1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
|
//
// Copyright 2012 Ettus Research LLC
//
// This program 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.
//
// This program 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 this program. If not, see <http://www.gnu.org/licenses/>.
//
#include "convert_common.hpp"
#include <uhd/utils/byteswap.hpp>
#include <emmintrin.h>
using namespace uhd::convert;
UHD_INLINE __m128i pack_sc32_4x_be(
const __m128 &in0, const __m128 &in1,
const __m128 &in2, const __m128 &in3,
const __m128 &scalar
){
__m128i tmpi0 = _mm_cvtps_epi32(_mm_mul_ps(in0, scalar));
tmpi0 = _mm_shuffle_epi32(tmpi0, _MM_SHUFFLE(1, 0, 3, 2));
__m128i tmpi1 = _mm_cvtps_epi32(_mm_mul_ps(in1, scalar));
tmpi1 = _mm_shuffle_epi32(tmpi1, _MM_SHUFFLE(1, 0, 3, 2));
const __m128i lo = _mm_packs_epi32(tmpi0, tmpi1);
__m128i tmpi2 = _mm_cvtps_epi32(_mm_mul_ps(in2, scalar));
tmpi2 = _mm_shuffle_epi32(tmpi2, _MM_SHUFFLE(1, 0, 3, 2));
__m128i tmpi3 = _mm_cvtps_epi32(_mm_mul_ps(in3, scalar));
tmpi3 = _mm_shuffle_epi32(tmpi3, _MM_SHUFFLE(1, 0, 3, 2));
const __m128i hi = _mm_packs_epi32(tmpi2, tmpi3);
return _mm_packs_epi16(lo, hi);
}
UHD_INLINE __m128i pack_sc32_4x_le(
const __m128 &in0, const __m128 &in1,
const __m128 &in2, const __m128 &in3,
const __m128 &scalar
){
__m128i tmpi0 = _mm_cvtps_epi32(_mm_mul_ps(in0, scalar));
tmpi0 = _mm_shuffle_epi32(tmpi0, _MM_SHUFFLE(2, 3, 0, 1));
__m128i tmpi1 = _mm_cvtps_epi32(_mm_mul_ps(in1, scalar));
tmpi1 = _mm_shuffle_epi32(tmpi1, _MM_SHUFFLE(2, 3, 0, 1));
const __m128i lo = _mm_packs_epi32(tmpi0, tmpi1);
__m128i tmpi2 = _mm_cvtps_epi32(_mm_mul_ps(in2, scalar));
tmpi2 = _mm_shuffle_epi32(tmpi2, _MM_SHUFFLE(2, 3, 0, 1));
__m128i tmpi3 = _mm_cvtps_epi32(_mm_mul_ps(in3, scalar));
tmpi3 = _mm_shuffle_epi32(tmpi3, _MM_SHUFFLE(2, 3, 0, 1));
const __m128i hi = _mm_packs_epi32(tmpi2, tmpi3);
return _mm_packs_epi16(lo, hi);
}
DECLARE_CONVERTER(fc32, 1, sc8_item32_be, 1, PRIORITY_SIMD){
const fc32_t *input = reinterpret_cast<const fc32_t *>(inputs[0]);
item32_t *output = reinterpret_cast<item32_t *>(outputs[0]);
const __m128 scalar = _mm_set_ps1(float(scale_factor));
#define convert_fc32_1_to_sc8_item32_1_bswap_guts(_al_) \
for (size_t j = 0; i+7 < nsamps; i+=8, j+=4){ \
/* load from input */ \
__m128 tmp0 = _mm_load ## _al_ ## ps(reinterpret_cast<const float *>(input+i+0)); \
__m128 tmp1 = _mm_load ## _al_ ## ps(reinterpret_cast<const float *>(input+i+2)); \
__m128 tmp2 = _mm_load ## _al_ ## ps(reinterpret_cast<const float *>(input+i+4)); \
__m128 tmp3 = _mm_load ## _al_ ## ps(reinterpret_cast<const float *>(input+i+6)); \
\
/* convert */ \
const __m128i tmpi = pack_sc32_4x_be(tmp0, tmp1, tmp2, tmp3, scalar); \
\
/* store to output */ \
_mm_storeu_si128(reinterpret_cast<__m128i *>(output+j), tmpi); \
} \
size_t i = 0;
//dispatch according to alignment
if ((size_t(input) & 0xf) == 0){
convert_fc32_1_to_sc8_item32_1_bswap_guts(_)
}
else{
convert_fc32_1_to_sc8_item32_1_bswap_guts(u_)
}
//convert remainder
const size_t num_pairs = nsamps/2;
for (size_t j = i/2; j < num_pairs; j++, i+=2){
const item32_t item = fc32_to_item32_sc8(input[i], input[i+1], scale_factor);
output[j] = uhd::byteswap(item);
}
if (nsamps != num_pairs*2){
const item32_t item = fc32_to_item32_sc8(input[nsamps-1], 0, scale_factor);
output[num_pairs] = uhd::byteswap(item);
}
}
DECLARE_CONVERTER(fc32, 1, sc8_item32_le, 1, PRIORITY_SIMD){
const fc32_t *input = reinterpret_cast<const fc32_t *>(inputs[0]);
item32_t *output = reinterpret_cast<item32_t *>(outputs[0]);
const __m128 scalar = _mm_set_ps1(float(scale_factor));
#define convert_fc32_1_to_sc8_item32_1_nswap_guts(_al_) \
for (size_t j = 0; i+7 < nsamps; i+=8, j+=4){ \
/* load from input */ \
__m128 tmp0 = _mm_load ## _al_ ## ps(reinterpret_cast<const float *>(input+i+0)); \
__m128 tmp1 = _mm_load ## _al_ ## ps(reinterpret_cast<const float *>(input+i+2)); \
__m128 tmp2 = _mm_load ## _al_ ## ps(reinterpret_cast<const float *>(input+i+4)); \
__m128 tmp3 = _mm_load ## _al_ ## ps(reinterpret_cast<const float *>(input+i+6)); \
\
/* convert */ \
const __m128i tmpi = pack_sc32_4x_le(tmp0, tmp1, tmp2, tmp3, scalar); \
\
/* store to output */ \
_mm_storeu_si128(reinterpret_cast<__m128i *>(output+j), tmpi); \
} \
size_t i = 0;
//dispatch according to alignment
if ((size_t(input) & 0xf) == 0){
convert_fc32_1_to_sc8_item32_1_nswap_guts(_)
}
else{
convert_fc32_1_to_sc8_item32_1_nswap_guts(u_)
}
//convert remainder
const size_t num_pairs = nsamps/2;
for (size_t j = i/2; j < num_pairs; j++, i+=2){
const item32_t item = fc32_to_item32_sc8(input[i], input[i+1], scale_factor);
output[j] = (item);
}
if (nsamps != num_pairs*2){
const item32_t item = fc32_to_item32_sc8(input[nsamps-1], 0, scale_factor);
output[num_pairs] = (item);
}
}
|
