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
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
|
;******************************************************************************
;* x86 optimized Format Conversion Utils
;* Copyright (c) 2008 Loren Merritt
;* Copyright (c) 2012 Justin Ruggles <justin.ruggles@gmail.com>
;*
;* This file is part of Libav.
;*
;* Libav 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.
;*
;* Libav 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 Libav; if not, write to the Free Software
;* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
;******************************************************************************
%include "x86inc.asm"
%include "x86util.asm"
%include "util.asm"
SECTION_RODATA 32
pf_s32_inv_scale: times 8 dd 0x30000000
pf_s32_scale: times 8 dd 0x4f000000
pf_s16_inv_scale: times 4 dd 0x38000000
pf_s16_scale: times 4 dd 0x47000000
SECTION_TEXT
;------------------------------------------------------------------------------
; void ff_conv_s16_to_s32(int32_t *dst, const int16_t *src, int len);
;------------------------------------------------------------------------------
INIT_XMM sse2
cglobal conv_s16_to_s32, 3,3,3, dst, src, len
lea lenq, [2*lend]
lea dstq, [dstq+2*lenq]
add srcq, lenq
neg lenq
.loop:
mova m2, [srcq+lenq]
pxor m0, m0
pxor m1, m1
punpcklwd m0, m2
punpckhwd m1, m2
mova [dstq+2*lenq ], m0
mova [dstq+2*lenq+mmsize], m1
add lenq, mmsize
jl .loop
REP_RET
;------------------------------------------------------------------------------
; void ff_conv_s16_to_flt(float *dst, const int16_t *src, int len);
;------------------------------------------------------------------------------
%macro CONV_S16_TO_FLT 0
cglobal conv_s16_to_flt, 3,3,3, dst, src, len
lea lenq, [2*lend]
add srcq, lenq
lea dstq, [dstq + 2*lenq]
neg lenq
mova m2, [pf_s16_inv_scale]
ALIGN 16
.loop:
mova m0, [srcq+lenq]
S16_TO_S32_SX 0, 1
cvtdq2ps m0, m0
cvtdq2ps m1, m1
mulps m0, m2
mulps m1, m2
mova [dstq+2*lenq ], m0
mova [dstq+2*lenq+mmsize], m1
add lenq, mmsize
jl .loop
REP_RET
%endmacro
INIT_XMM sse2
CONV_S16_TO_FLT
INIT_XMM sse4
CONV_S16_TO_FLT
;------------------------------------------------------------------------------
; void ff_conv_s32_to_s16(int16_t *dst, const int32_t *src, int len);
;------------------------------------------------------------------------------
%macro CONV_S32_TO_S16 0
cglobal conv_s32_to_s16, 3,3,4, dst, src, len
lea lenq, [2*lend]
lea srcq, [srcq+2*lenq]
add dstq, lenq
neg lenq
.loop:
mova m0, [srcq+2*lenq ]
mova m1, [srcq+2*lenq+ mmsize]
mova m2, [srcq+2*lenq+2*mmsize]
mova m3, [srcq+2*lenq+3*mmsize]
psrad m0, 16
psrad m1, 16
psrad m2, 16
psrad m3, 16
packssdw m0, m1
packssdw m2, m3
mova [dstq+lenq ], m0
mova [dstq+lenq+mmsize], m2
add lenq, mmsize*2
jl .loop
%if mmsize == 8
emms
RET
%else
REP_RET
%endif
%endmacro
INIT_MMX mmx
CONV_S32_TO_S16
INIT_XMM sse2
CONV_S32_TO_S16
;------------------------------------------------------------------------------
; void ff_conv_s32_to_flt(float *dst, const int32_t *src, int len);
;------------------------------------------------------------------------------
%macro CONV_S32_TO_FLT 0
cglobal conv_s32_to_flt, 3,3,3, dst, src, len
lea lenq, [4*lend]
add srcq, lenq
add dstq, lenq
neg lenq
mova m0, [pf_s32_inv_scale]
ALIGN 16
.loop:
cvtdq2ps m1, [srcq+lenq ]
cvtdq2ps m2, [srcq+lenq+mmsize]
mulps m1, m1, m0
