Home / Class/ deinterleave2 Class — pytorch Architecture

deinterleave2 Class — pytorch Architecture

Architecture documentation for the deinterleave2 class in vec512.h from the pytorch codebase.

Entity Profile

Source Code

aten/src/ATen/cpu/vec/vec512/vec512.h lines 203–221

template <>
std::pair<Vectorized<double>, Vectorized<double>> inline deinterleave2<double>(
    const Vectorized<double>& a,
    const Vectorized<double>& b) {
  // inputs:
  //   a = {a0, b0, a1, b1, a2, b2, a3, b3}
  //   b = {a4, b4, a5, b5, a6, b6, a7, b7}
  // output:
  //   return {a0, a1, a2, a3, a4, a5, a6, a7}
  //          {b0, b1, b2, b3, b4, b5, b6, b7}
  // The members of indices have been written in binary format for better
  // understandability
  __m512i idx1 = _mm512_set_epi64(14, 12, 10, 8, 6, 4, 2, 0);
  __m512i idx2 = _mm512_set_epi64(15, 13, 11, 9, 7, 5, 3, 1);

  return std::make_pair(
      _mm512_mask_permutex2var_pd(a, 0xff, idx1, b),
      _mm512_mask_permutex2var_pd(a, 0xff, idx2, b));
}

Analyze Your Own Codebase

Get architecture documentation, dependency graphs, and domain analysis for your codebase in minutes.

Try Supermodel Free