maximum Class — pytorch Architecture

Source Code

aten/src/ATen/cpu/vec/vec256/vsx/vec256_complex_float_vsx.h lines 629–644

template <>
Vectorized<ComplexFlt> inline maximum(
    const Vectorized<ComplexFlt>& a,
    const Vectorized<ComplexFlt>& b) {
  auto abs_a = a.abs_2_();
  auto abs_b = b.abs_2_();
  // auto mask = _mm256_cmp_ps(abs_a, abs_b, _CMP_LT_OQ);
  // auto max = _mm256_blendv_ps(a, b, mask);
  auto mask = abs_a.elwise_lt(abs_b);
  auto max = Vectorized<ComplexFlt>::elwise_blendv(a, b, mask);

  return max;
  // Exploit the fact that all-ones is a NaN.
  // auto isnan = _mm256_cmp_ps(abs_a, abs_b, _CMP_UNORD_Q);
  // return _mm256_or_ps(max, isnan);
}