arrayfire ☆
« Back to VersTracker
Description:
General purpose GPU library
Type: Formula  |  Latest Version: 3.10.0@2  |  Tracked Since: Dec 17, 2025
Links: Homepage  |  @ArrayFire  |  formulae.brew.sh
Category: Ai ml
Tags: gpu parallel-computing hpc cuda opencl math
Install: brew install arrayfire
About:
ArrayFire is a high-performance library for parallel computing that abstracts the complexity of writing GPU-accelerated code. It provides a unified interface that allows developers to target CPUs, GPUs, and other accelerators without rewriting source code. The library optimizes matrix operations, signal processing, and computer vision tasks for maximum throughput.
Key Features:
  • Unified API for CPU, CUDA, OpenCL, and oneAPI backends
  • Just-in-time compilation for optimized kernel fusion
  • Built-in computer vision and image processing functions
  • Support for sparse and dense linear algebra operations
Use Cases:
  • Accelerating deep learning model inference and training
  • Real-time computer vision and image processing pipelines
  • Large-scale scientific computing and data analysis
Alternatives:
  • OpenCL – ArrayFire provides higher-level abstractions than raw OpenCL APIs
  • cuBLAS/cuDNN – NVIDIA-specific libraries vs ArrayFire's cross-platform support
License: BSD-3-Clause
Dependencies: clblast, fftw, fmt, openblas, spdlog, opencl-icd-loader, pocl
Bottles available for: arm64_tahoe, arm64_sequoia, arm64_sonoma, sonoma, arm64_linux, x86_64_linux
Version History
Detected Version Rev Change Commit
Oct 29, 2025 8:34pm 1 VERSION_BUMP da1457bc
Oct 12, 2025 2:41am 1 VERSION_BUMP 204e9b51
Nov 18, 2024 3:37pm 4 VERSION_BUMP d1b63a8f
« Back to VersTracker  |  Browse Packages  |  About