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
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
|
/*
* Copyright (C) 2020 Linux Studio Plugins Project <https://lsp-plug.in/>
* (C) 2020 Vladimir Sadovnikov <sadko4u@gmail.com>
*
* This file is part of lsp-dsp-lib
* Created on: 31 мар. 2020 г.
*
* lsp-dsp-lib is free software: you can redistribute it and/or modify
* it under the terms of the GNU Lesser General Public License as published by
* the Free Software Foundation, either version 3 of the License, or
* any later version.
*
* lsp-dsp-lib 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 Lesser General Public License for more details.
*
* You should have received a copy of the GNU Lesser General Public License
* along with lsp-dsp-lib. If not, see <https://www.gnu.org/licenses/>.
*/
#ifndef PRIVATE_DSP_ARCH_AARCH64_ASIMD_FFT_PBUTTERFLY_H_
#define PRIVATE_DSP_ARCH_AARCH64_ASIMD_FFT_PBUTTERFLY_H_
#ifndef PRIVATE_DSP_ARCH_AARCH64_ASIMD_IMPL
#error "This header should not be included directly"
#endif /* PRIVATE_DSP_ARCH_AARCH64_ASIMD_IMPL */
namespace lsp
{
namespace asimd
{
#define PBUTTERFLY_RANK3(op1, op2) \
/* 8x butterflies */ \
__ASM_EMIT("ldp q28, q29, [%[XFFT_A], #0x00]") /* q28 = wr1, q29 = wr2 */ \
__ASM_EMIT("ldp q30, q31, [%[XFFT_A], #0x20]") /* q30 = wi1, q31 = wi2 */ \
__ASM_EMIT("subs %[blocks], %[blocks], #2") \
__ASM_EMIT("b.lo 2f") \
/* Calculate complex c = w * b */ \
__ASM_EMIT("1:") \
__ASM_EMIT("ldp q0, q4, [%[dst], #0x00]") /* v0 = ar1, v4 = ai1 */ \
__ASM_EMIT("ldp q1, q5, [%[dst], #0x20]") /* v1 = br1, v5 = bi1 */ \
__ASM_EMIT("ldp q2, q6, [%[dst], #0x40]") /* v2 = ai2, v6 = ai2 */ \
__ASM_EMIT("ldp q3, q7, [%[dst], #0x60]") /* v3 = bi2, v7 = bi2 */ \
/* Calc cr and ci */ \
__ASM_EMIT("fmul v16.4s, v28.4s, v1.4s") /* v16 = wr1 * br1 */ \
__ASM_EMIT("fmul v17.4s, v29.4s, v3.4s") /* v17 = wr2 * br2 */ \
__ASM_EMIT("fmul v18.4s, v28.4s, v5.4s") /* v18 = wr1 * bi1 */ \
__ASM_EMIT("fmul v19.4s, v29.4s, v7.4s") /* v19 = wr2 * bi2 */ \
__ASM_EMIT(op1 " v16.4s, v30.4s, v5.4s") /* v16 = wr1 * br1 +- wi1 * bi1 = cr1 */ \
__ASM_EMIT(op1 " v17.4s, v31.4s, v7.4s") /* v17 = wr2 * br2 +- wi2 * bi2 = cr2 */ \
__ASM_EMIT(op2 " v18.4s, v30.4s, v1.4s") /* v18 = wr1 * bi1 -+ wi1 * br1 = ci1 */ \
__ASM_EMIT(op2 " v19.4s, v31.4s, v3.4s") /* v19 = wr1 * bi1 -+ wi2 * br2 = ci2 */ \
/* Apply butterfly */ \
