Why can't get cuda kernel for vllm in nsight system

Developer Tools Nsight Systems Profiling Linux Targets
1

vLLM code

# SPDX-License-Identifier: Apache-2.0
import torch
from vllm import LLM, SamplingParams

if __name__ == "__main__":
    # Sample prompts.
    prompts = [
        "Hello, my name is",
        "The president of the United States is",
        "The capital of France is",
        "The future of AI is",
    ]
    # Create a sampling params object.
    sampling_params = SamplingParams(temperature=0.8, top_p=0.95)

    # Create an LLM.
    llm = LLM(model="/root/xiaobin/Qwen3-8B")
    # Generate texts from the prompts. The output is a list of RequestOutput objects
    # that contain the prompt, generated text, and other information.
    outputs = llm.generate(prompts, sampling_params)
    # Print the outputs.
    print("\nGenerated Outputs:\n" + "-" * 60)
    for output in outputs:
        prompt = output.prompt
        generated_text = output.outputs[0].text
        print(f"Prompt:    {prompt!r}")
        print(f"Output:    {generated_text!r}")
        print("-" * 60)

run nsight system to get report.

nsys profile -o report.nsys-rep --trace-fork-before-exec=true --cuda-graph-trace=node python test.py

Why can not get cuda information in nsight system?

image
image2952×1694 238 KB

2

@sanjivanis is this a HES/graph-node interference?

@byte.xiaobin can you try this again, explicitly asking for –trace=cuda-sw?