[hamradio-commits] [gnss-sdr] 156/236: Introducing a new resampler kernel for comparison
Carles Fernandez
carles_fernandez-guest at moszumanska.debian.org
Tue Apr 26 16:02:47 UTC 2016
This is an automated email from the git hooks/post-receive script.
carles_fernandez-guest pushed a commit to branch next
in repository gnss-sdr.
commit 57107cf86d6dcff721aae3e758dbce26ac8e2f6c
Author: Carles Fernandez <carles.fernandez at gmail.com>
Date: Fri Apr 1 12:41:00 2016 +0200
Introducing a new resampler kernel for comparison
---
...=> volk_gnsssdr_16ic_resamplerxnpuppet2_16ic.h} | 95 ++---
.../volk_gnsssdr_16ic_xn_resampler2_16ic_xn.h | 397 +++++++++++++++++++++
.../volk_gnsssdr_16ic_xn_resampler_16ic_xn.h | 11 +-
.../volk_gnsssdr_32fc_resamplerxnpuppet_32fc.h | 6 +-
.../volk_gnsssdr/lib/kernel_tests.h | 1 +
5 files changed, 453 insertions(+), 57 deletions(-)
diff --git a/src/algorithms/libs/volk_gnsssdr_module/volk_gnsssdr/kernels/volk_gnsssdr/volk_gnsssdr_32fc_resamplerxnpuppet_32fc.h b/src/algorithms/libs/volk_gnsssdr_module/volk_gnsssdr/kernels/volk_gnsssdr/volk_gnsssdr_16ic_resamplerxnpuppet2_16ic.h
similarity index 53%
copy from src/algorithms/libs/volk_gnsssdr_module/volk_gnsssdr/kernels/volk_gnsssdr/volk_gnsssdr_32fc_resamplerxnpuppet_32fc.h
copy to src/algorithms/libs/volk_gnsssdr_module/volk_gnsssdr/kernels/volk_gnsssdr/volk_gnsssdr_16ic_resamplerxnpuppet2_16ic.h
index 53b1044..a3958dc 100644
--- a/src/algorithms/libs/volk_gnsssdr_module/volk_gnsssdr/kernels/volk_gnsssdr/volk_gnsssdr_32fc_resamplerxnpuppet_32fc.h
+++ b/src/algorithms/libs/volk_gnsssdr_module/volk_gnsssdr/kernels/volk_gnsssdr/volk_gnsssdr_16ic_resamplerxnpuppet2_16ic.h
@@ -1,5 +1,5 @@
/*!
- * \file volk_gnsssdr_32fc_resamplerxnpuppet_32fc.h
+ * \file volk_gnsssdr_16ic_resamplerxnpuppet_16ic.h
* \brief VOLK_GNSSSDR puppet for the multiple 16-bit complex vector resampler kernel.
* \authors <ul>
* <li> Carles Fernandez Prades 2016 cfernandez at cttc dot cat
@@ -32,17 +32,17 @@
* -------------------------------------------------------------------------
*/
-#ifndef INCLUDED_volk_gnsssdr_32fc_resamplerxnpuppet_32fc_H
-#define INCLUDED_volk_gnsssdr_32fc_resamplerxnpuppet_32fc_H
+#ifndef INCLUDED_volk_gnsssdr_16ic_resamplerxnpuppet2_16ic_H
+#define INCLUDED_volk_gnsssdr_16ic_resamplerxnpuppet2_16ic_H
-#include "volk_gnsssdr/volk_gnsssdr_32fc_xn_resampler_32fc_xn.h"
+#include "volk_gnsssdr/volk_gnsssdr_16ic_xn_resampler2_16ic_xn.h"
#include <volk_gnsssdr/volk_gnsssdr_malloc.h>
#include <volk_gnsssdr/volk_gnsssdr_complex.h>
#include <volk_gnsssdr/volk_gnsssdr.h>
#include <string.h>
#ifdef LV_HAVE_GENERIC
-static inline void volk_gnsssdr_32fc_resamplerxnpuppet_32fc_generic(lv_32fc_t* result, const lv_32fc_t* local_code, unsigned int num_points)
+static inline void volk_gnsssdr_16ic_resamplerxnpuppet2_16ic_generic(lv_16sc_t* result, const lv_16sc_t* local_code, unsigned int num_points)
{
float code_phase_step_chips = 0.6;
int code_length_chips = 2046;
@@ -51,15 +51,15 @@ static inline void volk_gnsssdr_32fc_resamplerxnpuppet_32fc_generic(lv_32fc_t* r
float shifts_chips[3] = { -0.1, 0.0, 0.1 };
- lv_32fc_t** result_aux = (lv_32fc_t**)volk_gnsssdr_malloc(sizeof(lv_32fc_t*) * num_out_vectors, volk_gnsssdr_get_alignment());
+ lv_16sc_t** result_aux = (lv_16sc_t**)volk_gnsssdr_malloc(sizeof(lv_16sc_t*) * num_out_vectors, volk_gnsssdr_get_alignment());
for(unsigned int n = 0; n < num_out_vectors; n++)
{
- result_aux[n] = (lv_32fc_t*)volk_gnsssdr_malloc(sizeof(lv_32fc_t) * num_points, volk_gnsssdr_get_alignment());
+ result_aux[n] = (lv_16sc_t*)volk_gnsssdr_malloc(sizeof(lv_16sc_t) * num_points, volk_gnsssdr_get_alignment());
}
- volk_gnsssdr_32fc_xn_resampler_32fc_xn_generic(result_aux, local_code, rem_code_phase_chips, code_phase_step_chips, shifts_chips, code_length_chips, num_out_vectors, num_points);
+ volk_gnsssdr_16ic_xn_resampler2_16ic_xn_generic(result_aux, local_code, rem_code_phase_chips, code_phase_step_chips, shifts_chips, code_length_chips, num_out_vectors, num_points);
- memcpy((lv_32fc_t*)result, (lv_32fc_t*)result_aux[0], sizeof(lv_32fc_t) * num_points);
+ memcpy((lv_16sc_t*)result, (lv_16sc_t*)result_aux[0], sizeof(lv_16sc_t) * num_points);
for(unsigned int n = 0; n < num_out_vectors; n++)
{
@@ -72,7 +72,7 @@ static inline void volk_gnsssdr_32fc_resamplerxnpuppet_32fc_generic(lv_32fc_t* r
#ifdef LV_HAVE_SSE3
-static inline void volk_gnsssdr_32fc_resamplerxnpuppet_32fc_a_sse3(lv_32fc_t* result, const lv_32fc_t* local_code, unsigned int num_points)
+static inline void volk_gnsssdr_16ic_resamplerxnpuppet2_16ic_a_sse3(lv_16sc_t* result, const lv_16sc_t* local_code, unsigned int num_points)
{
float code_phase_step_chips = 0.6;
int code_length_chips = 2046;
@@ -81,15 +81,15 @@ static inline void volk_gnsssdr_32fc_resamplerxnpuppet_32fc_a_sse3(lv_32fc_t* re
float shifts_chips[3] = { -0.1, 0.0, 0.1 };
- lv_32fc_t** result_aux = (lv_32fc_t**)volk_gnsssdr_malloc(sizeof(lv_32fc_t*) * num_out_vectors, volk_gnsssdr_get_alignment());
+ lv_16sc_t** result_aux = (lv_16sc_t**)volk_gnsssdr_malloc(sizeof(lv_16sc_t*) * num_out_vectors, volk_gnsssdr_get_alignment());
for(unsigned int n = 0; n < num_out_vectors; n++)
