diff options
| author | Tom Rondeau <trondeau@vt.edu> | 2012-02-07 16:59:50 -0500 |
|---|---|---|
| committer | Tom Rondeau <trondeau@vt.edu> | 2012-02-13 14:57:27 -0500 |
| commit | cdb328758dca9fa494956c0e62f5e78adf613982 (patch) | |
| tree | 825a1021a2e3f62ca3372b5624f3ef56f38d9fe2 /volk | |
| parent | f34b496341ceb73baffee6f8bf84ed197ffeeaf0 (diff) | |
volk: fixed complex multiply and conjugate kernel to use xor for conjugation.
Diffstat (limited to 'volk')
| -rw-r--r-- | volk/include/volk/volk_32fc_x2_multiply_conjugate_32fc_a.h | 5 | ||||
| -rw-r--r-- | volk/include/volk/volk_32fc_x2_multiply_conjugate_32fc_u.h | 2 |
2 files changed, 3 insertions, 4 deletions
diff --git a/volk/include/volk/volk_32fc_x2_multiply_conjugate_32fc_a.h b/volk/include/volk/volk_32fc_x2_multiply_conjugate_32fc_a.h index 70476a8c7d..2a1bcbce0e 100644 --- a/volk/include/volk/volk_32fc_x2_multiply_conjugate_32fc_a.h +++ b/volk/include/volk/volk_32fc_x2_multiply_conjugate_32fc_a.h @@ -24,15 +24,14 @@ static inline void volk_32fc_x2_multiply_conjugate_32fc_a_sse3(lv_32fc_t* cVecto const lv_32fc_t* a = aVector; const lv_32fc_t* b = bVector; - __m128 conjugator = _mm_setr_ps(1, -1, 1, -1); + __m128 conjugator = _mm_setr_ps(0, -0.f, 0, -0.f); for(;number < halfPoints; number++){ x = _mm_load_ps((float*)a); // Load the ar + ai, br + bi as ar,ai,br,bi y = _mm_load_ps((float*)b); // Load the cr + ci, dr + di as cr,ci,dr,di - // FIXME: replace with xor for a faster implementation - y = _mm_mul_ps(y, conjugator); // conjugate y + y = _mm_xor_ps(y, conjugator); // conjugate y yl = _mm_moveldup_ps(y); // Load yl with cr,cr,dr,dr yh = _mm_movehdup_ps(y); // Load yh with ci,ci,di,di diff --git a/volk/include/volk/volk_32fc_x2_multiply_conjugate_32fc_u.h b/volk/include/volk/volk_32fc_x2_multiply_conjugate_32fc_u.h index fbaa29c174..92f6a051ee 100644 --- a/volk/include/volk/volk_32fc_x2_multiply_conjugate_32fc_u.h +++ b/volk/include/volk/volk_32fc_x2_multiply_conjugate_32fc_u.h @@ -24,7 +24,7 @@ static inline void volk_32fc_x2_multiply_conjugate_32fc_u_sse3(lv_32fc_t* cVecto const lv_32fc_t* a = aVector; const lv_32fc_t* b = bVector; - __m128 conjugator = _mm_set_ps(0, 0x80000000, 0, 0x80000000); + __m128 conjugator = _mm_setr_ps(0, -0.f, 0, -0.f); for(;number < halfPoints; number++){ |