[Bug Fix]Fix m_indices argument passing in BF16 grouped contiguous warmup

[laurxh]

Motivation

DeepGEMM >= 2.5.0 (PR deepseek-ai/DeepGEMM#280) renamed the 4th parameter of
m_grouped_bf16_gemm_nt_contiguous from m_indices to grouped_layout.
The BF16 grouped contiguous warmup executor in SGLang uses keyword argument
m_indices=, which causes a TypeError on startup:

TypeError: m_grouped_bf16_gemm_nt_contiguous(): incompatible function arguments. The following argument types are supported:                                            
   1. (a: torch.Tensor, b: torch.Tensor, d: torch.Tensor, grouped_layout: torch.Tensor, compiled_dims: str = 'nk', use_psum_layout: bool = False,                      
expected_m_for_psum_layout: typing.SupportsInt | None = None) -> None

Reproduce:

python -c "
from sglang.srt.layers.deep_gemm_wrapper.compile_utils import _BF16GroupedContWarmupExecutor
executor = _BF16GroupedContWarmupExecutor(max_m=128, n=128, k=128, num_groups=4)
executor.execute(m=128)
"

Modifications

• Changed m_indices=self.m_indices[:m] (keyword) to self.m_indices[:m] (positional)
  in _BF16GroupedContWarmupExecutor.execute() to match the DeepGEMM pybind11 API.

• The parameter functionality is unchanged — it is still an int32 tensor indicating which
  expert group each row belongs to. Only the parameter name was changed upstream.

Checklist

• Format your code according to the Format code with pre-commit.
• Add unit tests according to the Run and add unit tests.
• Update documentation according to Write documentations.
• Provide accuracy and speed benchmark results according to Test the accuracy and Benchmark the speed.
• Follow the SGLang code style guidance.

---

CI States

Latest PR Test (Base): ❌ Run #26806013714
Latest PR Test (Extra): ❌ Run #26806013517

[gemini-code-assist]

Warning

You have reached your daily quota limit. Please wait up to 24 hours and I will start processing your requests again!

[gemini-code-assist]

Warning

You have reached your daily quota limit. Please wait up to 24 hours and I will start processing your requests again!