{
- result_aux[n] = (lv_32fc_t*)volk_gnsssdr_malloc(sizeof(lv_32fc_t) * num_points, volk_gnsssdr_get_alignment());
+ result_aux[n] = (lv_16sc_t*)volk_gnsssdr_malloc(sizeof(lv_16sc_t) * num_points, volk_gnsssdr_get_alignment());
}
- volk_gnsssdr_32fc_xn_resampler_32fc_xn_a_sse3(result_aux, local_code, rem_code_phase_chips, code_phase_step_chips, shifts_chips, code_length_chips, num_out_vectors, num_points);
+ volk_gnsssdr_16ic_xn_resampler2_16ic_xn_a_sse3(result_aux, local_code, rem_code_phase_chips, code_phase_step_chips, shifts_chips, code_length_chips, num_out_vectors, num_points);
- memcpy((lv_32fc_t*)result, (lv_32fc_t*)result_aux[0], sizeof(lv_32fc_t) * num_points);
+ memcpy((lv_16sc_t*)result, (lv_16sc_t*)result_aux[0], sizeof(lv_16sc_t) * num_points);
for(unsigned int n = 0; n < num_out_vectors; n++)
{
@@ -101,7 +101,7 @@ static inline void volk_gnsssdr_32fc_resamplerxnpuppet_32fc_a_sse3(lv_32fc_t* re
#endif
#ifdef LV_HAVE_SSE3
-static inline void volk_gnsssdr_32fc_resamplerxnpuppet_32fc_u_sse3(lv_32fc_t* result, const lv_32fc_t* local_code, unsigned int num_points)
+static inline void volk_gnsssdr_16ic_resamplerxnpuppet2_16ic_u_sse3(lv_16sc_t* result, const lv_16sc_t* local_code, unsigned int num_points)
{
float code_phase_step_chips = 0.6;
int code_length_chips = 2046;
@@ -110,15 +110,15 @@ static inline void volk_gnsssdr_32fc_resamplerxnpuppet_32fc_u_sse3(lv_32fc_t* re
float shifts_chips[3] = { -0.1, 0.0, 0.1 };
- lv_32fc_t** result_aux = (lv_32fc_t**)volk_gnsssdr_malloc(sizeof(lv_32fc_t*) * num_out_vectors, volk_gnsssdr_get_alignment());
+ lv_16sc_t** result_aux = (lv_16sc_t**)volk_gnsssdr_malloc(sizeof(lv_16sc_t*) * num_out_vectors, volk_gnsssdr_get_alignment());
for(unsigned int n = 0; n < num_out_vectors; n++)
{
- result_aux[n] = (lv_32fc_t*)volk_gnsssdr_malloc(sizeof(lv_32fc_t) * num_points, volk_gnsssdr_get_alignment());
+ result_aux[n] = (lv_16sc_t*)volk_gnsssdr_malloc(sizeof(lv_16sc_t) * num_points, volk_gnsssdr_get_alignment());
}
- volk_gnsssdr_32fc_xn_resampler_32fc_xn_a_sse3(result_aux, local_code, rem_code_phase_chips, code_phase_step_chips, shifts_chips, code_length_chips, num_out_vectors, num_points);
+ volk_gnsssdr_16ic_xn_resampler2_16ic_xn_u_sse3(result_aux, local_code, rem_code_phase_chips, code_phase_step_chips, shifts_chips, code_length_chips, num_out_vectors, num_points);
- memcpy((lv_32fc_t*)result, (lv_32fc_t*)result_aux[0], sizeof(lv_32fc_t) * num_points);
+ memcpy((lv_16sc_t*)result, (lv_16sc_t*)result_aux[0], sizeof(lv_16sc_t) * num_points);
for(unsigned int n = 0; n < num_out_vectors; n++)
{
@@ -131,7 +131,7 @@ static inline void volk_gnsssdr_32fc_resamplerxnpuppet_32fc_u_sse3(lv_32fc_t* re
#ifdef LV_HAVE_SSE4_1
-static inline void volk_gnsssdr_32fc_resamplerxnpuppet_32fc_u_sse4_1(lv_32fc_t* result, const lv_32fc_t* local_code, unsigned int num_points)
+static inline void volk_gnsssdr_16ic_resamplerxnpuppet2_16ic_u_sse4_1(lv_16sc_t* result, const lv_16sc_t* local_code, unsigned int num_points)
{
float code_phase_step_chips = 0.6;
int code_length_chips = 2046;
@@ -140,15 +140,15 @@ static inline void volk_gnsssdr_32fc_resamplerxnpuppet_32fc_u_sse4_1(lv_32fc_t*
float shifts_chips[3] = { -0.1, 0.0, 0.1 };
- lv_32fc_t** result_aux = (lv_32fc_t**)volk_gnsssdr_malloc(sizeof(lv_32fc_t*) * num_out_vectors, volk_gnsssdr_get_alignment());
+ lv_16sc_t** result_aux = (lv_16sc_t**)volk_gnsssdr_malloc(sizeof(lv_16sc_t*) * num_out_vectors, volk_gnsssdr_get_alignment());
for(unsigned int n = 0; n < num_out_vectors; n++)
{
- result_aux[n] = (lv_32fc_t*)volk_gnsssdr_malloc(sizeof(lv_32fc_t) * num_points, volk_gnsssdr_get_alignment());
+ result_aux[n] = (lv_16sc_t*)volk_gnsssdr_malloc(sizeof(lv_16sc_t) * num_points, volk_gnsssdr_get_alignment());
}
- volk_gnsssdr_32fc_xn_resampler_32fc_xn_a_sse4_1(result_aux, local_code, rem_code_phase_chips, code_phase_step_chips, shifts_chips, code_length_chips, num_out_vectors, num_points);
+ volk_gnsssdr_16ic_xn_resampler2_16ic_xn_u_sse4_1(result_aux, local_code, rem_code_phase_chips, code_phase_step_chips, shifts_chips, code_length_chips, num_out_vectors, num_points);
- memcpy((lv_32fc_t*)result, (lv_32fc_t*)result_aux[0], sizeof(lv_32fc_t) * num_points);
+ memcpy((lv_16sc_t*)result, (lv_16sc_t*)result_aux[0], sizeof(lv_16sc_t) * num_points);
for(unsigned int n = 0; n < num_out_vectors; n++)
{
@@ -159,8 +159,9 @@ static inline void volk_gnsssdr_32fc_resamplerxnpuppet_32fc_u_sse4_1(lv_32fc_t*
#endif
+
#ifdef LV_HAVE_SSE4_1
-static inline void volk_gnsssdr_32fc_resamplerxnpuppet_32fc_a_sse4_1(lv_32fc_t* result, const lv_32fc_t* local_code, unsigned int num_points)
+static inline void volk_gnsssdr_16ic_resamplerxnpuppet2_16ic_a_sse4_1(lv_16sc_t* result, const lv_16sc_t* local_code, unsigned int num_points)
{
float code_phase_step_chips = 0.6;
int code_length_chips = 2046;
@@ -169,15 +170,15 @@ static inline void volk_gnsssdr_32fc_resamplerxnpuppet_32fc_a_sse4_1(lv_32fc_t*
float shifts_chips[3] = { -0.1, 0.0, 0.1 };
- lv_32fc_t** result_aux = (lv_32fc_t**)volk_gnsssdr_malloc(sizeof(lv_32fc_t*) * num_out_vectors, volk_gnsssdr_get_alignment());
