70 #ifndef INCLUDED_volk_32f_s32f_normalize_a_H 71 #define INCLUDED_volk_32f_s32f_normalize_a_H 77 #include <immintrin.h> 80 unsigned int number = 0;
81 float* inputPtr = vecBuffer;
83 const float invScalar = 1.0 / scalar;
84 __m256 vecScalar = _mm256_set1_ps(invScalar);
88 const uint64_t eighthPoints = num_points / 8;
89 for(;number < eighthPoints; number++){
91 input1 = _mm256_load_ps(inputPtr);
93 input1 = _mm256_mul_ps(input1, vecScalar);
95 _mm256_store_ps(inputPtr, input1);
100 number = eighthPoints*8;
101 for(; number < num_points; number++){
102 *inputPtr *= invScalar;
109 #include <xmmintrin.h> 112 unsigned int number = 0;
113 float* inputPtr = vecBuffer;
115 const float invScalar = 1.0 / scalar;
116 __m128 vecScalar = _mm_set_ps1(invScalar);
120 const uint64_t quarterPoints = num_points / 4;
121 for(;number < quarterPoints; number++){
123 input1 = _mm_load_ps(inputPtr);
125 input1 = _mm_mul_ps(input1, vecScalar);
127 _mm_store_ps(inputPtr, input1);
132 number = quarterPoints*4;
133 for(; number < num_points; number++){
134 *inputPtr *= invScalar;
140 #ifdef LV_HAVE_GENERIC 143 unsigned int number = 0;
144 float* inputPtr = vecBuffer;
145 const float invScalar = 1.0 / scalar;
146 for(number = 0; number < num_points; number++){
147 *inputPtr *= invScalar;
155 extern void volk_32f_s32f_normalize_a_orc_impl(
float* dst,
float* src,
const float scalar,
unsigned int num_points);
156 static inline void volk_32f_s32f_normalize_u_orc(
float* vecBuffer,
const float scalar,
unsigned int num_points){
157 float invscalar = 1.0 / scalar;
158 volk_32f_s32f_normalize_a_orc_impl(vecBuffer, vecBuffer, invscalar, num_points);
164 #ifndef INCLUDED_volk_32f_s32f_normalize_u_H 165 #define INCLUDED_volk_32f_s32f_normalize_u_H 167 #include <inttypes.h> 170 #include <immintrin.h> 173 unsigned int number = 0;
174 float* inputPtr = vecBuffer;
176 const float invScalar = 1.0 / scalar;
177 __m256 vecScalar = _mm256_set1_ps(invScalar);
181 const uint64_t eighthPoints = num_points / 8;
182 for(;number < eighthPoints; number++){
184 input1 = _mm256_loadu_ps(inputPtr);
186 input1 = _mm256_mul_ps(input1, vecScalar);
188 _mm256_storeu_ps(inputPtr, input1);
193 number = eighthPoints*8;
194 for(; number < num_points; number++){
195 *inputPtr *= invScalar;
static void volk_32f_s32f_normalize_generic(float *vecBuffer, const float scalar, unsigned int num_points)
Definition: volk_32f_s32f_normalize.h:142
static void volk_32f_s32f_normalize_u_avx(float *vecBuffer, const float scalar, unsigned int num_points)
Definition: volk_32f_s32f_normalize.h:172
static void volk_32f_s32f_normalize_a_avx(float *vecBuffer, const float scalar, unsigned int num_points)
Definition: volk_32f_s32f_normalize.h:79
static void volk_32f_s32f_normalize_a_sse(float *vecBuffer, const float scalar, unsigned int num_points)
Definition: volk_32f_s32f_normalize.h:111