onednn ☆
« Back to VersTracker
Description:
Basic building blocks for deep learning applications
Type: Formula  |  Latest Version: 3.10.2@0  |  Tracked Since: Dec 2, 2025
Links: Homepage  |  formulae.brew.sh
Category: Ai ml
Tags: ai ml library intel deep-learning performance
Install: brew install onednn
About:
oneDNN (Intel oneAPI Deep Neural Network Library) is an open-source performance library for deep learning applications. It provides highly optimized building blocks for convolution, matrix multiplication, and other primitives for CPUs and GPUs. Its main value is accelerating AI frameworks like TensorFlow and PyTorch on Intel hardware.
Key Features:
  • Optimized primitives for CPUs and GPUs
  • Support for INT8 and BF16 data types
  • Graph API for complex operations
  • Cross-platform support
Use Cases:
  • Accelerating deep learning inference and training
  • Optimizing AI models for Intel hardware
  • Integrating high-performance kernels into custom frameworks
Alternatives:
  • OpenBLAS – Focuses on linear algebra (BLAS) rather than deep learning primitives
  • cuDNN – NVIDIA's equivalent, but locked to NVIDIA GPUs
Version History
Detected Version Rev Change Commit
Dec 2, 2025 6:46pm 0 VERSION_BUMP 9d040f7c
Dec 2, 2025 5:29pm 0 VERSION_BUMP 2a67b791
Nov 7, 2025 8:56pm 0 VERSION_BUMP 06ecd964
Oct 15, 2025 11:20pm 0 VERSION_BUMP 7bf48c2e
Sep 16, 2025 9:16am 0 VERSION_BUMP 8968c4b7
Oct 26, 2024 9:34pm 0 VERSION_BUMP 8fb52536
« Back to VersTracker  |  Browse Packages  |  About