+ lv_16sc_t** result_aux = (lv_16sc_t**)volk_gnsssdr_malloc(sizeof(lv_16sc_t*) * num_out_vectors, volk_gnsssdr_get_alignment());
for(unsigned int n = 0; n < num_out_vectors; n++)
{
- result_aux[n] = (lv_32fc_t*)volk_gnsssdr_malloc(sizeof(lv_32fc_t) * num_points, volk_gnsssdr_get_alignment());
+ result_aux[n] = (lv_16sc_t*)volk_gnsssdr_malloc(sizeof(lv_16sc_t) * num_points, volk_gnsssdr_get_alignment());
}
- volk_gnsssdr_32fc_xn_resampler_32fc_xn_a_sse4_1(result_aux, local_code, rem_code_phase_chips, code_phase_step_chips, shifts_chips, code_length_chips, num_out_vectors, num_points);
+ volk_gnsssdr_16ic_xn_resampler2_16ic_xn_a_sse4_1(result_aux, local_code, rem_code_phase_chips, code_phase_step_chips, shifts_chips, code_length_chips, num_out_vectors, num_points);
- memcpy((lv_32fc_t*)result, (lv_32fc_t*)result_aux[0], sizeof(lv_32fc_t) * num_points);
+ memcpy((lv_16sc_t*)result, (lv_16sc_t*)result_aux[0], sizeof(lv_16sc_t) * num_points);
for(unsigned int n = 0; n < num_out_vectors; n++)
{
@@ -188,8 +189,9 @@ static inline void volk_gnsssdr_32fc_resamplerxnpuppet_32fc_a_sse4_1(lv_32fc_t*
#endif
+
#ifdef LV_HAVE_AVX
-static inline void volk_gnsssdr_32fc_resamplerxnpuppet_32fc_a_avx(lv_32fc_t* result, const lv_32fc_t* local_code, unsigned int num_points)
+static inline void volk_gnsssdr_16ic_resamplerxnpuppet2_16ic_u_avx(lv_16sc_t* result, const lv_16sc_t* local_code, unsigned int num_points)
{
float code_phase_step_chips = 0.6;
int code_length_chips = 2046;
@@ -198,15 +200,15 @@ static inline void volk_gnsssdr_32fc_resamplerxnpuppet_32fc_a_avx(lv_32fc_t* res
float shifts_chips[3] = { -0.1, 0.0, 0.1 };
- lv_32fc_t** result_aux = (lv_32fc_t**)volk_gnsssdr_malloc(sizeof(lv_32fc_t*) * num_out_vectors, volk_gnsssdr_get_alignment());
+ lv_16sc_t** result_aux = (lv_16sc_t**)volk_gnsssdr_malloc(sizeof(lv_16sc_t*) * num_out_vectors, volk_gnsssdr_get_alignment());
for(unsigned int n = 0; n < num_out_vectors; n++)
{
- result_aux[n] = (lv_32fc_t*)volk_gnsssdr_malloc(sizeof(lv_32fc_t) * num_points, volk_gnsssdr_get_alignment());
+ result_aux[n] = (lv_16sc_t*)volk_gnsssdr_malloc(sizeof(lv_16sc_t) * num_points, volk_gnsssdr_get_alignment());
}
- volk_gnsssdr_32fc_xn_resampler_32fc_xn_a_avx(result_aux, local_code, rem_code_phase_chips, code_phase_step_chips, shifts_chips, code_length_chips, num_out_vectors, num_points);
+ volk_gnsssdr_16ic_xn_resampler2_16ic_xn_u_avx(result_aux, local_code, rem_code_phase_chips, code_phase_step_chips, shifts_chips, code_length_chips, num_out_vectors, num_points);
- memcpy((lv_32fc_t*)result, (lv_32fc_t*)result_aux[0], sizeof(lv_32fc_t) * num_points);
+ memcpy((lv_16sc_t*)result, (lv_16sc_t*)result_aux[0], sizeof(lv_16sc_t) * num_points);
for(unsigned int n = 0; n < num_out_vectors; n++)
{
@@ -214,11 +216,12 @@ static inline void volk_gnsssdr_32fc_resamplerxnpuppet_32fc_a_avx(lv_32fc_t* res
}
volk_gnsssdr_free(result_aux);
}
+
#endif
#ifdef LV_HAVE_AVX
-static inline void volk_gnsssdr_32fc_resamplerxnpuppet_32fc_u_avx(lv_32fc_t* result, const lv_32fc_t* local_code, unsigned int num_points)
+static inline void volk_gnsssdr_16ic_resamplerxnpuppet2_16ic_a_avx(lv_16sc_t* result, const lv_16sc_t* local_code, unsigned int num_points)
{
float code_phase_step_chips = 0.6;
int code_length_chips = 2046;
@@ -227,15 +230,15 @@ static inline void volk_gnsssdr_32fc_resamplerxnpuppet_32fc_u_avx(lv_32fc_t* res
float shifts_chips[3] = { -0.1, 0.0, 0.1 };
- lv_32fc_t** result_aux = (lv_32fc_t**)volk_gnsssdr_malloc(sizeof(lv_32fc_t*) * num_out_vectors, volk_gnsssdr_get_alignment());
+ lv_16sc_t** result_aux = (lv_16sc_t**)volk_gnsssdr_malloc(sizeof(lv_16sc_t*) * num_out_vectors, volk_gnsssdr_get_alignment());
for(unsigned int n = 0; n < num_out_vectors; n++)
{
- result_aux[n] = (lv_32fc_t*)volk_gnsssdr_malloc(sizeof(lv_32fc_t) * num_points, volk_gnsssdr_get_alignment());
+ result_aux[n] = (lv_16sc_t*)volk_gnsssdr_malloc(sizeof(lv_16sc_t) * num_points, volk_gnsssdr_get_alignment());
}
- volk_gnsssdr_32fc_xn_resampler_32fc_xn_a_avx(result_aux, local_code, rem_code_phase_chips, code_phase_step_chips, shifts_chips, code_length_chips, num_out_vectors, num_points);
+ volk_gnsssdr_16ic_xn_resampler2_16ic_xn_a_avx(result_aux, local_code, rem_code_phase_chips, code_phase_step_chips, shifts_chips, code_length_chips, num_out_vectors, num_points);
- memcpy((lv_32fc_t*)result, (lv_32fc_t*)result_aux[0], sizeof(lv_32fc_t) * num_points);
+ memcpy((lv_16sc_t*)result, (lv_16sc_t*)result_aux[0], sizeof(lv_16sc_t) * num_points);
for(unsigned int n = 0; n < num_out_vectors; n++)
{
@@ -243,11 +246,12 @@ static inline void volk_gnsssdr_32fc_resamplerxnpuppet_32fc_u_avx(lv_32fc_t* res
}
volk_gnsssdr_free(result_aux);
}
+
#endif
#ifdef LV_HAVE_NEON
-static inline void volk_gnsssdr_32fc_resamplerxnpuppet_32fc_neon(lv_32fc_t* result, const lv_32fc_t* local_code, unsigned int num_points)
+static inline void volk_gnsssdr_16ic_resamplerxnpuppet2_16ic_neon(lv_16sc_t* result, const lv_16sc_t* local_code, unsigned int num_points)
{
float code_phase_step_chips = 0.6;
int code_length_chips = 2046;
@@ -256,15 +260,15 @@ static inline void volk_gnsssdr_32fc_resamplerxnpuppet_32fc_neon(lv_32fc_t* resu
float shifts_chips[3] = { -0.1, 0.0, 0.1 };
