Intel Compute Runtime 24.13.29138.7 Brings Improved OpenCL/OpenGL Sharing

As the first new release to Intel's open-source Compute Runtime stack in about one month for this OpenCL and Level Zero compute support, Intel Compute Runtime 24.13.29138.7 was released this morning with much improved OpenCL/OpenGL sharing and interoperability on Linux, out-of-the-box support for the Xe kernel graphics driver, new optimizations, and many other changes.

Intel Compute Runtime 24.13.29138.7 was released a few minutes ago for further enhancing Intel's open-source GPU compute capabilities. This user-space driver stack continues to support Skylake and newer in a production capacity with OpenCL 3.0 and Level Zero 1.3 support. In digging through all the commits in this new version, some of the highlights include:

- Proper reporting of the Intel Arc A310 LP Graphics rather than being reported just as an A310 (non-LP).

- Adding the new device ID for the Intel Data Center GPU Flex 170V graphics card (PCI device ID 0x56C2).

- Support for ze_kernel_max_group_size_ext_properties_t to report the kernel max group size.

- Support for attaching metadata for vm_bind in the Xe driver.

- Enabling command buffer pre-allocation per command queue on Xe graphics and later.

- Adding several missing Alder Lake N device IDs.

- Improve Linux OpenCL and OpenGL sharing support. The existing OpenCL/OpenGL sharing for Intel graphics on Linux hasn't been too practical due to bugs and only supporting a few texture formats. The new code supports the official interoperability extension from Mesa for GLX and EGL contexts and can share buffers and just not textures as well as supporting more formats. DaVinci Resolve and other software have been tested with this improved OpenCL/OpenGL sharing support.

- Xe DRM driver detection is enabled by default for using this alternative kernel graphics driver to i915 on Linux 6.8+ kernels.

- Enabling Link-Time Optimization (LTO) for supported compiler configurations.

- Enabling localPreferred and other optimizations for DG2/Alchemist GPUs.

- Initial support for synchronized dispatch.

- A fix for Linux 6.8 kernel support.

Downloads and more details on today's Intel Compute Runtime stack update via GitHub.