# arrayfire/arrayfire

ArrayFire: a general purpose GPU library.

Repository: https://github.com/arrayfire/arrayfire
Canonical: https://ross.abutalabs.com/products/arrayfire
Homepage: https://arrayfire.com
Language: C++
License: BSD-3-Clause
License Family: permissive
Topics: arrayfire, c-plus-plus, gpu, cuda, opencl, gpgpu, hpc, performance, scientific-computing, cpp, c
Last push: 2026-03-07T03:06:52+00:00

## Health v2 (maintenance only)
Score: 57/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 71, release rhythm 14, longevity 100
- inputs: {"age_days": 4327, "days_push": 179, "days_rel": 362, "gap_med": null, "n_releases_24m": 1}
- flags: none
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 4902, forks 558 (observed 2026-08-28T04:09:02.138935+00:00)

## What it is
ArrayFire is a general-purpose tensor/numerical computing library for C, C++, and Python that accelerates array operations on GPUs (CUDA, OpenCL, oneAPI) and CPUs. It provides hundreds of functions for linear algebra, image processing, computer vision, signal processing, statistics, and machine learning behind a simple af::array abstraction.

## Use cases
- accelerate numerical array computations on GPU
- run linear algebra on CUDA or OpenCL devices
- GPU-accelerated image processing in C++
- port scientific computing code across CPU and GPU backends
- build high-performance machine learning primitives
- write cross-platform GPGPU code without writing kernels

## When to choose
- you need GPU-accelerated tensor math in C/C++ without writing CUDA kernels
- you want one API that targets CUDA, OpenCL, oneAPI, and CPU backends
- you need fast linear algebra, FFT, image, or signal processing routines
- you are doing HPC or scientific computing on NVIDIA, AMD, or Intel GPUs

## When to avoid
- you need a Python-first deep learning framework with autograd and neural network layers
- you want a modern actively evolving ecosystem - development activity has slowed
- your workload is small and CPU-only, where GPU overhead outweighs gains
- you need distributed multi-node computing rather than single-device acceleration

## Facets
- artifact type: library
- maturity: stable
- function: machine-learning, image-processing, computer-vision, gpu-computing, math, data-science
- domain: gpu-computing, machine-learning, image-processing, computer-vision, performance
- platform: cpp, c, windows, cross-platform
- tags: tensor-library, gpgpu, opencl, oneapi, linear-algebra, signal-processing, hpc, scientific-computing, algorithms, cuda, linux, macos, gpu

## Member repositories
- arrayfire/arrayfire (main) score 57

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:09:02.138935+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 2026-08-29T18:18:08.370114+00:00, confidence not recorded.
  - readme: https://github.com/arrayfire/arrayfire (fetched 2026-08-28T04:09:02.138935+00:00, sha 2c1c66858238)
  - homepage: https://arrayfire.com (fetched 2026-08-29T09:00:18.004052+00:00, sha b298f69a2793)
- Data as of 2026-08-30T08:39:29.467469+00:00.