- lv_32fc_t** result_aux = (lv_32fc_t**)volk_gnsssdr_malloc(sizeof(lv_32fc_t*) * num_out_vectors, volk_gnsssdr_get_alignment());
+ lv_16sc_t** result_aux = (lv_16sc_t**)volk_gnsssdr_malloc(sizeof(lv_16sc_t*) * num_out_vectors, volk_gnsssdr_get_alignment());
for(unsigned int n = 0; n < num_out_vectors; n++)
{
- result_aux[n] = (lv_32fc_t*)volk_gnsssdr_malloc(sizeof(lv_32fc_t) * num_points, volk_gnsssdr_get_alignment());
+ result_aux[n] = (lv_16sc_t*)volk_gnsssdr_malloc(sizeof(lv_16sc_t) * num_points, volk_gnsssdr_get_alignment());
}
- volk_gnsssdr_32fc_xn_resampler_32fc_xn_neon(result_aux, local_code, rem_code_phase_chips, code_phase_step_chips, shifts_chips, code_length_chips, num_out_vectors, num_points);
+ volk_gnsssdr_16ic_xn_resampler2_16ic_xn_neon(result_aux, local_code, rem_code_phase_chips, code_phase_step_chips, shifts_chips, code_length_chips, num_out_vectors, num_points);
- memcpy((lv_32fc_t*)result, (lv_32fc_t*)result_aux[0], sizeof(lv_32fc_t) * num_points);
+ memcpy((lv_16sc_t*)result, (lv_16sc_t*)result_aux[0], sizeof(lv_16sc_t) * num_points);
for(unsigned int n = 0; n < num_out_vectors; n++)
{
@@ -272,6 +276,7 @@ static inline void volk_gnsssdr_32fc_resamplerxnpuppet_32fc_neon(lv_32fc_t* resu
}
volk_gnsssdr_free(result_aux);
}
+
#endif
-#endif // INCLUDED_volk_gnsssdr_32fc_resamplerpuppet_32fc_H
+#endif // INCLUDED_volk_gnsssdr_16ic_resamplerpuppet_16ic_H
diff --git a/src/algorithms/libs/volk_gnsssdr_module/volk_gnsssdr/kernels/volk_gnsssdr/volk_gnsssdr_16ic_xn_resampler2_16ic_xn.h b/src/algorithms/libs/volk_gnsssdr_module/volk_gnsssdr/kernels/volk_gnsssdr/volk_gnsssdr_16ic_xn_resampler2_16ic_xn.h
new file mode 100644
index 0000000..f460b64
--- /dev/null
+++ b/src/algorithms/libs/volk_gnsssdr_module/volk_gnsssdr/kernels/volk_gnsssdr/volk_gnsssdr_16ic_xn_resampler2_16ic_xn.h
@@ -0,0 +1,397 @@
+/*!
+ * \file volk_gnsssdr_16ic_xn_resampler_16ic_xn.h
+ * \brief VOLK_GNSSSDR kernel: Resamples N 16 bits integer short complex vectors using zero hold resample algorithm.
+ * \authors <ul>
+ * <li> Javier Arribas, 2015. jarribas(at)cttc.es
+ * </ul>
+ *
+ * VOLK_GNSSSDR kernel that esamples N 16 bits integer short complex vectors using zero hold resample algorithm.
+ * It is optimized to resample a sigle GNSS local code signal replica into N vectors fractional-resampled and fractional-delayed
+ * (i.e. it creates the Early, Prompt, and Late code replicas)
+ *
+ * -------------------------------------------------------------------------
+ *
+ * Copyright (C) 2010-2015 (see AUTHORS file for a list of contributors)
+ *
+ * GNSS-SDR is a software defined Global Navigation
+ * Satellite Systems receiver
+ *
+ * This file is part of GNSS-SDR.
+ *
+ * GNSS-SDR is free software: you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation, either version 3 of the License, or
+ * (at your option) any later version.
+ *
+ * GNSS-SDR is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with GNSS-SDR. If not, see <http://www.gnu.org/licenses/>.
+ *
+ * -------------------------------------------------------------------------
+ */
+
+/*!
+ * \page volk_gnsssdr_16ic_xn_resampler_16ic_xn
+ *
+ * \b Overview
+ *
+ * Resamples a complex vector (16-bit integer each component), providing \p num_out_vectors outputs.
+ *
+ * <b>Dispatcher Prototype</b>
+ * \code
+ * void volk_gnsssdr_16ic_xn_resampler_16ic_xn(lv_16sc_t** result, const lv_16sc_t* local_code, float* rem_code_phase_chips, float code_phase_step_chips, unsigned int code_length_chips, int num_out_vectors, unsigned int num_output_samples)
+ * \endcode
+ *
+ * \b Inputs
+ * \li local_code: One of the vectors to be multiplied.
+ * \li rem_code_phase_chips: Remnant code phase [chips].
+ * \li code_phase_step_chips: Phase increment per sample [chips/sample].
+ * \li code_length_chips: Code length in chips.
+ * \li num_out_vectors Number of output vectors.
+ * \li num_output_samples: The number of data values to be in the resampled vector.
+ *
+ * \b Outputs
+ * \li result: Pointer to a vector of pointers where the results will be stored.
+ *
+ */
+
+#ifndef INCLUDED_volk_gnsssdr_16ic_xn_resampler2_16ic_xn_H
+#define INCLUDED_volk_gnsssdr_16ic_xn_resampler2_16ic_xn_H
+
+#include <math.h>
+#include <volk_gnsssdr/volk_gnsssdr_common.h>
+#include <volk_gnsssdr/volk_gnsssdr_complex.h>
+
+
+#ifdef LV_HAVE_GENERIC
+
+static inline void volk_gnsssdr_16ic_xn_resampler2_16ic_xn_generic(lv_16sc_t** result, const lv_16sc_t* local_code, float rem_code_phase_chips, float code_phase_step_chips, float* shifts_chips, unsigned int code_length_chips, int num_out_vectors, unsigned int num_output_samples)
+{
+ int local_code_chip_index;
+ for (int current_correlator_tap = 0; current_correlator_tap < num_out_vectors; current_correlator_tap++)
+ {
+ for (int n = 0; n < num_output_samples; n++)
+ {
+ // resample code for current tap
+ local_code_chip_index = (int)floor(code_phase_step_chips * (float)n + shifts_chips[current_correlator_tap] - rem_code_phase_chips);
+ local_code_chip_index = local_code_chip_index % code_length_chips;
+ //Take into account that in multitap correlators, the shifts can be negative!