mulps m2, m2, m0
mova [dstq+lenq ], m1
mova [dstq+lenq+mmsize], m2
add lenq, mmsize*2
jl .loop
REP_RET
%endmacro
INIT_XMM sse2
CONV_S32_TO_FLT
%if HAVE_AVX
INIT_YMM avx
CONV_S32_TO_FLT
%endif
;------------------------------------------------------------------------------
; void ff_conv_flt_to_s16(int16_t *dst, const float *src, int len);
;------------------------------------------------------------------------------
INIT_XMM sse2
cglobal conv_flt_to_s16, 3,3,5, dst, src, len
lea lenq, [2*lend]
lea srcq, [srcq+2*lenq]
add dstq, lenq
neg lenq
mova m4, [pf_s16_scale]
.loop:
mova m0, [srcq+2*lenq ]
mova m1, [srcq+2*lenq+1*mmsize]
mova m2, [srcq+2*lenq+2*mmsize]
mova m3, [srcq+2*lenq+3*mmsize]
mulps m0, m4
mulps m1, m4
mulps m2, m4
mulps m3, m4
cvtps2dq m0, m0
cvtps2dq m1, m1
cvtps2dq m2, m2
cvtps2dq m3, m3
packssdw m0, m1
packssdw m2, m3
mova [dstq+lenq ], m0
mova [dstq+lenq+mmsize], m2
add lenq, mmsize*2
jl .loop
REP_RET
;------------------------------------------------------------------------------
; void ff_conv_flt_to_s32(int32_t *dst, const float *src, int len);
;------------------------------------------------------------------------------
%macro CONV_FLT_TO_S32 0
cglobal conv_flt_to_s32, 3,3,5, dst, src, len
lea lenq, [lend*4]
add srcq, lenq
add dstq, lenq
neg lenq
mova m4, [pf_s32_scale]
.loop:
mulps m0, m4, [srcq+lenq ]
mulps m1, m4, [srcq+lenq+1*mmsize]
mulps m2, m4, [srcq+lenq+2*mmsize]
mulps m3, m4, [srcq+lenq+3*mmsize]
cvtps2dq m0, m0
cvtps2dq m1, m1
cvtps2dq m2, m2
cvtps2dq m3, m3
mova [dstq+lenq ], m0
mova [dstq+lenq+1*mmsize], m1
mova [dstq+lenq+2*mmsize], m2
mova [dstq+lenq+3*mmsize], m3
add lenq, mmsize*4
jl .loop
REP_RET
%endmacro
INIT_XMM sse2
CONV_FLT_TO_S32
%if HAVE_AVX
INIT_YMM avx
CONV_FLT_TO_S32
%endif
;------------------------------------------------------------------------------
; void ff_conv_s16p_to_s16_2ch(int16_t *dst, int16_t *const *src, int len,
; int channels);
;------------------------------------------------------------------------------
%macro CONV_S16P_TO_S16_2CH 0
cglobal conv_s16p_to_s16_2ch, 3,4,5, dst, src0, len, src1
mov src1q, [src0q+gprsize]
mov src0q, [src0q ]
lea lenq, [2*lend]
add src0q, lenq
add src1q, lenq
lea dstq, [dstq+2*lenq]
neg lenq
.loop
mova m0, [src0q+lenq ]
mova m1, [src1q+lenq ]
mova m2, [src0q+lenq+mmsize]
mova m3, [src1q+lenq+mmsize]
SBUTTERFLY2 wd, 0, 1, 4
SBUTTERFLY2 wd, 2, 3, 4
mova [dstq+2*lenq+0*mmsize], m0
mova [dstq+2*lenq+1*mmsize], m1
mova [dstq+2*lenq+2*mmsize], m2
mova [dstq+2*lenq+3*mmsize], m3
add lenq, 2*mmsize
jl .loop
REP_RET
%endmacro
INIT_XMM sse2
CONV_S16P_TO_S16_2CH
%if HAVE_AVX
INIT_XMM avx
CONV_S16P_TO_S16_2CH
%endif
;------------------------------------------------------------------------------
; void ff_conv_s16p_to_s16_6ch(int16_t *dst, int16_t *const *src, int len,
; int channels);
;------------------------------------------------------------------------------
;------------------------------------------------------------------------------
; NOTE: In the 6-channel functions, len could be used as an index on x86-64
; instead of just a counter, which would avoid incrementing the
; pointers, but the extra complexity and amount of code is not worth
; the small gain. On x86-32 there are not enough registers to use len
; as an index without keeping two of the pointers on the stack and
; loading them in each iteration.
