For upstreaming AMD CPU optimization in OneDNN, which is widely used in all framework like PyTorch, TF, vLLM - View it on GitHub