+ if (local_code_chip_index < 0) local_code_chip_index += code_length_chips;
+ result[current_correlator_tap][n] = local_code[local_code_chip_index];
+ }
+ }
+}
+
+#endif /*LV_HAVE_GENERIC*/
+
+
+#ifdef LV_HAVE_SSE4_1
+#include <smmintrin.h>
+static inline void volk_gnsssdr_16ic_xn_resampler2_16ic_xn_a_sse4_1(lv_16sc_t** result, const lv_16sc_t* local_code, float rem_code_phase_chips, float code_phase_step_chips, float* shifts_chips, unsigned int code_length_chips, int num_out_vectors, unsigned int num_output_samples)
+{
+ lv_16sc_t** _result = result;
+ const unsigned int quarterPoints = num_output_samples / 4;
+
+ const __m128 fours = _mm_set1_ps(4.0f);
+ const __m128 rem_code_phase_chips_reg = _mm_set_ps1(rem_code_phase_chips);
+ const __m128 code_phase_step_chips_reg = _mm_set_ps1(code_phase_step_chips);
+
+ __VOLK_ATTR_ALIGNED(16) int local_code_chip_index[4];
+ int local_code_chip_index_;
+
+ const __m128i zeros = _mm_setzero_si128();
+ const __m128 code_length_chips_reg_f = _mm_set_ps1((float)code_length_chips);
+ const __m128i code_length_chips_reg_i = _mm_set1_epi32((int)code_length_chips);
+ __m128i local_code_chip_index_reg, aux_i, negatives, i;
+ __m128 aux, aux2, shifts_chips_reg, c, cTrunc, base;
+
+ for (int current_correlator_tap = 0; current_correlator_tap < num_out_vectors; current_correlator_tap++)
+ {
+ shifts_chips_reg = _mm_set_ps1((float)shifts_chips[current_correlator_tap]);
+ aux2 = _mm_sub_ps(shifts_chips_reg, rem_code_phase_chips_reg);
+ __m128 indexn = _mm_set_ps(3.0f, 2.0f, 1.0f, 0.0f);
+ for(unsigned int n = 0; n < quarterPoints; n++)
+ {
+ aux = _mm_mul_ps(code_phase_step_chips_reg, indexn);
+ aux = _mm_add_ps(aux, aux2);
+ // floor
+ aux = _mm_floor_ps(aux);
+
+ // fmod
+ c = _mm_div_ps(aux, code_length_chips_reg_f);
+ i = _mm_cvttps_epi32(c);
+ cTrunc = _mm_cvtepi32_ps(i);
+ base = _mm_mul_ps(cTrunc, code_length_chips_reg_f);
+ local_code_chip_index_reg = _mm_cvtps_epi32(_mm_sub_ps(aux, base));
+
+ negatives = _mm_cmplt_epi32(local_code_chip_index_reg, zeros);
+ aux_i = _mm_and_si128(code_length_chips_reg_i, negatives);
+ local_code_chip_index_reg = _mm_add_epi32(local_code_chip_index_reg, aux_i);
+ _mm_store_si128((__m128i*)local_code_chip_index, local_code_chip_index_reg);
+ for(unsigned int k = 0; k < 4; ++k)
+ {
+ _result[current_correlator_tap][n * 4 + k] = local_code[local_code_chip_index[k]];
+ }
+ indexn = _mm_add_ps(indexn, fours);
+ }
+ for(unsigned int n = quarterPoints * 4; n < num_output_samples; n++)
+ {
+ // resample code for current tap
+ local_code_chip_index_ = (int)floor(code_phase_step_chips * (float)n + shifts_chips[current_correlator_tap] - rem_code_phase_chips);
+ local_code_chip_index_ = local_code_chip_index_ % code_length_chips;
+ //Take into account that in multitap correlators, the shifts can be negative!
+ if (local_code_chip_index_ < 0) local_code_chip_index_ += code_length_chips;
+ _result[current_correlator_tap][n] = local_code[local_code_chip_index_];
+ }
+ }
+}
+#endif
+
+
+#ifdef LV_HAVE_SSE4_1
+#include <smmintrin.h>
+static inline void volk_gnsssdr_16ic_xn_resampler2_16ic_xn_u_sse4_1(lv_16sc_t** result, const lv_16sc_t* local_code, float rem_code_phase_chips, float code_phase_step_chips, float* shifts_chips, unsigned int code_length_chips, int num_out_vectors, unsigned int num_output_samples)
+{
+ volk_gnsssdr_16ic_xn_resampler2_16ic_xn_a_sse4_1(result, local_code, rem_code_phase_chips, code_phase_step_chips, shifts_chips, code_length_chips, num_out_vectors, num_output_samples);
+}
+
+#endif
+
+
+#ifdef LV_HAVE_SSE3
+#include <pmmintrin.h>
+static inline void volk_gnsssdr_16ic_xn_resampler2_16ic_xn_a_sse3(lv_16sc_t** result, const lv_16sc_t* local_code, float rem_code_phase_chips, float code_phase_step_chips, float* shifts_chips, unsigned int code_length_chips, int num_out_vectors, unsigned int num_output_samples)
+{
+ lv_16sc_t** _result = result;
+ const unsigned int quarterPoints = num_output_samples / 4;
+
+ const __m128 ones = _mm_set1_ps(1.0f);
+ const __m128 fours = _mm_set1_ps(4.0f);
+ const __m128 rem_code_phase_chips_reg = _mm_set_ps1(rem_code_phase_chips);
+ const __m128 code_phase_step_chips_reg = _mm_set_ps1(code_phase_step_chips);
+
+ __VOLK_ATTR_ALIGNED(16) int local_code_chip_index[4];
+ int local_code_chip_index_;
+
+ const __m128i zeros = _mm_setzero_si128();
+ const __m128 code_length_chips_reg_f = _mm_set_ps1((float)code_length_chips);
+ const __m128i code_length_chips_reg_i = _mm_set1_epi32((int)code_length_chips);
+ __m128i local_code_chip_index_reg, aux_i, negatives, i;
+ __m128 aux, aux2, shifts_chips_reg, fi, igx, j, c, cTrunc, base;
+
+ for (int current_correlator_tap = 0; current_correlator_tap < num_out_vectors; current_correlator_tap++)
+ {
+ shifts_chips_reg = _mm_set_ps1((float)shifts_chips[current_correlator_tap]);
+ aux2 = _mm_sub_ps(shifts_chips_reg, rem_code_phase_chips_reg);
+ __m128 indexn = _mm_set_ps(3.0f, 2.0f, 1.0f, 0.0f);
+ for(unsigned int n = 0; n < quarterPoints; n++)
+ {
+ aux = _mm_mul_ps(code_phase_step_chips_reg, indexn);
+ aux = _mm_add_ps(aux, aux2);
+ // floor
+ i = _mm_cvttps_epi32(aux);
+ fi = _mm_cvtepi32_ps(i);
+ igx = _mm_cmpgt_ps(fi, aux);
+ j = _mm_and_ps(igx, ones);
+ aux = _mm_sub_ps(fi, j);
+ // fmod
+ c = _mm_div_ps(aux, code_length_chips_reg_f);
+ i = _mm_cvttps_epi32(c);
+ cTrunc = _mm_cvtepi32_ps(i);
+ base = _mm_mul_ps(cTrunc, code_length_chips_reg_f);
+ local_code_chip_index_reg = _mm_cvtps_epi32(_mm_sub_ps(aux, base));
+
+ negatives = _mm_cmplt_epi32(local_code_chip_index_reg, zeros);
+ aux_i = _mm_and_si128(code_length_chips_reg_i, negatives);
+ local_code_chip_index_reg = _mm_add_epi32(local_code_chip_index_reg, aux_i);
+ _mm_store_si128((__m128i*)local_code_chip_index, local_code_chip_index_reg);
+ for(unsigned int k = 0; k < 4; ++k)
+ {
+ _result[current_correlator_tap][n * 4 + k] = local_code[local_code_chip_index[k]];
+ }
+ indexn = _mm_add_ps(indexn, fours);
+ }
+ for(unsigned int n = quarterPoints * 4; n < num_output_samples; n++)
+ {
+ // resample code for current tap
+ local_code_chip_index_ = (int)floor(code_phase_step_chips * (float)n + shifts_chips[current_correlator_tap] - rem_code_phase_chips);
+ local_code_chip_index_ = local_code_chip_index_ % code_length_chips;
+ //Take into account that in multitap correlators, the shifts can be negative!
