inline float64x4_t conjugate(const float64x4_t ymm)
{
static const float64x4_t sign_mask = _mm256_setr_pd(+0.0, -0.0, -0.0, -0.0);
float64x4_t xor0 = _mm256_xor_pd(ymm, sign_mask);
return xor0;
}
__SIMDd _SIMD_neg_pd(__SIMDd a)
{
#ifdef USE_SSE
return _mm_xor_pd(a, _mm_set1_pd(-0.0f));
#elif defined USE_AVX
return _mm256_xor_pd(a, _mm_set1_pd(-0.0f));
#elif defined USE_IBM
return vec_neg(a);
#endif
}
/*!
* \brief Compute the negative of each element in the given vector
* \return a vector containing the negative of each input element
*/
ETL_STATIC_INLINE(avx_simd_double) minus(avx_simd_double x) {
return _mm256_xor_pd(x.value, _mm256_set1_pd(-0.));
}
inline vector4d operator~(const vector4d& rhs)
{
return _mm256_xor_pd(rhs, _mm256_castsi256_pd(_mm256_set1_epi32(-1)));
}
inline vector4d operator^(const vector4d& lhs, const vector4d& rhs)
{
return _mm256_xor_pd(lhs, rhs);
}
BI_FORCE_INLINE inline const avx_double operator-(const avx_double& o) {
avx_double res;
res.packed = _mm256_xor_pd(_mm256_set1_pd(-0.0), o.packed);
return res;
}
