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
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
|
/*
* Copyright © 2022 Rémi Denis-Courmont.
*
* This file is part of FFmpeg.
*
* FFmpeg 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 2.1 of the License, or (at your option) any later version.
*
* FFmpeg 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 FFmpeg; if not, write to the Free Software
* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
*/
#include "asm.S"
// (a0) = (a1) * (a2) [0..a3-1]
func ff_vector_fmul_rvv, zve32f, zba
lpad 0
1:
vsetvli t0, a3, e32, m8, ta, ma
vle32.v v16, (a1)
sub a3, a3, t0
vle32.v v24, (a2)
sh2add a1, t0, a1
vfmul.vv v16, v16, v24
sh2add a2, t0, a2
vse32.v v16, (a0)
sh2add a0, t0, a0
bnez a3, 1b
ret
endfunc
// (a0) += (a1) * fa0 [0..a2-1]
func ff_vector_fmac_scalar_rvv, zve32f, zba
lpad 0
NOHWF fmv.w.x fa0, a2
NOHWF mv a2, a3
1:
vsetvli t0, a2, e32, m8, ta, ma
slli t1, t0, 2
vle32.v v24, (a1)
sub a2, a2, t0
vle32.v v16, (a0)
sh2add a1, t0, a1
vfmacc.vf v16, fa0, v24
vse32.v v16, (a0)
sh2add a0, t0, a0
bnez a2, 1b
ret
endfunc
// (a0) = (a1) * fa0 [0..a2-1]
func ff_vector_fmul_scalar_rvv, zve32f, zba
lpad 0
NOHWF fmv.w.x fa0, a2
NOHWF mv a2, a3
1:
vsetvli t0, a2, e32, m8, ta, ma
vle32.v v16, (a1)
sub a2, a2, t0
vfmul.vf v16, v16, fa0
sh2add a1, t0, a1
vse32.v v16, (a0)
sh2add a0, t0, a0
bnez a2, 1b
ret
endfunc
func ff_vector_fmul_window_rvv, zve32f, zba
lpad 0
// a0: dst, a1: src0, a2: src1, a3: window, a4: length
// e16/m2 and e32/m4 are possible but slower due to gather.
vsetvli t0, zero, e16, m1, ta, ma
sh2add a2, a4, a2
vid.v v0
sh3add t3, a4, a3
vadd.vi v0, v0, 1
sh3add t0, a4, a0
1:
vsetvli t2, a4, e16, m1, ta, ma
slli t4, t2, 2
vrsub.vx v2, v0, t2
sub t3, t3, t4
vsetvli zero, zero, e32, m2, ta, ma
sub a2, a2, t4
vle32.v v8, (t3)
sub t0, t0, t4
vle32.v v4, (a2)
sub a4, a4, t2
vrgatherei16.vv v28, v8, v2
vle32.v v16, (a1)
add a1, a1, t4
vrgatherei16.vv v20, v4, v2
vle32.v v24, (a3)
add a3, a3, t4
vfmul.vv v12, v16, v28
vfmul.vv v16, v16, v24
vfnmsac.vv v12, v20, v24
vfmacc.vv v16, v20, v28
vrgatherei16.vv v8, v16, v2
vse32.v v12, (a0)
add a0, a0, t4
vse32.v v8, (t0)
bnez a4, 1b
ret
endfunc
// (a0) = (a1) * (a2) + (a3) [0..a4-1]
func ff_vector_fmul_add_rvv, zve32f, zba
lpad 0
1:
vsetvli t0, a4, e32, m8, ta, ma
vle32.v v8, (a1)
sub a4, a4, t0
vle32.v v16, (a2)
sh2add a1, t0, a1
vle32.v v24, (a3)
sh2add a2, t0, a2
vfmadd.vv v8, v16, v24
sh2add a3, t0, a3
vse32.v v8, (a0)
sh2add a0, t0, a0
bnez a4, 1b
ret
endfunc
// TODO factor vrsub, separate last iteration?
