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
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
|
;******************************************************************************
;* SIMD optimized MPEG-4 Parametric Stereo decoding functions
;*
;* Copyright (C) 2015 James Almer
;*
;* 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 "libavutil/x86/x86util.asm"
SECTION_RODATA
ps_p1m1p1m1: dd 0, 0x80000000, 0, 0x80000000
SECTION .text
;*************************************************************************
;void ff_ps_add_squares_<opt>(float *dst, const float (*src)[2], int n);
;*************************************************************************
%macro PS_ADD_SQUARES 1
cglobal ps_add_squares, 3, 3, %1, dst, src, n
shl nd, 3
add srcq, nq
neg nq
align 16
.loop:
movaps m0, [srcq+nq]
movaps m1, [srcq+nq+mmsize]
mulps m0, m0
mulps m1, m1
HADDPS m0, m1, m2
addps m0, [dstq]
movaps [dstq], m0
add dstq, mmsize
add nq, mmsize*2
jl .loop
REP_RET
%endmacro
INIT_XMM sse
PS_ADD_SQUARES 2
INIT_XMM sse3
PS_ADD_SQUARES 3
;*******************************************************************
;void ff_ps_mul_pair_single_sse(float (*dst)[2], float (*src0)[2],
; float *src1, int n);
;*******************************************************************
INIT_XMM sse
cglobal ps_mul_pair_single, 4, 4, 4, dst, src1, src2, n
shl nd, 3
add src1q, nq
add dstq, nq
neg nq
align 16
.loop:
movu m0, [src1q+nq]
movu m1, [src1q+nq+mmsize]
mova m2, [src2q]
mova m3, m2
unpcklps m2, m2
unpckhps m3, m3
mulps m0, m2
mulps m1, m3
mova [dstq+nq], m0
mova [dstq+nq+mmsize], m1
add src2q, mmsize
add nq, mmsize*2
jl .loop
REP_RET
;***********************************************************************
;void ff_ps_stereo_interpolate_sse3(float (*l)[2], float (*r)[2],
; float h[2][4], float h_step[2][4],
; int len);
;***********************************************************************
INIT_XMM sse3
cglobal ps_stereo_interpolate, 5, 5, 6, l, r, h, h_step, n
movaps m0, [hq]
movaps m1, [h_stepq]
unpcklps m4, m0, m0
unpckhps m0, m0
unpcklps m5, m1, m1
unpckhps m1, m1
shl nd, 3
add lq, nq
add rq, nq
neg nq
align 16
.loop:
addps m4, m5
addps m0, m1
movddup m2, [lq+nq]
movddup m3, [rq+nq]
mulps m2, m4
mulps m3, m0
addps m2, m3
movsd [lq+nq], m2
movhps [rq+nq], m2
add nq, 8
jl .loop
REP_RET
;***************************************************************************
;void ps_stereo_interpolate_ipdopd_sse3(float (*l)[2], float (*r)[2],
; float h[2][4], float h_step[2][4],
; int len);
;***************************************************************************
INIT_XMM sse3
cglobal ps_stereo_interpolate_ipdopd, 5, 5, 10, l, r, h, h_step, n
movaps m0, [hq]
movaps m1, [hq+mmsize]
%if ARCH_X86_64
movaps m8, [h_stepq]
movaps m9, [h_stepq+mmsize]
%define H_STEP0 m8
%define H_STEP1 m9
%else
%define H_STEP0 [h_stepq]
%define H_STEP1 [h_stepq+mmsize]
%endif
shl nd, 3
add lq, nq
add rq, nq
neg nq
align 16
.loop:
addps m0, H_STEP0
addps m1, H_STEP1
movddup m2, [lq+nq]
movddup m3, [rq+nq]
shufps m4, m2, m2, q2301
shufps m5, m3, m3, q2301
unpcklps m6, m0, m0
unpckhps m7, m0, m0
mulps m2, m6
mulps m3, m7
unpcklps m6, m1, m1
unpckhps m7, m1, m1
mulps m4, m6
mulps m5, m7
addps m2, m3
addsubps m2, m4
addsubps m2, m5
movsd [lq+nq], m2
movhps [rq+nq], m2
add nq, 8
jl .loop
REP_RET
;***********************************************************
;void ps_hybrid_synthesis_deint_sse4(float out[2][38][64],
; float (*in)[32][2],
; int i, int len)
;***********************************************************
%macro HYBRID_SYNTHESIS_DEINT 0
