cpp_impl() — pytorch Function Reference

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

Function python CoreTensor Dispatcher calls 1 called by 1

Entity Profile

CoreTensor→ Dispatcher→ cpp_impl() — pytorch Function Reference

Dependency Diagram

graph TD
  8b199137_d342_0675_061e_74abc2d1f308["cpp_impl()"]
  eb2d4816_acac_e28b_9e97_87aef6f8f594["cpp_impl()"]
  eb2d4816_acac_e28b_9e97_87aef6f8f594 -->|calls| 8b199137_d342_0675_061e_74abc2d1f308
  eb2d4816_acac_e28b_9e97_87aef6f8f594["cpp_impl()"]
  8b199137_d342_0675_061e_74abc2d1f308 -->|calls| eb2d4816_acac_e28b_9e97_87aef6f8f594
  style 8b199137_d342_0675_061e_74abc2d1f308 fill:#6366f1,stroke:#818cf8,color:#fff

Relationship Graph

Source Code

aten/src/ATen/native/transformers/cuda/mem_eff_attention/kernels/generate_kernels.py lines 217–223

    def cpp_impl(self) -> str:
        return KERNEL_IMPL_TEMPLATE.format(
            CPP_CLASS=self.cpp_class,
            NAME=self.name,
            SM=self.sm_range[0],
            SM_MAX=self.sm_range[1],
        )