marl
A hybrid thread/fiber task scheduler for C++11 that enables efficient execution of blocking tasks.
What is marl?
Marl is a hybrid thread and fiber task scheduler written in C++11 that enables efficient execution of tasks across multiple threads. It solves the problem of blocking tasks by using fibers to yield execution, allowing other tasks to run while waiting, thus maximizing hardware thread utilization.
C++ developers building high-performance applications that require efficient task scheduling, such as game engines, servers, or real-time systems needing to handle blocking operations without wasting threads.
Developers choose Marl for its hybrid approach that combines threads and fibers, providing efficient blocking task handling with a simple API, cross-platform support, and no external dependencies, making it a lightweight yet powerful concurrency solution.
Overview
A hybrid thread / fiber task scheduler written in C++ 11
Use Cases
Best For
- Implementing efficient task-based parallelism in C++ applications
- Handling blocking I/O operations without wasting thread resources
- Building cross-platform concurrent systems with consistent behavior
- Optimizing thread utilization in game engines or real-time systems
- Simplifying concurrency patterns with a fluent scheduling API
- Managing worker pools with built-in synchronization primitives
Not Ideal For
- Real-time systems requiring deterministic task scheduling and strict timing guarantees
- Projects heavily reliant on standard C++ threading primitives like std::thread without willingness to refactor
- Small applications with minimal concurrency needs where setup overhead outweighs benefits
- Environments where frequent external blocking calls cannot be easily avoided or wrapped
Pros & Cons
Pros
Combines threads and fibers to yield on blocking tasks, preventing thread wastage and maximizing hardware utilization as highlighted in the key features.
Runs on multiple OSes including Windows, macOS, Linux, and mobile platforms across various architectures, as listed in the README's supported systems.
Self-contained library with no required external dependencies for core functionality, simplifying integration, though unit tests need googletest.
Provides an intuitive interface with marl::schedule() and defer() for minimal boilerplate, demonstrated in the example code for task management.
Cons
Requires precise binding and unbinding of the scheduler to threads to avoid undefined behavior or destructor blocking, as warned in usage recommendations.
Cannot safely use standard blocking functions like std::mutex::lock in tasks without risking deadlocks, necessitating workarounds like marl::blocking_call.
Recommends a single scheduler instance per process lifetime, which can be limiting for modular or dynamic application architectures.
Frequently Asked Questions
Related Projects
concurrentqueueA fast multi-producer, multi-consumer lock-free concurrent queue for C++11
TaskflowA General-purpose Task-parallel Programming System in C++
ThreadPoolA simple C++11 Thread Pool implementation
ArrayFireArrayFire: a general purpose GPU library.
Found a gem we're missing?
Open-Awesome is built by the community, for the community. Submit a project, suggest an awesome list, or help improve the catalog on GitHub.