+ if (local_code_chip_index_ < 0) local_code_chip_index_ += code_length_chips;
+ _result[current_correlator_tap][n] = local_code[local_code_chip_index_];
+ }
+ }
+}
+
+#endif
+
+
+#ifdef LV_HAVE_SSE3
+#include <pmmintrin.h>
+static inline void volk_gnsssdr_16ic_xn_resampler2_16ic_xn_u_sse3(lv_16sc_t** result, const lv_16sc_t* local_code, float rem_code_phase_chips, float code_phase_step_chips, float* shifts_chips, unsigned int code_length_chips, int num_out_vectors, unsigned int num_output_samples)
+{
+ volk_gnsssdr_16ic_xn_resampler2_16ic_xn_a_sse3(result, local_code, rem_code_phase_chips, code_phase_step_chips, shifts_chips, code_length_chips, num_out_vectors, num_output_samples);
+}
+
+#endif
+
+
+#ifdef LV_HAVE_AVX
+#include <immintrin.h>
+static inline void volk_gnsssdr_16ic_xn_resampler2_16ic_xn_a_avx(lv_16sc_t** result, const lv_16sc_t* local_code, float rem_code_phase_chips, float code_phase_step_chips, float* shifts_chips, unsigned int code_length_chips, int num_out_vectors, unsigned int num_output_samples)
+{
+ lv_16sc_t** _result = result;
+ const unsigned int avx_iters = num_output_samples / 8;
+
+ const __m256 eights = _mm256_set1_ps(8.0f);
+ const __m256 rem_code_phase_chips_reg = _mm256_set1_ps(rem_code_phase_chips);
+ const __m256 code_phase_step_chips_reg = _mm256_set1_ps(code_phase_step_chips);
+
+ __VOLK_ATTR_ALIGNED(32) int local_code_chip_index[8];
+ int local_code_chip_index_;
+
+ const __m256 zeros = _mm256_setzero_ps();
+ const __m256 code_length_chips_reg_f = _mm256_set1_ps((float)code_length_chips);
+
+ __m256i local_code_chip_index_reg, i;
+ __m256 aux, aux2, shifts_chips_reg, c, cTrunc, base, negatives;
+
+ for (int current_correlator_tap = 0; current_correlator_tap < num_out_vectors; current_correlator_tap++)
+ {
+ shifts_chips_reg = _mm256_set1_ps((float)shifts_chips[current_correlator_tap]);
+ aux2 = _mm256_sub_ps(shifts_chips_reg, rem_code_phase_chips_reg);
+ __m256 indexn = _mm256_set_ps(7.0f, 6.0f, 5.0f, 4.0f, 3.0f, 2.0f, 1.0f, 0.0f);
+ for(unsigned int n = 0; n < avx_iters; n++)
+ {
+ aux = _mm256_mul_ps(code_phase_step_chips_reg, indexn);
+ aux = _mm256_add_ps(aux, aux2);
+ // floor
+ aux = _mm256_floor_ps(aux);
+
+ // fmod
+ c = _mm256_div_ps(aux, code_length_chips_reg_f);
+ i = _mm256_cvttps_epi32(c);
+ cTrunc = _mm256_cvtepi32_ps(i);
+ base = _mm256_mul_ps(cTrunc, code_length_chips_reg_f);
+ aux = _mm256_sub_ps(aux, base);
+
+ negatives = _mm256_cmp_ps(aux, zeros, 0x01);
+ aux2 = _mm256_and_ps(code_length_chips_reg_f, negatives);
+ local_code_chip_index_reg = _mm256_cvtps_epi32(_mm256_add_ps(aux, aux2));
+ _mm256_store_si256((__m256i*)local_code_chip_index, local_code_chip_index_reg);
+ for(unsigned int k = 0; k < 8; ++k)
+ {
+ _result[current_correlator_tap][n * 8 + k] = local_code[local_code_chip_index[k]];
+ }
+ indexn = _mm256_add_ps(indexn, eights);
+ }
+
+ for(unsigned int n = avx_iters * 8; n < num_output_samples; n++)
+ {
+ // resample code for current tap
+ local_code_chip_index_ = (int)floor(code_phase_step_chips * (float)n + shifts_chips[current_correlator_tap] - rem_code_phase_chips);
+ local_code_chip_index_ = local_code_chip_index_ % code_length_chips;
+ //Take into account that in multitap correlators, the shifts can be negative!
