4 ms·LLM Inference with VLLM and AMD: Achieving LLM Inference Parity with Nvidia2 points by pstan 3y ago