First problem, it would need to have to be open source, and from the information provided, only V2. It might work with OpenCL using Cycles and improve performance a bit. The new version 4.0 support more, but openGL is not there, and is NOT open source.
If Cycles can run your AMD card using OPENCL then it logically should be able to do the same, but there are lots of "if".
Lots of people would like to have EEVEE accelerated (Viewports and render) with the new RT functionality. Blender need to change it API to Vulkan to be able to achieve that. Once Vulkan is implemented there a way to class the "RT" (Real time trace) functions from theses card, as there is no way to access them from OpenGL.
So in the long term, we could have massive acceleration coming to Blender EEVEE if there is a Vulkan version and that the new API is used to make use the RT functions from theses cards (NVIDIA, AMD)
NOTE: I think the Blender devs have already implemented EMBREE from Intel that do the same kind of calculation. Still this lib relies on the CPU.
2 Comments
Hi, looking at this, and the quick answer is no. Here is the link to this: gpuopen.com/radeon-rays/
First problem, it would need to have to be open source, and from the information provided, only V2. It might work with OpenCL using Cycles and improve performance a bit. The new version 4.0 support more, but openGL is not there, and is NOT open source.
If Cycles can run your AMD card using OPENCL then it logically should be able to do the same, but there are lots of "if".
Lots of people would like to have EEVEE accelerated (Viewports and render) with the new RT functionality. Blender need to change it API to Vulkan to be able to achieve that. Once Vulkan is implemented there a way to class the "RT" (Real time trace) functions from theses card, as there is no way to access them from OpenGL.
So in the long term, we could have massive acceleration coming to Blender EEVEE if there is a Vulkan version and that the new API is used to make use the RT functions from theses cards (NVIDIA, AMD)
NOTE: I think the Blender devs have already implemented EMBREE from Intel that do the same kind of calculation. Still this lib relies on the CPU.
code.blender.org/2020/09/faster-motion-blur-with-intel-embree/
developer.blender.org/D3682
how many time you ask/propose same thing?