cglobal ps_hybrid_synthesis_deint, 3, 7, 5, out, in, i, len, out0, out1, tmp
%if cpuflag(sse4)
%define MOVH movsd
%else
%define MOVH movlps
%endif
movsxdifnidn iq, id
mov lend, 32 << 3
lea outq, [outq+iq*4]
mov tmpd, id
shl tmpd, 8
add inq, tmpq
mov tmpd, 64
sub tmpd, id
mov id, tmpd
test id, 1
jne .loop4
test id, 2
jne .loop8
align 16
.loop16:
mov out0q, outq
mov out1q, 38*64*4
add out1q, out0q
mov tmpd, lend
.inner_loop16:
movaps m0, [inq]
movaps m1, [inq+lenq]
movaps m2, [inq+lenq*2]
movaps m3, [inq+3*32*2*4]
TRANSPOSE4x4PS 0, 1, 2, 3, 4
movaps [out0q], m0
movaps [out1q], m1
movaps [out0q+lenq], m2
movaps [out1q+lenq], m3
lea out0q, [out0q+lenq*2]
lea out1q, [out1q+lenq*2]
add inq, mmsize
sub tmpd, mmsize
jg .inner_loop16
add outq, 16
add inq, 3*32*2*4
sub id, 4
jg .loop16
RET
align 16
.loop8:
mov out0q, outq
mov out1q, 38*64*4
add out1q, out0q
mov tmpd, lend
.inner_loop8:
movaps m0, [inq]
movaps m1, [inq+lenq]
SBUTTERFLYPS 0, 1, 2
SBUTTERFLYPD 0, 1, 2
MOVH [out0q], m0
MOVH [out1q], m1
movhps [out0q+lenq], m0
movhps [out1q+lenq], m1
lea out0q, [out0q+lenq*2]
lea out1q, [out1q+lenq*2]
add inq, mmsize
sub tmpd, mmsize
jg .inner_loop8
add outq, 8
add inq, lenq
sub id, 2
jg .loop16
RET
align 16
.loop4:
mov out0q, outq
mov out1q, 38*64*4
add out1q, out0q
mov tmpd, lend
.inner_loop4:
movaps m0, [inq]
movss [out0q], m0
%if cpuflag(sse4)
extractps [out1q], m0, 1
extractps [out0q+lenq], m0, 2
extractps [out1q+lenq], m0, 3
%else
movhlps m1, m0
movss [out0q+lenq], m1
shufps m0, m0, 0xb1
movss [out1q], m0
movhlps m1, m0
movss [out1q+lenq], m1
%endif
lea out0q, [out0q+lenq*2]
lea out1q, [out1q+lenq*2]
add inq, mmsize
sub tmpd, mmsize
jg .inner_loop4
add outq, 4
sub id, 1
test id, 2
jne .loop8
cmp id, 4
jge .loop16
RET
%endmacro
INIT_XMM sse
HYBRID_SYNTHESIS_DEINT
INIT_XMM sse4
HYBRID_SYNTHESIS_DEINT
;*******************************************************************
;void ff_ps_hybrid_analysis_<opt>(float (*out)[2], float (*in)[2],
; const float (*filter)[8][2],
; int stride, int n);
;*******************************************************************
%macro PS_HYBRID_ANALYSIS_LOOP 3
movu %1, [inq+mmsize*%3]
movu m1, [inq+mmsize*(5-%3)+8]
%if cpuflag(sse3)
pshufd %2, %1, q2301
pshufd m4, m1, q0123
pshufd m1, m1, q1032
pshufd m2, [filterq+nq+mmsize*%3], q2301
addsubps %2, m4
addsubps %1, m1
%else
mova m2, [filterq+nq+mmsize*%3]
mova %2, %1
mova m4, m1
shufps %2, %2, q2301
shufps m4, m4, q0123
shufps m1, m1, q1032
shufps m2, m2, q2301
xorps m4, m7
xorps m1, m7
subps %2, m4
subps %1, m1
%endif
mulps %2, m2
mulps %1, m2
%if %3
addps m3, %2
addps m0, %1
%endif
%endmacro
%macro PS_HYBRID_ANALYSIS 0
cglobal ps_hybrid_analysis, 5, 5, 8, out, in, filter, stride, n
%if cpuflag(sse3)
%define MOVH movsd
%else
%define MOVH movlps
%endif
shl strided, 3
shl nd, 6
add filterq, nq
neg nq
mova m7, [ps_p1m1p1m1]
align 16
.loop:
PS_HYBRID_ANALYSIS_LOOP m0, m3, 0
PS_HYBRID_ANALYSIS_LOOP m5, m6, 1
PS_HYBRID_ANALYSIS_LOOP m5, m6, 2
%if cpuflag(sse3)
pshufd m3, m3, q2301
xorps m0, m7
hsubps m3, m0
pshufd m1, m3, q0020
pshufd m3, m3, q0031
addps m1, m3
movsd m2, [inq+6*8]
%else
mova m1, m3
mova m2, m0
shufps m1, m1, q2301
shufps m2, m2, q2301
subps m1, m3
addps m2, m0
unpcklps m3, m1, m2
unpckhps m1, m2
addps m1, m3
movu m2, [inq+6*8] ; faster than movlps and no risk of overread
%endif
movss m3, [filterq+nq+8*6]
SPLATD m3
mulps m2, m3
addps m1, m2
MOVH [outq], m1
add outq, strideq
add nq, 64
jl .loop
REP_RET
%endmacro
INIT_XMM sse
PS_HYBRID_ANALYSIS
INIT_XMM sse3
PS_HYBRID_ANALYSIS
|
