# taskflow/taskflow

A General-purpose Task-parallel Programming System in C++

Repository: https://github.com/taskflow/taskflow
Canonical: https://ross.abutalabs.com/products/taskflow
Homepage: https://taskflow.github.io
Language: C++
License: NOASSERTION
License Family: other
Topics: parallel-programming, threadpool, concurrent-programming, high-performance-computing, multicore-programming, multi-threading, taskparallelism, multithreading, parallel-computing, work-stealing, gpu-programming, heterogeneous-parallel-programming, cuda-programming, parallel, taskflow
Last push: 2026-08-06T17:29:28+00:00

## Health v2 (maintenance only)
Score: 86/100 (v2, computed 2026-09-03T02:39:23.370411+00:00)
- activity 96, release rhythm 65, longevity 100
- inputs: {"age_days": 3059, "days_push": 27, "days_rel": 75, "gap_med": 118, "n_releases_24m": 6}
- flags: no_license
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 12126, forks 1403 (observed 2026-08-28T04:10:51.795445+00:00)

## What it is
Taskflow is a header-only C++ library for building task-parallel programs using task dependency graphs, with a work-stealing executor for CPUs and GPUs (CUDA). It includes a web-based profiler (TFProf) for visualizing and profiling taskflow executions.

## Use cases
- parallelize a C++ application across CPU cores with a task graph
- express task dependencies like fork-join and pipeline patterns
- run recursive parallel algorithms with subflow graphs
- accelerate workloads on NVIDIA GPUs with CUDA tasking
- profile and visualize parallel task execution
- scale machine learning or simulation workloads to millions of tasks

## When to choose
- you need high-performance task parallelism in modern C++ with minimal setup
- you want a header-only library with no installation burden
- you need both CPU and GPU (CUDA) heterogeneous parallelism
- you want built-in profiling and visualization of task graphs

## When to avoid
- you need distributed multi-node parallelism rather than a single machine
- your project is not C++ or targets an older C++ standard
- you only need simple thread pools without task dependency graphs

## Facets
- artifact type: library
- maturity: stable
- function: concurrency, developer-tools, benchmarking, monitoring
- domain: developer-tools, performance, gpu-computing, machine-learning
- platform: cpp, windows, cross-platform
- tags: task-parallelism, work-stealing, thread-pool, header-only, cuda, heterogeneous-computing, task-graph, profiler, linux, macos, gpu

## Member repositories
- taskflow/taskflow (main) score 86

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:10:51.795445+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-29T17:14:55.287468+00:00, confidence not recorded.
  - readme: https://github.com/taskflow/taskflow (fetched 2026-08-28T04:10:51.795445+00:00, sha b54d802e8d84)
  - homepage: https://taskflow.github.io (fetched 2026-08-29T08:12:07.849230+00:00, sha f78872414a81)
- Data as of 2026-08-30T08:39:29.467469+00:00.