GNU Radio 3.6.5 C++ API
|
00001 #ifndef INCLUDED_volk_32f_x2_subtract_32f_a_H 00002 #define INCLUDED_volk_32f_x2_subtract_32f_a_H 00003 00004 #include <inttypes.h> 00005 #include <stdio.h> 00006 00007 #ifdef LV_HAVE_SSE 00008 #include <xmmintrin.h> 00009 /*! 00010 \brief Subtracts bVector form aVector and store their results in the cVector 00011 \param cVector The vector where the results will be stored 00012 \param aVector The initial vector 00013 \param bVector The vector to be subtracted 00014 \param num_points The number of values in aVector and bVector to be subtracted together and stored into cVector 00015 */ 00016 static inline void volk_32f_x2_subtract_32f_a_sse(float* cVector, const float* aVector, const float* bVector, unsigned int num_points){ 00017 unsigned int number = 0; 00018 const unsigned int quarterPoints = num_points / 4; 00019 00020 float* cPtr = cVector; 00021 const float* aPtr = aVector; 00022 const float* bPtr= bVector; 00023 00024 __m128 aVal, bVal, cVal; 00025 for(;number < quarterPoints; number++){ 00026 00027 aVal = _mm_load_ps(aPtr); 00028 bVal = _mm_load_ps(bPtr); 00029 00030 cVal = _mm_sub_ps(aVal, bVal); 00031 00032 _mm_store_ps(cPtr,cVal); // Store the results back into the C container 00033 00034 aPtr += 4; 00035 bPtr += 4; 00036 cPtr += 4; 00037 } 00038 00039 number = quarterPoints * 4; 00040 for(;number < num_points; number++){ 00041 *cPtr++ = (*aPtr++) - (*bPtr++); 00042 } 00043 } 00044 #endif /* LV_HAVE_SSE */ 00045 00046 #ifdef LV_HAVE_GENERIC 00047 /*! 00048 \brief Subtracts bVector form aVector and store their results in the cVector 00049 \param cVector The vector where the results will be stored 00050 \param aVector The initial vector 00051 \param bVector The vector to be subtracted 00052 \param num_points The number of values in aVector and bVector to be subtracted together and stored into cVector 00053 */ 00054 static inline void volk_32f_x2_subtract_32f_a_generic(float* cVector, const float* aVector, const float* bVector, unsigned int num_points){ 00055 float* cPtr = cVector; 00056 const float* aPtr = aVector; 00057 const float* bPtr= bVector; 00058 unsigned int number = 0; 00059 00060 for(number = 0; number < num_points; number++){ 00061 *cPtr++ = (*aPtr++) - (*bPtr++); 00062 } 00063 } 00064 #endif /* LV_HAVE_GENERIC */ 00065 00066 #ifdef LV_HAVE_ORC 00067 /*! 00068 \brief Subtracts bVector form aVector and store their results in the cVector 00069 \param cVector The vector where the results will be stored 00070 \param aVector The initial vector 00071 \param bVector The vector to be subtracted 00072 \param num_points The number of values in aVector and bVector to be subtracted together and stored into cVector 00073 */ 00074 extern void volk_32f_x2_subtract_32f_a_orc_impl(float* cVector, const float* aVector, const float* bVector, unsigned int num_points); 00075 static inline void volk_32f_x2_subtract_32f_a_orc(float* cVector, const float* aVector, const float* bVector, unsigned int num_points){ 00076 volk_32f_x2_subtract_32f_a_orc_impl(cVector, aVector, bVector, num_points); 00077 } 00078 #endif /* LV_HAVE_ORC */ 00079 00080 00081 #endif /* INCLUDED_volk_32f_x2_subtract_32f_a_H */