Checklist

• If this is not a feature request but a general question, please start a discussion at https://github.com/sgl-project/sglang/discussions. Otherwise, it will be closed.
• Please use English. Otherwise, it will be closed.

Motivation

Pipeline Parallelism (PP) is currently disabled in Piecewise CUDA Graph, limiting scalability for large models. Regular CUDA Graph already supports PP, so we should implement the same support for piecewise CUDA graph.

Tasks

1. Remove PP check in can_run_piecewise_cuda_graph()
2. Add pp_proxy_tensors support in replay_prepare() and replay()
3. Handle PPProxyTensors output (follow CudaGraphRunner pattern)
   Related: [Feature] Roadmap for Prefill (Piecewise) CUDA Graph #11490

Related resources

No response