// (a0) = (a1) * reverse(a2) [0..a3-1]
func ff_vector_fmul_reverse_rvv, zve32f, zba
lpad 0
// e16/m4 and e32/m8 are possible but slower due to gather.
vsetvli t0, zero, e16, m1, ta, ma
sh2add a2, a3, a2
vid.v v0
vadd.vi v0, v0, 1
1:
vsetvli t0, a3, e16, m1, ta, ma
slli t1, t0, 2
vrsub.vx v4, v0, t0 // v4[i] = [VL-1, VL-2... 1, 0]
sub a2, a2, t1
vsetvli zero, zero, e32, m2, ta, ma
vle32.v v8, (a2)
sub a3, a3, t0
vle32.v v16, (a1)
add a1, a1, t1
vrgatherei16.vv v24, v8, v4 // v24 = reverse(v8)
vfmul.vv v16, v16, v24
vse32.v v16, (a0)
add a0, a0, t1
bnez a3, 1b
ret
endfunc
// (a0) = (a0) + (a1), (a1) = (a0) - (a1) [0..a2-1]
func ff_butterflies_float_rvv, zve32f, zba
lpad 0
1:
vsetvli t0, a2, e32, m8, ta, ma
vle32.v v16, (a0)
sub a2, a2, t0
vle32.v v24, (a1)
vfadd.vv v0, v16, v24
vfsub.vv v8, v16, v24
vse32.v v0, (a0)
sh2add a0, t0, a0
vse32.v v8, (a1)
sh2add a1, t0, a1
bnez a2, 1b
ret
endfunc
// a0 = (a0).(a1) [0..a2-1]
func ff_scalarproduct_float_rvv, zve32f, zba
lpad 0
vsetvli t0, zero, e32, m8, ta, ma
vmv.v.x v8, zero
vmv.s.x v0, zero
1:
vsetvli t0, a2, e32, m8, tu, ma
vle32.v v16, (a0)
sub a2, a2, t0
vle32.v v24, (a1)
sh2add a0, t0, a0
vfmacc.vv v8, v16, v24
sh2add a1, t0, a1
bnez a2, 1b
vsetvli t0, zero, e32, m8, ta, ma
vfredusum.vs v0, v8, v0
vfmv.f.s fa0, v0
NOHWF fmv.x.w a0, fa0
ret
endfunc
// (a0) = (a1) * (a2) [0..a3-1]
func ff_vector_dmul_rvv, zve64d, zba
lpad 0
1:
vsetvli t0, a3, e64, m8, ta, ma
vle64.v v16, (a1)
sub a3, a3, t0
vle64.v v24, (a2)
sh3add a1, t0, a1
vfmul.vv v16, v16, v24
sh3add a2, t0, a2
vse64.v v16, (a0)
sh3add a0, t0, a0
bnez a3, 1b
ret
endfunc
// (a0) += (a1) * fa0 [0..a2-1]
func ff_vector_dmac_scalar_rvv, zve64d, zba
lpad 0
NOHWD fmv.d.x fa0, a2
NOHWD mv a2, a3
1:
vsetvli t0, a2, e64, m8, ta, ma
vle64.v v24, (a1)
sub a2, a2, t0
vle64.v v16, (a0)
sh3add a1, t0, a1
vfmacc.vf v16, fa0, v24
vse64.v v16, (a0)
sh3add a0, t0, a0
bnez a2, 1b
ret
endfunc
// (a0) = (a1) * fa0 [0..a2-1]
func ff_vector_dmul_scalar_rvv, zve64d, zba
lpad 0
NOHWD fmv.d.x fa0, a2
NOHWD mv a2, a3
1:
vsetvli t0, a2, e64, m8, ta, ma
vle64.v v16, (a1)
sub a2, a2, t0
vfmul.vf v16, v16, fa0
sh3add a1, t0, a1
vse64.v v16, (a0)
sh3add a0, t0, a0
bnez a2, 1b
ret
endfunc
func ff_scalarproduct_double_rvv, zve64f, zba
lpad 0
vsetvli t0, zero, e64, m8, ta, ma
vmv.v.x v8, zero
vmv.s.x v0, zero
1:
vsetvli t0, a2, e64, m8, tu, ma
vle64.v v16, (a0)
sub a2, a2, t0
vle64.v v24, (a1)
sh3add a0, t0, a0
vfmacc.vv v8, v16, v24
sh3add a1, t0, a1
bnez a2, 1b
vsetvli t0, zero, e64, m8, ta, ma
vfredusum.vs v0, v8, v0
vfmv.f.s fa0, v0
NOHWD fmv.x.w a0, fa0
ret
endfunc
|