__ASM_EMIT("fsub v1.4s, v0.4s, v16.4s") /* v1 = ar1 - cr1 */ \
__ASM_EMIT("fsub v3.4s, v2.4s, v17.4s") /* v3 = ar2 - cr2 */ \
__ASM_EMIT("fsub v5.4s, v4.4s, v18.4s") /* v5 = ai1 - ci1 */ \
__ASM_EMIT("fsub v7.4s, v6.4s, v19.4s") /* v7 = ai2 - ci2 */ \
__ASM_EMIT("fadd v0.4s, v0.4s, v16.4s") /* v0 = ar1 + cr1 */ \
__ASM_EMIT("fadd v2.4s, v2.4s, v17.4s") /* v2 = ar2 + cr2 */ \
__ASM_EMIT("fadd v4.4s, v4.4s, v18.4s") /* v4 = ai1 + ci1 */ \
__ASM_EMIT("fadd v6.4s, v6.4s, v19.4s") /* v6 = ai2 + ci2 */ \
__ASM_EMIT("stp q0, q4, [%[dst], #0x00]") \
__ASM_EMIT("stp q1, q5, [%[dst], #0x20]") \
__ASM_EMIT("stp q2, q6, [%[dst], #0x40]") \
__ASM_EMIT("stp q3, q7, [%[dst], #0x60]") \
__ASM_EMIT("subs %[blocks], %[blocks], #2") \
__ASM_EMIT("add %[dst], %[dst], #0x80") \
__ASM_EMIT("b.ge 1b") \
__ASM_EMIT("2:") \
/* 4x butterflies */ \
__ASM_EMIT("adds %[blocks], %[blocks], #1") \
__ASM_EMIT("b.lo 4f") \
__ASM_EMIT("ldp q0, q4, [%[dst], #0x00]") /* v0 = ar1, v4 = ai1 */ \
__ASM_EMIT("ldp q1, q5, [%[dst], #0x20]") /* v1 = br1, v5 = bi1 */ \
/* Calc cr and ci */ \
__ASM_EMIT("fmul v16.4s, v28.4s, v1.4s") /* v16 = wr1 * br1 */ \
__ASM_EMIT("fmul v18.4s, v28.4s, v5.4s") /* v18 = wr1 * bi1 */ \
__ASM_EMIT(op1 " v16.4s, v30.4s, v5.4s") /* v16 = wr1 * br1 +- wi1 * bi1 = cr1 */ \
__ASM_EMIT(op2 " v18.4s, v30.4s, v1.4s") /* v18 = wr1 * bi1 -+ wi1 * br1 = ci1 */ \
/* Apply butterfly */ \
__ASM_EMIT("fsub v1.4s, v0.4s, v16.4s") /* v1 = ar1 - cr1 */ \
__ASM_EMIT("fsub v5.4s, v4.4s, v18.4s") /* v5 = ai1 - ci1 */ \
__ASM_EMIT("fadd v0.4s, v0.4s, v16.4s") /* v0 = ar1 + cr1 */ \
__ASM_EMIT("fadd v4.4s, v4.4s, v18.4s") /* v4 = ai1 + ci1 */ \
__ASM_EMIT("stp q0, q4, [%[dst], #0x00]") \
__ASM_EMIT("stp q1, q5, [%[dst], #0x20]") \
__ASM_EMIT("4:")
void packed_direct_butterfly_rank3(float *dst, size_t blocks)
{
ARCH_AARCH64_ASM(
PBUTTERFLY_RANK3("fmla", "fmls")
: [dst] "+r" (dst), [blocks] "+r" (blocks)
: [XFFT_A] "r" (&XFFT_A[0])
: "cc", "memory",
"v0", "v1", "v2", "v3",
"v4", "v5", "v6", "v7",
"v16", "v17", "v18", "v19",
"v28", "v29", "v30", "v31"
);
}
void packed_reverse_butterfly_rank3(float *dst, size_t blocks)
{
ARCH_AARCH64_ASM(
PBUTTERFLY_RANK3("fmls", "fmla")
: [dst] "+r" (dst),
[blocks] "+r" (blocks)
: [XFFT_A] "r" (&XFFT_A[0])
: "cc", "memory",
