name() — pytorch Function Reference

Architecture documentation for the name() function in generate_kernels.py from the pytorch codebase.

Function python ComputeKernels QuantizedCPU calls 1 called by 1

Entity Profile

ComputeKernels→ QuantizedCPU→ name() — pytorch Function Reference

Dependency Diagram

graph TD
  43bf4dfa_33b6_275e_b989_8f50c5c2b9cd["name()"]
  d8d80e47_3bf7_91f3_187b_0bebaae577c7["name()"]
  d8d80e47_3bf7_91f3_187b_0bebaae577c7 -->|calls| 43bf4dfa_33b6_275e_b989_8f50c5c2b9cd
  d8d80e47_3bf7_91f3_187b_0bebaae577c7["name()"]
  43bf4dfa_33b6_275e_b989_8f50c5c2b9cd -->|calls| d8d80e47_3bf7_91f3_187b_0bebaae577c7
  style 43bf4dfa_33b6_275e_b989_8f50c5c2b9cd fill:#6366f1,stroke:#818cf8,color:#fff

Relationship Graph

Source Code

aten/src/ATen/native/transformers/cuda/mem_eff_attention/kernels/generate_kernels.py lines 182–190

    def name(self) -> str:
        dropout_suffix = "_dropout" if self.apply_dropout else ""
        seqlen_aligned_suffix = (
            "_seqaligned" if self.keys_queries_aligned_to_blocksizes else ""
        )
        return (
            f"fmha_cutlassB_{self.dtype}_{self._aligned_suffix}"
            f"_{self.block_i}x{self.block_j}_k{self.max_k}{dropout_suffix}{seqlen_aligned_suffix}_sm{self.sm_range[0]}"
        )