+ if (local_code_chip_index_ < 0) local_code_chip_index_ += code_length_chips;
+ _result[current_correlator_tap][n] = local_code[local_code_chip_index_];
+ }
+ }
+ _mm256_zeroupper();
+}
+
+#endif
+
+
+#ifdef LV_HAVE_AVX
+#include <immintrin.h>
+static inline void volk_gnsssdr_16ic_xn_resampler2_16ic_xn_u_avx(lv_16sc_t** result, const lv_16sc_t* local_code, float rem_code_phase_chips, float code_phase_step_chips, float* shifts_chips, unsigned int code_length_chips, int num_out_vectors, unsigned int num_output_samples)
+{
+ volk_gnsssdr_16ic_xn_resampler2_16ic_xn_a_avx(result, local_code, rem_code_phase_chips, code_phase_step_chips, shifts_chips, code_length_chips, num_out_vectors, num_output_samples);
+}
+
+#endif
+
+
+#ifdef LV_HAVE_NEON
+#include <arm_neon.h>
+#include <volk_gnsssdr/volk_gnsssdr_neon_intrinsics.h>
+static inline void volk_gnsssdr_16ic_xn_resampler2_16ic_xn_neon(lv_16sc_t** result, const lv_16sc_t* local_code, float rem_code_phase_chips, float code_phase_step_chips, float* shifts_chips, unsigned int code_length_chips, int num_out_vectors, unsigned int num_output_samples)
+{
+ lv_16sc_t** _result = result;
+ const unsigned int neon_iters = num_output_samples / 4;
+ const int32x4_t ones = vdupq_n_s32(1);
+ const float32x4_t fours = vdupq_n_f32(4.0f);
+ const float32x4_t rem_code_phase_chips_reg = vdupq_n_f32(rem_code_phase_chips);
+ const float32x4_t code_phase_step_chips_reg = vdupq_n_f32(code_phase_step_chips);
+
+ __attribute__((aligned(16))) int local_code_chip_index[4];
+ int local_code_chip_index_;
+
+ const int32x4_t zeros = vdupq_n_s32(0);
+ const float32x4_t code_length_chips_reg_f = vdupq_n_f32((float)code_length_chips);
+ const int32x4_t code_length_chips_reg_i = vdupq_n_s32((int)code_length_chips);
+ int32x4_t local_code_chip_index_reg, aux_i, negatives, i;
+ float32x4_t aux, aux2, shifts_chips_reg, fi, c, j, cTrunc, base;
+ __attribute__((aligned(16))) float vec[4] = { 3.0f, 2.0f, 1.0f, 0.0f };
+ uint32x4_t igx;
+ float32x4_t half = vdupq_n_f32(0.5f);
+ float32x4_t sign, PlusHalf, Round;
+ for (int current_correlator_tap = 0; current_correlator_tap < num_out_vectors; current_correlator_tap++)
+ {
+ shifts_chips_reg = vdupq_n_f32((float)shifts_chips[current_correlator_tap]);
+ aux2 = vsubq_f32(shifts_chips_reg, rem_code_phase_chips_reg);
+ float32x4_t indexn = vld1q_f32((float*)vec);
+
+ for(unsigned int n = 0; n < neon_iters; n++)
+ {
+ aux = vmulq_f32(code_phase_step_chips_reg, indexn);
+ aux = vaddq_f32(aux, aux2);
+ // floor
+ i = vcvtq_s32_f32(aux);
+ fi = vcvtq_f32_s32(i);
+ igx = vcgtq_f32(fi, aux);
+ j = vreinterpretq_f32_s32(vandq_s32(vreinterpretq_s32_u32(igx), ones));
+ aux = vsubq_f32(fi, j);
+
+ // fmod
+ c = vdivq_f32(aux, code_length_chips_reg_f);
+ i = vcvtq_s32_f32(c);
+ cTrunc = vcvtq_f32_s32(i);
+ base = vmulq_f32(cTrunc, code_length_chips_reg_f);
+ aux = vsubq_f32(aux, base);
+ sign = vcvtq_f32_u32((vshrq_n_u32(vreinterpretq_u32_f32(aux), 31)));
+ PlusHalf = vaddq_f32(aux, half);
+ Round = vsubq_f32(PlusHalf, sign);
+ local_code_chip_index_reg = vcvtq_s32_f32(Round);
+
+ negatives = vreinterpretq_s32_u32(vcltq_s32(local_code_chip_index_reg, zeros));
+ aux_i = vandq_s32(code_length_chips_reg_i, negatives);
+ local_code_chip_index_reg = vaddq_s32(local_code_chip_index_reg, aux_i);
+ vst1q_s32((int32_t*)local_code_chip_index, local_code_chip_index_reg);
+ for(unsigned int k = 0; k < 4; ++k)
+ {
+ _result[current_correlator_tap][n * 4 + k] = local_code[local_code_chip_index[k]];
+ }
+ indexn = vaddq_f32(indexn, fours);
+ }
+ for(unsigned int n = neon_iters * 4; n < num_output_samples; n++)
+ {
+ // resample code for current tap
+ local_code_chip_index_ = (int)floor(code_phase_step_chips * (float)n + shifts_chips[current_correlator_tap] - rem_code_phase_chips);
+ local_code_chip_index_ = local_code_chip_index_ % code_length_chips;
+ //Take into account that in multitap correlators, the shifts can be negative!
+ if (local_code_chip_index_ < 0) local_code_chip_index_ += code_length_chips;
+ _result[current_correlator_tap][n] = local_code[local_code_chip_index_];
+ }
+ }
+}
+
+#endif
+
+
+#endif /*INCLUDED_volk_gnsssdr_16ic_xn_resampler_16ic_xn_H*/
+
diff --git a/src/algorithms/libs/volk_gnsssdr_module/volk_gnsssdr/kernels/volk_gnsssdr/volk_gnsssdr_16ic_xn_resampler_16ic_xn.h b/src/algorithms/libs/volk_gnsssdr_module/volk_gnsssdr/kernels/volk_gnsssdr/volk_gnsssdr_16ic_xn_resampler_16ic_xn.h
index e84f445..9478a4f 100644
--- a/src/algorithms/libs/volk_gnsssdr_module/volk_gnsssdr/kernels/volk_gnsssdr/volk_gnsssdr_16ic_xn_resampler_16ic_xn.h
+++ b/src/algorithms/libs/volk_gnsssdr_module/volk_gnsssdr/kernels/volk_gnsssdr/volk_gnsssdr_16ic_xn_resampler_16ic_xn.h
@@ -66,15 +66,9 @@
#include <volk_gnsssdr/volk_gnsssdr_common.h>
#include <volk_gnsssdr/volk_gnsssdr_complex.h>
-//#pragma STDC FENV_ACCESS ON
#ifdef LV_HAVE_GENERIC
-//int round_int( float r ) {
-// return (r > 0.0) ? (r + 0.5) : (r - 0.5);
-//}
-
-
static inline void volk_gnsssdr_16ic_xn_resampler_16ic_xn_generic(lv_16sc_t** result, const lv_16sc_t* local_code, float* rem_code_phase_chips, float code_phase_step_chips, unsigned int code_length_chips, int num_out_vectors, unsigned int num_output_samples)
{
int local_code_chip_index;
@@ -91,7 +85,6 @@ static inline void volk_gnsssdr_16ic_xn_resampler_16ic_xn_generic(lv_16sc_t** re
