Intel® Distribution of OpenVINO™ Toolkit
What's New in Version 2022.2
The latest version (2022.2) of this toolkit makes it easier for developers everywhere to start innovating. Build with a cleaner API and more integrations. Optimize with broader model and hardware support. Deploy with improved portability and performance.
Note This is a standard release intended for developers who prefer the very latest features and leading performance. Standard releases will continue to be made available three to four times a year. Long Term Support (LTS) releases are also available. A new LTS version is released every year and is supported for two years (one year of bug fixes, and two years for security patches).
Broader Support
Optimize and deploy with ease across an expanded range of deep learning models including natural language processing (NLP). Access AI acceleration across an expanded range of hardware.
- New initial support for 13th generation Intel® Core™ desktop processor (formerly code named Raptor Lake).
- New preview support for Intel’s discrete graphics cards for deep learning inferencing workloads in intelligent cloud, edge, and media analytics workloads, Intel® Data Center GPU Flex Series and Intel® Arc™ GPU, enabling hundreds of models to work on the GPU.
- New ability to test your model performance with preview support for 4th generation Intel® Xeon® processors (formerly code named Sapphire Rapids).
- Get reduced memory consumption when using dynamic input shapes on CPUs. This improves efficiency for NLP applications.
Cleaner API and More Integrations
It’s simpler to adopt and maintain your code. The new version requires fewer code changes, provides more options when integrating with frameworks, and minimizes conversions.
- More performance optimization choices for ONNX* (Open Neural Network Exchange) runtime developers through the OpenVINO™ Execution Provider. It’s now easier to add the Intel Distribution of OpenVINO toolkit with minimal code changes.
- Model inferencing in PyTorch* is accelerated with the new OpenVINO™ Integration with Torch-ORT. PyTorch developers can more seamlessly integrate with OpenVINO and get performance gains with less code changes.
- More deep learning models get support and improved inferencing performance in the OpenVINO™ integration with TensorFlow*.
- OpenVINO API 2.0 was introduced in 2022.1. Old inference engine and nGraph APIs are available. For more information, see the 2022.1 Release Notes and Transition Guide.
Note These framework integrations aren't included in the install packages. The products are for users who haven't installed the standard OpenVINO toolkit. For more information, visit the respective links.
Portability and Performance
See a performance boost quickly with automatic device discovery, load balancing, and dynamic inferencing parallelism across CPU, GPU, and more.
- New cumulative throughput and a selection of compute mode added to the AUTO functionality. This enables multiple accelerators (like multiple GPUs) to be used at once to maximize inferencing performance, adding to the AUTO base functionality of automatic discovery an configuration.
- Expert optimization is built into the toolkit. Increase device performance with an autobatching functionality to automatically tune and customize the proper throughput settings for your system configuration and deep learning model. Get scalable parallelism and optimized memory usage.
Get Started
Optimize, fine-tune, and run comprehensive AI inference using the included model optimizer and runtime and development tools.
Sign Up for Exclusive News, Tips & Releases
Be among the first to learn about everything new with the Intel® Distribution of OpenVINO™ toolkit. By signing up, you get:
• Early Access product updates and releases
• Exclusive invitations to webinars and events
• Training and tutorial resources
• Contest announcements
• Other breaking news