vllm的V1为什么删除了multi step特性

[lianjie]

V1 has lower end-to-end performance than V0(--num-scheduler-steps=8)

Testing Environment: vllm version: 0.8.5 Operating System: Ubuntu 22.04 GPU: Nvidia A100 Performance Comparison: For Llama/qwen2 series models, the end-to-end inference performance of V1 is 5% to 10% lower than that of V0. Version 0 configuration: --num-scheduler-steps=8 Question: Is there any community that has relevant performance comparison data?