result[current_vector][n] = local_code[local_code_chip_index];
}
}
- //std::cout<<std::endl;
}
#endif /*LV_HAVE_GENERIC*/
@@ -102,7 +95,7 @@ static inline void volk_gnsssdr_16ic_xn_resampler_16ic_xn_generic(lv_16sc_t** re
static inline void volk_gnsssdr_16ic_xn_resampler_16ic_xn_a_sse2(lv_16sc_t** result, const lv_16sc_t* local_code, float* rem_code_phase_chips ,float code_phase_step_chips, unsigned int code_length_chips, int num_out_vectors, unsigned int num_output_samples)
{
- _MM_SET_ROUNDING_MODE (_MM_ROUND_NEAREST);//_MM_ROUND_NEAREST, _MM_ROUND_DOWN, _MM_ROUND_UP, _MM_ROUND_TOWARD_ZERO
+ _MM_SET_ROUNDING_MODE(_MM_ROUND_NEAREST);//_MM_ROUND_NEAREST, _MM_ROUND_DOWN, _MM_ROUND_UP, _MM_ROUND_TOWARD_ZERO
unsigned int number;
const unsigned int quarterPoints = num_output_samples / 4;
@@ -195,7 +188,7 @@ static inline void volk_gnsssdr_16ic_xn_resampler_16ic_xn_a_sse2(lv_16sc_t** res
static inline void volk_gnsssdr_16ic_xn_resampler_16ic_xn_u_sse2(lv_16sc_t** result, const lv_16sc_t* local_code, float* rem_code_phase_chips ,float code_phase_step_chips, unsigned int code_length_chips, int num_out_vectors, unsigned int num_output_samples)
{
- _MM_SET_ROUNDING_MODE (_MM_ROUND_NEAREST);//_MM_ROUND_NEAREST, _MM_ROUND_DOWN, _MM_ROUND_UP, _MM_ROUND_TOWARD_ZERO
+ _MM_SET_ROUNDING_MODE(_MM_ROUND_NEAREST);//_MM_ROUND_NEAREST, _MM_ROUND_DOWN, _MM_ROUND_UP, _MM_ROUND_TOWARD_ZERO
unsigned int number;
const unsigned int quarterPoints = num_output_samples / 4;
diff --git a/src/algorithms/libs/volk_gnsssdr_module/volk_gnsssdr/kernels/volk_gnsssdr/volk_gnsssdr_32fc_resamplerxnpuppet_32fc.h b/src/algorithms/libs/volk_gnsssdr_module/volk_gnsssdr/kernels/volk_gnsssdr/volk_gnsssdr_32fc_resamplerxnpuppet_32fc.h
index 53b1044..1b50d2e 100644
--- a/src/algorithms/libs/volk_gnsssdr_module/volk_gnsssdr/kernels/volk_gnsssdr/volk_gnsssdr_32fc_resamplerxnpuppet_32fc.h
+++ b/src/algorithms/libs/volk_gnsssdr_module/volk_gnsssdr/kernels/volk_gnsssdr/volk_gnsssdr_32fc_resamplerxnpuppet_32fc.h
@@ -116,7 +116,7 @@ static inline void volk_gnsssdr_32fc_resamplerxnpuppet_32fc_u_sse3(lv_32fc_t* re
result_aux[n] = (lv_32fc_t*)volk_gnsssdr_malloc(sizeof(lv_32fc_t) * num_points, volk_gnsssdr_get_alignment());
}
- volk_gnsssdr_32fc_xn_resampler_32fc_xn_a_sse3(result_aux, local_code, rem_code_phase_chips, code_phase_step_chips, shifts_chips, code_length_chips, num_out_vectors, num_points);
+ volk_gnsssdr_32fc_xn_resampler_32fc_xn_u_sse3(result_aux, local_code, rem_code_phase_chips, code_phase_step_chips, shifts_chips, code_length_chips, num_out_vectors, num_points);
memcpy((lv_32fc_t*)result, (lv_32fc_t*)result_aux[0], sizeof(lv_32fc_t) * num_points);
@@ -146,7 +146,7 @@ static inline void volk_gnsssdr_32fc_resamplerxnpuppet_32fc_u_sse4_1(lv_32fc_t*
result_aux[n] = (lv_32fc_t*)volk_gnsssdr_malloc(sizeof(lv_32fc_t) * num_points, volk_gnsssdr_get_alignment());
}
- volk_gnsssdr_32fc_xn_resampler_32fc_xn_a_sse4_1(result_aux, local_code, rem_code_phase_chips, code_phase_step_chips, shifts_chips, code_length_chips, num_out_vectors, num_points);
+ volk_gnsssdr_32fc_xn_resampler_32fc_xn_u_sse4_1(result_aux, local_code, rem_code_phase_chips, code_phase_step_chips, shifts_chips, code_length_chips, num_out_vectors, num_points);
memcpy((lv_32fc_t*)result, (lv_32fc_t*)result_aux[0], sizeof(lv_32fc_t) * num_points);
@@ -233,7 +233,7 @@ static inline void volk_gnsssdr_32fc_resamplerxnpuppet_32fc_u_avx(lv_32fc_t* res
result_aux[n] = (lv_32fc_t*)volk_gnsssdr_malloc(sizeof(lv_32fc_t) * num_points, volk_gnsssdr_get_alignment());
}
- volk_gnsssdr_32fc_xn_resampler_32fc_xn_a_avx(result_aux, local_code, rem_code_phase_chips, code_phase_step_chips, shifts_chips, code_length_chips, num_out_vectors, num_points);
+ volk_gnsssdr_32fc_xn_resampler_32fc_xn_u_avx(result_aux, local_code, rem_code_phase_chips, code_phase_step_chips, shifts_chips, code_length_chips, num_out_vectors, num_points);
memcpy((lv_32fc_t*)result, (lv_32fc_t*)result_aux[0], sizeof(lv_32fc_t) * num_points);
diff --git a/src/algorithms/libs/volk_gnsssdr_module/volk_gnsssdr/lib/kernel_tests.h b/src/algorithms/libs/volk_gnsssdr_module/volk_gnsssdr/lib/kernel_tests.h
index 2f68173..f94169c 100644
--- a/src/algorithms/libs/volk_gnsssdr_module/volk_gnsssdr/lib/kernel_tests.h
+++ b/src/algorithms/libs/volk_gnsssdr_module/volk_gnsssdr/lib/kernel_tests.h
@@ -87,6 +87,7 @@ std::vector<volk_gnsssdr_test_case_t> init_test_list(volk_gnsssdr_test_params_t
(VOLK_INIT_PUPP(volk_gnsssdr_16ic_rotatorpuppet_16ic, volk_gnsssdr_16ic_s32fc_x2_rotator_16ic, test_params_int1))
(VOLK_INIT_PUPP(volk_gnsssdr_16ic_resamplerpuppet_16ic, volk_gnsssdr_16ic_resampler_16ic, test_params))
(VOLK_INIT_PUPP(volk_gnsssdr_16ic_resamplerxnpuppet_16ic, volk_gnsssdr_16ic_xn_resampler_16ic_xn, test_params))
+ (VOLK_INIT_PUPP(volk_gnsssdr_16ic_resamplerxnpuppet2_16ic, volk_gnsssdr_16ic_xn_resampler2_16ic_xn, test_params))
(VOLK_INIT_PUPP(volk_gnsssdr_32fc_resamplerxnpuppet_32fc, volk_gnsssdr_32fc_xn_resampler_32fc_xn, test_params))
(VOLK_INIT_PUPP(volk_gnsssdr_16ic_x2_dotprodxnpuppet_16ic, volk_gnsssdr_16ic_x2_dot_prod_16ic_xn, test_params))
(VOLK_INIT_PUPP(volk_gnsssdr_16ic_x2_rotator_dotprodxnpuppet_16ic, volk_gnsssdr_16ic_x2_rotator_dot_prod_16ic_xn, test_params_int16))
--
Alioth's /usr/local/bin/git-commit-notice on /srv/git.debian.org/git/pkg-hamradio/gnss-sdr.git
More information about the pkg-hamradio-commits
mailing list