Function AsGpuKernel

tensorflow/stream_executor/gpu/gpu_kernel.h:92–94

Given a platform-independent kernel datatype, returns the (const) internal CUDA platform implementation pointer.

// Given a platform-independent kernel datatype, returns the (const) internal
// CUDA platform implementation pointer.
inline const GpuKernel* AsGpuKernel(const KernelBase* kernel) {
  return static_cast<const GpuKernel*>(kernel->implementation());
}

// Given a platform-independent kernel datatype, returns the (non-const)
// internal CUDA platform implementation pointer.