;------------------------------------------------------------------------------
%macro CONV_S16P_TO_S16_6CH 0
%if ARCH_X86_64
cglobal conv_s16p_to_s16_6ch, 3,8,7, dst, src0, len, src1, src2, src3, src4, src5
%else
cglobal conv_s16p_to_s16_6ch, 2,7,7, dst, src0, src1, src2, src3, src4, src5
%define lend dword r2m
%endif
mov src1q, [src0q+1*gprsize]
mov src2q, [src0q+2*gprsize]
mov src3q, [src0q+3*gprsize]
mov src4q, [src0q+4*gprsize]
mov src5q, [src0q+5*gprsize]
mov src0q, [src0q]
sub src1q, src0q
sub src2q, src0q
sub src3q, src0q
sub src4q, src0q
sub src5q, src0q
.loop:
%if cpuflag(sse2slow)
movq m0, [src0q ] ; m0 = 0, 6, 12, 18, x, x, x, x
movq m1, [src0q+src1q] ; m1 = 1, 7, 13, 19, x, x, x, x
movq m2, [src0q+src2q] ; m2 = 2, 8, 14, 20, x, x, x, x
movq m3, [src0q+src3q] ; m3 = 3, 9, 15, 21, x, x, x, x
movq m4, [src0q+src4q] ; m4 = 4, 10, 16, 22, x, x, x, x
movq m5, [src0q+src5q] ; m5 = 5, 11, 17, 23, x, x, x, x
; unpack words:
punpcklwd m0, m1 ; m0 = 0, 1, 6, 7, 12, 13, 18, 19
punpcklwd m2, m3 ; m2 = 4, 5, 10, 11, 16, 17, 22, 23
punpcklwd m4, m5 ; m4 = 2, 3, 8, 9, 14, 15, 20, 21
; blend dwords
shufps m1, m0, m2, q2020 ; m1 = 0, 1, 12, 13, 2, 3, 14, 15
shufps m0, m4, q2031 ; m0 = 6, 7, 18, 19, 4, 5, 16, 17
shufps m2, m4, q3131 ; m2 = 8, 9, 20, 21, 10, 11, 22, 23
; shuffle dwords
pshufd m0, m0, q1302 ; m0 = 4, 5, 6, 7, 16, 17, 18, 19
pshufd m1, m1, q3120 ; m1 = 0, 1, 2, 3, 12, 13, 14, 15
pshufd m2, m2, q3120 ; m2 = 8, 9, 10, 11, 20, 21, 22, 23
movq [dstq+0*mmsize/2], m1
movq [dstq+1*mmsize/2], m0
movq [dstq+2*mmsize/2], m2
movhps [dstq+3*mmsize/2], m1
movhps [dstq+4*mmsize/2], m0
movhps [dstq+5*mmsize/2], m2
add src0q, mmsize/2
add dstq, mmsize*3
sub lend, mmsize/4
%else
mova m0, [src0q ] ; m0 = 0, 6, 12, 18, 24, 30, 36, 42
mova m1, [src0q+src1q] ; m1 = 1, 7, 13, 19, 25, 31, 37, 43
mova m2, [src0q+src2q] ; m2 = 2, 8, 14, 20, 26, 32, 38, 44
mova m3, [src0q+src3q] ; m3 = 3, 9, 15, 21, 27, 33, 39, 45
mova m4, [src0q+src4q] ; m4 = 4, 10, 16, 22, 28, 34, 40, 46
mova m5, [src0q+src5q] ; m5 = 5, 11, 17, 23, 29, 35, 41, 47
; unpack words:
SBUTTERFLY2 wd, 0, 1, 6 ; m0 = 0, 1, 6, 7, 12, 13, 18, 19
; m1 = 24, 25, 30, 31, 36, 37, 42, 43
SBUTTERFLY2 wd, 2, 3, 6 ; m2 = 2, 3, 8, 9, 14, 15, 20, 21
; m3 = 26, 27, 32, 33, 38, 39, 44, 45
SBUTTERFLY2 wd, 4, 5, 6 ; m4 = 4, 5, 10, 11, 16, 17, 22, 23
; m5 = 28, 29, 34, 35, 40, 41, 46, 47
; blend dwords
shufps m6, m0, m2, q2020 ; m6 = 0, 1, 12, 13, 2, 3, 14, 15
shufps m0, m4, q2031 ; m0 = 6, 7, 18, 19, 4, 5, 16, 17
shufps m2, m4, q3131 ; m2 = 8, 9, 20, 21, 10, 11, 22, 23
SWAP 4,6 ; m4 = 0, 1, 12, 13, 2, 3, 14, 15
shufps m6, m1, m3, q2020 ; m6 = 24, 25, 36, 37, 26, 27, 38, 39
shufps m1, m5, q2031 ; m1 = 30, 31, 42, 43, 28, 29, 40, 41
shufps m3, m5, q3131 ; m3 = 32, 33, 44, 45, 34, 35, 46, 47
