intel/compute-runtime

Intel® Graphics Compute Runtime for oneAPI Level Zero and OpenCL™ Driver

What it solves

This project provides the essential compute API support for Intel graphics hardware, allowing developers to run compute-intensive workloads on Intel GPUs (such as HD Graphics and Xe architectures) using standardized interfaces.

How it works

Known as NEO, the runtime acts as a driver layer that implements the oneAPI Level Zero and OpenCL APIs. It bridges the gap between high-level compute applications and the Intel GPU hardware, enabling hardware acceleration for general-purpose computing tasks.

Who it’s for

It is designed for developers building high-performance compute applications that target Intel graphics hardware across a wide range of platforms, from legacy Gen8 devices to modern architectures like Lunar Lake and Panther Lake.

Highlights

  • Dual API Support: Implements both oneAPI Level Zero and OpenCL standards.
  • Broad Hardware Compatibility: Supports a vast array of Intel GPU architectures, including integrated and discrete graphics.
  • Cross-Platform Availability: Available for Linux distributions and Windows Subsystem for Linux (WSL).
  • Open Source: Distributed under the MIT License for transparency and and community contribution.

Related

  • Project
  • Project
  • Project
  • Project
  • Project