16 ms·A collection of reproducible LLM inference engine benchmarks: SGLang vs. vLLM1 points by zhwu 1y ago