SWAP 5,6 ; m5 = 24, 25, 36, 37, 26, 27, 38, 39
; shuffle dwords
pshufd m0, m0, q1302 ; m0 = 4, 5, 6, 7, 16, 17, 18, 19
pshufd m2, m2, q3120 ; m2 = 8, 9, 10, 11, 20, 21, 22, 23
pshufd m4, m4, q3120 ; m4 = 0, 1, 2, 3, 12, 13, 14, 15
pshufd m1, m1, q1302 ; m1 = 28, 29, 30, 31, 40, 41, 42, 43
pshufd m3, m3, q3120 ; m3 = 32, 33, 34, 35, 44, 45, 46, 47
pshufd m5, m5, q3120 ; m5 = 24, 25, 26, 27, 36, 37, 38, 39
; shuffle qwords
punpcklqdq m6, m4, m0 ; m6 = 0, 1, 2, 3, 4, 5, 6, 7
punpckhqdq m0, m2 ; m0 = 16, 17, 18, 19, 20, 21, 22, 23
shufps m2, m4, q3210 ; m2 = 8, 9, 10, 11, 12, 13, 14, 15
SWAP 4,6 ; m4 = 0, 1, 2, 3, 4, 5, 6, 7
punpcklqdq m6, m5, m1 ; m6 = 24, 25, 26, 27, 28, 29, 30, 31
punpckhqdq m1, m3 ; m1 = 40, 41, 42, 43, 44, 45, 46, 47
shufps m3, m5, q3210 ; m3 = 32, 33, 34, 35, 36, 37, 38, 39
SWAP 5,6 ; m5 = 24, 25, 26, 27, 28, 29, 30, 31
mova [dstq+0*mmsize], m4
mova [dstq+1*mmsize], m2
mova [dstq+2*mmsize], m0
mova [dstq+3*mmsize], m5
mova [dstq+4*mmsize], m3
mova [dstq+5*mmsize], m1
add src0q, mmsize
add dstq, mmsize*6
sub lend, mmsize/2
%endif
jg .loop
REP_RET
%endmacro
INIT_XMM sse2
CONV_S16P_TO_S16_6CH
INIT_XMM sse2slow
CONV_S16P_TO_S16_6CH
%if HAVE_AVX
INIT_XMM avx
CONV_S16P_TO_S16_6CH
%endif
;-----------------------------------------------------------------------------
; void ff_conv_fltp_to_flt_6ch(float *dst, float *const *src, int len,
; int channels);
;-----------------------------------------------------------------------------
%macro CONV_FLTP_TO_FLT_6CH 0
cglobal conv_fltp_to_flt_6ch, 2,8,7, dst, src, src1, src2, src3, src4, src5, len
%if ARCH_X86_64
mov lend, r2d
%else
%define lend dword r2m
%endif
mov src1q, [srcq+1*gprsize]
mov src2q, [srcq+2*gprsize]
mov src3q, [srcq+3*gprsize]
mov src4q, [srcq+4*gprsize]
mov src5q, [srcq+5*gprsize]
mov srcq, [srcq]
sub src1q, srcq
sub src2q, srcq
sub src3q, srcq
sub src4q, srcq
sub src5q, srcq
.loop:
mova m0, [srcq ]
mova m1, [srcq+src1q]
mova m2, [srcq+src2q]
mova m3, [srcq+src3q]
mova m4, [srcq+src4q]
mova m5, [srcq+src5q]
%if cpuflag(sse4)
SBUTTERFLYPS 0, 1, 6
SBUTTERFLYPS 2, 3, 6
SBUTTERFLYPS 4, 5, 6
blendps m6, m4, m0, 1100b
movlhps m0, m2
movhlps m4, m2
blendps m2, m5, m1, 1100b
movlhps m1, m3
movhlps m5, m3
movaps [dstq ], m0
movaps [dstq+16], m6
movaps [dstq+32], m4
movaps [dstq+48], m1
movaps [dstq+64], m2
movaps [dstq+80], m5
%else ; mmx
SBUTTERFLY dq, 0, 1, 6
SBUTTERFLY dq, 2, 3, 6
SBUTTERFLY dq, 4, 5, 6
movq [dstq ], m0
movq [dstq+ 8], m2
movq [dstq+16], m4
movq [dstq+24], m1
movq [dstq+32], m3
movq [dstq+40], m5
%endif
add srcq, mmsize
add dstq, mmsize*6
sub lend, mmsize/4
jg .loop
%if mmsize == 8
emms
RET
%else
REP_RET
%endif
%endmacro
INIT_MMX mmx
CONV_FLTP_TO_FLT_6CH
INIT_XMM sse4
CONV_FLTP_TO_FLT_6CH
%if HAVE_AVX
INIT_XMM avx
CONV_FLTP_TO_FLT_6CH
%endif
|
