It was designed in #915 (read just the OP and the linked PRs at the end) and the implementation pretty much follows it closely, at least for the Metal backend.

It was designed in #915 (read just the OP and the linked PRs at the end) and the implementation pretty much follows it closely, at least for the Metal backend. The CUDA and OpenCL backends are currently slightly coupled in ggml as they started developing before #915, but I think we'll resolve this eventually. #915 - https://github.com/ggerganov/llama.cpp/discussions/915 https://github.com/ggerganov/llama.cpp/discussions/915