview

WAMR includes a stack-based interpreter for minimal footprint, an AOT compiler that translates WASM to native code offline, and an optional LLVM-based JIT for runtime compilation. The runtime manages linear memory, tables, and module instances inside a sandboxed execution environment. A built-in app management framework supports dynamic loading and lifecycle control of WASM apps.

Self-Hosting & Configuration

• Build with CMake; select execution mode via cmake flags (-DWAMR_BUILD_INTERP, -DWAMR_BUILD_AOT)
• Cross-compile for Zephyr, NuttX, RT-Thread, or bare-metal ARM targets
• Use wamrc to ahead-of-time compile WASM to native for production deployment
• Enable WASI for filesystem and network access in supported environments
• Embed in a host app by linking libiwasm and calling wasm_runtime_* APIs

Key Features

• Runs on devices with as little as 50KB RAM (interpreter mode)
• Three execution modes cover the spectrum from MCU to server
• WASI support for portable system access
• Dynamic module loading and unloading at runtime
• Hardware-sandboxed execution isolates untrusted code

Comparison with Similar Tools

• Wasmtime — full-featured Bytecode Alliance runtime; WAMR targets smaller footprint
• Wasmer — developer-friendly with package registry; WAMR focuses on embedded constraints
• wasm3 — pure interpreter with tiny footprint; WAMR adds AOT and JIT for higher throughput
• WasmEdge — cloud-native WASM runtime; WAMR targets deeper embedded use cases

FAQ

Q: What is the minimum hardware for WAMR? A: The interpreter runs on Cortex-M4 with 50KB RAM. AOT mode needs more flash but runs faster.

Q: Can I use WAMR in a cloud environment? A: Yes. WAMR runs on Linux, macOS, and Windows and is used in edge computing and serverless platforms.

Q: What languages can I compile to WASM for WAMR? A: C, C++, Rust, Go (TinyGo), AssemblyScript, and any language with a WASM compilation target.

Q: Is WAMR production-ready? A: Yes. It is used in production by Intel, Xiaomi, Midea, and other companies for IoT and edge workloads.

Sources

• https://github.com/bytecodealliance/wasm-micro-runtime
• https://bytecodealliance.org/