How come AMD still doesn't have support for Octane? https://render.otoy.com/forum/viewtopic.php?f=9&t=70524
Does Octane need to have OpenCL library implemented?
We stopped AMD CUDA development around 3.06.2 (although we yet release AMD OctaneBench 3.06.2 for RNDR node testing). Currently we are porting Octane over to a new framework (RNDR SDK) that cross compiles to Metal/Vulkan/CUDA to support all GPUs on all platforms (even iOS). Octane 2019.2 and later will ship with significant portions of the engine moved to this new framework, but you will still need a CUDA device (even if it is a network slave) to run the other 60% of Octane still in CUDA. At some point we'd like to port all of Octane over to the new framework, but until then, 2019.2+ can start to leverage AMD/Intel/Apple devices to offload some portions of the rendering work. So far, what we have ported from CUDA to the new framework seems to be faster, even on the same NVIDIA GPU (comparing CUDA vs. VK for example).
Would GCN1 card like 7870HD be able to run Octane then ?
Thank you for your queries. I think Octane team already mentioned couple of important points in their reply:
As this is a third-party software, I guess they can provide you more information regarding your queries.
By the way, we have a separate ProRender community, you may like to check it as well.
On Mac there will be Octane support. AMD on PC will probably be next. But it is not sure which GPUs will have support.
Octane X for Metal is now nearly feature complete and able to load and render all production scenes from our test suite at pixel perfect parity with the latest Octane 2020 builds. This includes all OSL shaders, AI denoising, random walk SSS, net render, perfect linear multi-GPU scaling on the new Mac Pro