Intel oneAPI GPU Rendering Appears Ready For Blender 3.3
As I've written about before, for months Intel engineers with the Blender community have been working on oneAPI GPU rendering support for the Cycles engine to complement the NVIDIA CUDA/OptiX and AMD HIP targets. Since Blender 3.0 dropped OpenCL, this oneAPI target will be the only option for GPU acceleration with Intel discrete GPUs for this 3D modeling software.
The oneAPI code for Blender has now been audtied/approved after going through a few rounds of review in recent months.
The work-in-progress Blender 3.3 release notes mention this Intel GPU rendering support. It does note that only Intel Arc GPUs are supported for these current and future Intel dGPUs -- unfortunately, it doesn't work out having Blender GPU acceleration for existing Intel integrated graphics.
At the moment this oneAPI GPU rendering support is Windows-only but Intel is working on supporting Linux in time for the final Blender 3.3 release. The Intel Compute Runtime 22.10.22597 is the version expected for Linux users to be ready for Blender 3.3.
The kernels are also compiled at first-run which can be quite slow but they too hope to have this fixed for Blender 3.3 final, "Kernels are compiled when rendering for the first time. Currently this can be slow, for example 15 minutes. We are working to bundle compiled kernels with Blender for the final release instead."
It's great seeing this get squared away for Blender 3.3 and in time for the upcoming Arc Graphics desktop GPUs launching this year. Blender 3.3 stable is expected to be released in early September.