About this project

Intel Video Processing Library (Intel VPL) is a programming interface for accessing hardware-accelerated video decode, encode, and processing capabilities on Intel GPUs. The project is intended for workloads such as AI visual inference, media delivery, cloud gaming, and virtual desktop infrastructure. This repository contains the VPL API header files, the VPL Dispatcher, and examples showing how to use the API. It does not provide a complete hardware implementation by itself. Users must install at least one runtime: Intel VPL GPU Runtime for Intel Iris Xe graphics and newer platforms, or Intel Media SDK for legacy Intel graphics. The dispatcher loads the selected runtime and forwards application function calls to it. The documented platform support ranges from older Broadwell-era systems through current and future Intel graphics platforms. Older platforms generally use Intel Media SDK with DirectX 9/11 support, while newer platforms use the VPL GPU Runtime, typically with DirectX 11. On Tiger Lake and DG1, when both runtimes are installed, the dispatcher prefers the VPL runtime unless the application explicitly requests Direct3D 9 through the acceleration-mode filter. Intel VPL can be built from source or installed through Linux packages where available. Setup scripts are provided for Linux and Windows to help CMake and pkg-config locate the library and headers. The repository documents CMake integration through find_package(VPL) and VPL::dispatcher, as well as direct compilation with pkg-config. VPL Tools are no longer included here and are maintained separately in the libvpl-tools repository. The project is released under the MIT License. The API options are treated as the stable public interface, while debug parameters, environment variables, and internal data structures may change.