"v0", "v1", "v2", "v3",
"v4", "v5", "v6", "v7",
"v16", "v17", "v18", "v19",
"v28", "v29", "v30", "v31"
);
}
#undef PBUTTERFLY_RANK3
#define BUTTERFLY_RANK4(op1, op2) \
__ASM_EMIT("1:") \
/* Initialize sub-loop */ \
__ASM_EMIT("mov %[pairs], #1") /* pairs = 1 */ \
__ASM_EMIT("ldp q28, q29, [%[XFFT_A], #0x00]") /* q28 = wr1, q29 = wr2 */ \
__ASM_EMIT("ldp q30, q31, [%[XFFT_A], #0x20]") /* q30 = wi1, q31 = wi2 */ \
__ASM_EMIT("ldp q24, q25, [%[XFFT_W], #0x00]") /* q24 = dr, q25 = di */ \
__ASM_EMIT("lsl %[pairs], %[pairs], %[rank]") /* pairs = 1 << rank */ \
__ASM_EMIT("add %[b], %[a], %[pairs], LSL #5") /* b = &a[pairs*8] */ \
/* 8x butterflies */ \
/* Calculate complex c = w * b */ \
__ASM_EMIT("3:") \
__ASM_EMIT("ldp q0, q2, [%[a], #0x00]") /* v0 = ar1, v2 = ai1 */ \
__ASM_EMIT("ldp q1, q3, [%[a], #0x20]") /* v1 = ar2, v3 = ai2 */ \
__ASM_EMIT("ldp q4, q6, [%[b], #0x00]") /* v4 = br1, v6 = bi1 */ \
__ASM_EMIT("ldp q5, q7, [%[b], #0x20]") /* v5 = br2, v7 = bi2 */ \
/* Calc cr and ci */ \
__ASM_EMIT("fmul v16.4s, v28.4s, v4.4s") /* v16 = wr1 * br1 */ \
__ASM_EMIT("fmul v17.4s, v29.4s, v5.4s") /* v17 = wr2 * br2 */ \
__ASM_EMIT("fmul v18.4s, v28.4s, v6.4s") /* v18 = wr1 * bi1 */ \
__ASM_EMIT("fmul v19.4s, v29.4s, v7.4s") /* v19 = wr2 * bi2 */ \
__ASM_EMIT(op1 " v16.4s, v30.4s, v6.4s") /* v16 = wr1 * br1 +- wi1 * bi1 = cr1 */ \
__ASM_EMIT(op1 " v17.4s, v31.4s, v7.4s") /* v17 = wr2 * br2 +- wi2 * bi2 = cr2 */ \
__ASM_EMIT(op2 " v18.4s, v30.4s, v4.4s") /* v18 = wr1 * bi1 -+ wi1 * br1 = ci1 */ \
__ASM_EMIT(op2 " v19.4s, v31.4s, v5.4s") /* v19 = wr1 * bi1 -+ wi2 * br2 = ci2 */ \
/* Apply butterfly */ \
__ASM_EMIT("fsub v4.4s, v0.4s, v16.4s") /* v4 = ar1 - cr1 */ \
__ASM_EMIT("fsub v5.4s, v1.4s, v17.4s") /* v5 = ar2 - cr2 */ \
__ASM_EMIT("fsub v6.4s, v2.4s, v18.4s") /* v6 = ai1 - ci1 */ \
__ASM_EMIT("fsub v7.4s, v3.4s, v19.4s") /* v7 = ai2 - ci2 */ \
__ASM_EMIT("fadd v0.4s, v0.4s, v16.4s") /* v0 = ar1 + cr1 */ \
__ASM_EMIT("fadd v1.4s, v1.4s, v17.4s") /* v1 = ar2 + cr2 */ \
__ASM_EMIT("fadd v2.4s, v2.4s, v18.4s") /* v2 = ai1 + ci1 */ \
__ASM_EMIT("fadd v3.4s, v3.4s, v19.4s") /* v3 = ai2 + ci2 */ \
__ASM_EMIT("stp q0, q2, [%[a], #0x00]") \
__ASM_EMIT("stp q1, q3, [%[a], #0x20]") \
__ASM_EMIT("stp q4, q6, [%[b], #0x00]") \
__ASM_EMIT("stp q5, q7, [%[b], #0x20]") \
__ASM_EMIT("subs %[pairs], %[pairs], #2") \
__ASM_EMIT("add %[a], %[a], #0x40") \
__ASM_EMIT("add %[b], %[b], #0x40") \
__ASM_EMIT("b.le 4f") \
/* Prepare next loop: rotate angle */ \
