Checklist

• 1. I have searched related issues but cannot get the expected help.
• 2. The bug has not been fixed in the latest version.
• 3. Please note that if the bug-related issue you submitted lacks corresponding environment info and a minimal reproducible demo, it will be challenging for us to reproduce and resolve the issue, reducing the likelihood of receiving feedback.
• 4. If the issue you raised is not a bug but a question, please raise a discussion at https://github.com/sgl-project/sglang/discussions/new/choose Otherwise, it will be closed.
• 5. Please use English, otherwise it will be closed.

Describe the bug

flashinfer-ai/flashinfer#969 claims that the flashinfer mla backend can be sped up after removal of

  with self.device as device:
      stream = torch.cuda.current_stream(device).cuda_stream

in fast_mla_decode_plan of flashinfer_mla_backend.py

We need to test its performance after removal.

Reproduction

python3 -m sglang.launch_server --model deepseek-ai/DeepSeek-R1 --tp 8 --trust-remote-code --enable-flashinfer-mla

Environment

GPU: H200 * 8
Latest version of sglang and flashinfer

Related PR

#5208 #5538