Taskflow
A General-purpose Task-parallel Programming System
A General-purpose Task-parallel Programming System
Taskflow enables developers to express a broad range of computational patterns using task graph-based approaches with minimal programming effort.
Taskflow offers parallel algorithm primitives for quickly expressing common parallel algorithm patterns.
Taskflow develops a highly efficient system runtime optimized for latency, energy efficiency, and throughput.
Taskflow has demonstrated promising performance in large parallel applications with millions of CPU and GPU tasks.
We have successfully applied Taskflow to assist developers in the implementation and deployment of parallel CAD algorithms and software that scale to millions of tasks on manycore CPUs and GPUs.
Explore Details »We have successfully applied Taskflow to accelerate the simulation of quantum circuits in both static and dynamic environments using scalable task parallelism.
Explore Details »We have successfully applied Taskflow to design efficient model- and data-parallel algorithms for scaling up large-scale machine learning workloads that incorporate billions of parameters.
Explore Details »Taskflow is open-source on GitHub. Check out the code from https://github.com/taskflow/taskflow or download the latest releases.
Taskflow handbook is the primary resource for understanding features, concepts, and application programming interface (API).
Taskflow profiler provides the visualization and tooling you need to profile Taskflow programs on the web.
Taskflow showcase presentation gives you a quick head start to understand the project motivation and important features.
Taskflow issue tracker lets you report bugs, request new features, or contribute to the project by fixing open issues.
Taskflow mailing list keeps you in contact with the project community and stay up-to-date with newest features.
Please refer to and cite the following paper if you are using Taskflow in your scientific computing projects:
Schedule a meeting and benefit from our experience in parallel and heterogeneous computing