__ASM_EMIT("fmul v16.4s, v28.4s, v25.4s") /* v16 = wr1 * di */ \
__ASM_EMIT("fmul v17.4s, v29.4s, v25.4s") /* v17 = wr2 * di */ \
__ASM_EMIT("fmul v18.4s, v30.4s, v25.4s") /* v18 = wi1 * di */ \
__ASM_EMIT("fmul v19.4s, v31.4s, v25.4s") /* v19 = wi2 * di */ \
__ASM_EMIT("fmul v28.4s, v28.4s, v24.4s") /* v28 = wr1 * dr */ \
__ASM_EMIT("fmul v29.4s, v29.4s, v24.4s") /* v29 = wr2 * dr */ \
__ASM_EMIT("fmul v30.4s, v30.4s, v24.4s") /* v30 = wi1 * dr */ \
__ASM_EMIT("fmul v31.4s, v31.4s, v24.4s") /* v31 = wi2 * dr */ \
__ASM_EMIT("fsub v28.4s, v28.4s, v18.4s") /* v28 = wr1*dr - wi1*di */ \
__ASM_EMIT("fsub v29.4s, v29.4s, v19.4s") /* v29 = wr2*dr - wi2*di */ \
__ASM_EMIT("fadd v30.4s, v30.4s, v16.4s") /* v30 = wi1*dr + wr1*di */ \
__ASM_EMIT("fadd v31.4s, v31.4s, v17.4s") /* v31 = wi2*dr + wr2*di */ \
__ASM_EMIT("b 3b") \
__ASM_EMIT("4:") \
__ASM_EMIT("subs %[blocks], %[blocks], #1") \
__ASM_EMIT("mov %[a], %[b]") \
__ASM_EMIT("b.gt 1b")
void packed_direct_butterfly_rank4p(float *dst, size_t rank, size_t blocks)
{
IF_ARCH_AARCH64(
rank -= 3;
const float *xfft_a = &XFFT_A[rank << 4];
const float *xfft_dw = &XFFT_DW[rank << 3];
float *b;
size_t pairs;
)
ARCH_AARCH64_ASM(
BUTTERFLY_RANK4("fmla", "fmls")
: [a] "+r" (dst), [b] "=&r" (b),
[blocks] "+r" (blocks), [pairs] "=&r" (pairs)
: [XFFT_A] "r" (xfft_a), [XFFT_W] "r" (xfft_dw),
[rank] "r" (rank)
: "cc", "memory",
"v0", "v1", "v2", "v3",
"v4", "v5", "v6", "v7",
"v16", "v17", "v18", "v19",
"v24", "v25",
"v28", "v29", "v30", "v31"
);
}
void packed_reverse_butterfly_rank4p(float *dst, size_t rank, size_t blocks)
{
IF_ARCH_AARCH64(
rank -= 3;
const float *xfft_a = &XFFT_A[rank << 4];
const float *xfft_dw = &XFFT_DW[rank << 3];
float *b;
size_t pairs;
)
ARCH_AARCH64_ASM(
BUTTERFLY_RANK4("fmls", "fmla")
: [a] "+r" (dst), [b] "=&r" (b),
[blocks] "+r" (blocks), [pairs] "=&r" (pairs)
: [XFFT_A] "r" (xfft_a), [XFFT_W] "r" (xfft_dw),
[rank] "r" (rank)
: "cc", "memory",
"v0", "v1", "v2", "v3",
"v4", "v5", "v6", "v7",
"v16", "v17", "v18", "v19",
"v24", "v25",
"v28", "v29", "v30", "v31"
);
}
#undef PBUTTERFLY_RANK4
}
}
#endif /* PRIVATE_DSP_ARCH_AARCH64_ASIMD_FFT_PBUTTERFLY_H_ */
|
