Blender's "Cycles X" Showing Nice Performance But Dropping OpenCL Support

[-MacNuke-]

OpenCL was killed by AMD because their compiler was a buggy piece of ####, complex kernels just didn't work, so why use opencl if it's not going to work on amd anyway?

Imagine being AMD at that time. Your direct competitor has like 80% marked share and does not support a specific API version people would need for basic needs. Yeah sure. You put ALL your resources into THAT API because you think when you will support it everyone will ditch the marked leader and move to your GPU.

Yeah... sure...

[Nille_kungen]

They could have written "We only care about nvidia performance and design for it from scratch"
Don't expect others to ever be treated as first class citizens.

[tildearrow]

OpenCL was killed by AMD because their compiler was a buggy piece of ####, complex kernels just didn't work, so why use opencl if it's not going to work on amd anyway?

OpenCL was killed by Apple because they decided to deprecate both CL and GL in macOS Mojave, plus it was heavy and hard to set up...

[blacknova]

Disappointing to say the least. Why are the spending so much resources on proprietary solutions like CUDA and Optix ?

Cause it just works? Most people prefer predictable results after all.

[ms178]

Wasn't Vulkan compute less advanced than OpenCL last time I read about it?! I haven't heard about any exciting news there in a long time. Would the compute capabilities found there be enough for an advanced renderer as Cycles? And what says this decision about the state of OpenCL compute in general?! To me it seems that it is not the cross-vendor compute API of choice anymore, at least for developers. OneAPI (which is basically a SYCL implementation with extras) could be the next thing (when hipSYCL on AMD is fully mature and the SYCL implementation for Nvidia was already mentioned in this thread). AMD seems to back ROCm still officially, I wonder if they will revise this sooner or later - with Intel opening up oneAPI and all the ressources behind it, that might be more attractive over the long term.

[paupav]

I'm guessing that decision was made due to the tight integration of Blender project with Nvidia.

[Sdar]

OpenCL was killed by Nvidia because they wanted people to use their CUDA.

OpenCL was killed by AMD because their compiler was a buggy piece of ####, complex kernels just didn't work, so why use opencl if it's not going to work on amd anyway?

[Grinch]

Disappointing to say the least. Why are the spending so much resources on proprietary solutions like CUDA and Optix ?

They should be going all in on Vulkan rendering, which is being supported by all GPU vendors and is a open standard. I have a NVidia graphics card, and I use CUDA to great effect, but I don't want to be locked into NVidia, which is what Blender is actively supporting.

[-MacNuke-]

And to think coder and others berated me when I rightfully said on another thread that the market had rejected OpenCL for OneAPI and CUDA

Well... Nvidia did not support OpenCL 1.2 for a very long time. How should OpenCL have gained any ground when the GPU marked leader did not support the version of OpenCL that made sense for most applications?

OpenCL was killed by Nvidia because they wanted people to use their CUDA.

[tildearrow]

It's hard to blame then starting with CUDA when it's the only one that already works.

The Vulkan implementation is going to have to be started from scratch. I would also start with what's already available.

...at the cost of pushing the non-greedy standard out of the market? No thanks....

Unless they start on a Vulkan renderer or a CUDA on AMD wrapper I am not interested.