AI Frameworks
Get performance gains ranging up to 10x to 100x for popular deep-learning and machine-learning frameworks through drop-in Intel® optimizations.
AI frameworks provide data scientists, AI developers, and researchers the building blocks to architect, train, validate, and deploy models, through a high-level programming interface. All major frameworks for deep learning and classical machine learning have been optimized by using oneAPI libraries that provide optimal performance across Intel® CPUs and XPUs. These Intel® software optimizations help deliver orders of magnitude performance gains over stock implementations of the same frameworks. As a framework user, you can reap all performance and productivity benefits through drop-in acceleration without the need to learn new APIs or low-level foundational libraries.
Deep-Learning Frameworks
Intel® Optimization for TensorFlow*
TensorFlow* is a widely used deep-learning framework that's based on Python*. It's designed for flexible implementation and extensibility on modern deep neural networks.
Intel is collaborating with Google* to optimize its performance on platforms based on the Intel® Xeon® processor. The platforms use the Intel® oneAPI Deep Neural Network Library (oneDNN), an open-source, cross-platform performance library for deep-learning applications. These optimizations are directly upstreamed and made available in the official TensorFlow release via a simple flag update, which enables developers to seamlessly benefit from the Intel® optimizations.
The latest version of Intel® Optimization for TensorFlow* is included as part of the Intel® oneAPI AI Analytics Toolkit (AI Kit). This kit provides a comprehensive and interoperable set of AI software libraries to accelerate end-to-end data science and machine-learning workflows.
Documentation
Intel® Optimization for PyTorch*
The PyTorch* for Python* package provides one of the fastest implementations of dynamic neural networks to achieve speed and flexibility. Intel and Facebook* extensively collaborated to:
- Include many Intel optimizations in this popular framework
- Provide superior PyTorch performance on Intel® architectures, most notably Intel® Xeon® Scalable processors
The optimizations are built using oneDNN to provide cross-platform support and acceleration.
Intel also provides Intel® Extension for PyTorch* for more capabilities that have not yet been upstreamed, including:
- Support for automatic mixed precision
- Customized operators
- Fusion patterns
This optimization adds bindings with Intel® oneAPI Collective Communications Library (oneCCL) for efficient distributed training and is a consolidated package. It provides the best out-of-box experience to get all of the performance benefits from PyTorch. The package has the latest versions of:
- Stock PyTorch with Intel® optimizations
- Intel Extension for PyTorch
- oneCCL
Intel® Optimization for Pytorch* is made available as part of the AI Kit that provides a comprehensive and interoperable set of AI software libraries to accelerate end-to-end data science and machine-learning workflows.
Apache MXNet*
This open-source, deep-learning framework is highly portable, lightweight, and designed to offer efficiency and flexibility through imperative and symbolic programming. MXNet* includes built-in support for Intel optimizations to achieve high performance on Intel Xeon Scalable processors.
Documentation
PaddlePaddle*
This open-source, deep-learning Python* framework from Baidu* is known for user-friendly, scalable operations. Built using oneDNN, this popular framework provides fast performance on Intel Xeon Scalable processors and a large collection of tools to help AI developers.
Machine-Learning Frameworks
Intel® Extension for Scikit-learn*
Scikit-learn* is one of the most widely used Python packages for data science and machine learning. Intel provides a seamless way to speed up the many algorithms of scikit-learn on Intel® CPUs and GPUs through the Intel® Extension for Scikit-learn*. This extension package dynamically patches scikit-learn estimators to use Intel® oneAPI Data Analytics Library (oneDAL) as the underlying solver. It achieves the speedup for machine learning algorithms on Intel architectures, both single and multi-nodes.
The latest version of Intel Extension for Scikit-learn is also included as part of the AI Kit. It provides a comprehensive and interoperable set of AI software libraries to accelerate end-to-end data science and machine-learning workflows.
XGBoost Optimized by Intel
This is a well-known machine-learning package for gradient-boosted decision trees. It includes seamless, drop-in acceleration for Intel architectures to significantly speed up model training and improve accuracy for better predictions. In collaboration with XGBoost community, Intel has been directly upstreaming many optimizations to provide superior performance on Intel CPUs.
The latest version of XGBoost that Intel optimizes is included as part of the AI Kit. It provides a comprehensive and interoperable set of AI software libraries to accelerate end-to-end data science and machine-learning workflows.