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
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
|
///*****************************************************************************
//*
//* Copyright (C) 2012 Ittiam Systems Pvt Ltd, Bangalore
//*
//* Licensed under the Apache License, Version 2.0 (the "License");
//* you may not use this file except in compliance with the License.
//* You may obtain a copy of the License at:
//*
//* http://www.apache.org/licenses/LICENSE-2.0
//*
//* Unless required by applicable law or agreed to in writing, software
//* distributed under the License is distributed on an "AS IS" BASIS,
//* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
//* See the License for the specific language governing permissions and
//* limitations under the License.
//*
//*****************************************************************************/
///**
//*******************************************************************************
//* @file
//* ihevc_intra_pred_filters_dc.s
//*
//* @brief
//* contains function definitions for intra prediction dc filtering.
//* functions are coded using neon intrinsics and can be compiled using
//* rvct
//*
//* @author
//* akshaya mukund
//*
//* @par list of functions:
//*
//*
//* @remarks
//* none
//*
//*******************************************************************************
//*/
///**
//*******************************************************************************
//*
//* @brief
//* luma intraprediction filter for dc input
//*
//* @par description:
//*
//* @param[in] pu1_ref
//* uword8 pointer to the source
//*
//* @param[out] pu1_dst
//* uword8 pointer to the destination
//*
//* @param[in] src_strd
//* integer source stride
//*
//* @param[in] dst_strd
//* integer destination stride
//*
//* @param[in] pi1_coeff
//* word8 pointer to the planar coefficients
//*
//* @param[in] nt
//* size of tranform block
//*
//* @param[in] mode
//* type of filtering
//*
//* @returns
//*
//* @remarks
//* none
//*
//*******************************************************************************
//*/
//void ihevc_intra_pred_luma_dc(uword8 *pu1_ref,
// word32 src_strd,
// uword8 *pu1_dst,
// word32 dst_strd,
// word32 nt,
// word32 mode)
//
//**************variables vs registers*****************************************
//x0 => *pu1_ref
//x1 => src_strd
//x2 => *pu1_dst
//x3 => dst_strd
//stack contents from #40
// nt
// mode
// pi1_coeff
.text
.align 4
.include "ihevc_neon_macros.s"
.globl ihevc_intra_pred_luma_dc_av8
.type ihevc_intra_pred_luma_dc_av8, %function
ihevc_intra_pred_luma_dc_av8:
// stmfd sp!, {x4-x12, x14} //stack stores the values of the arguments
stp x19, x20,[sp,#-16]!
//********** testing
//mov x6, #128
//b prologue_cpy_32
//********** testing
mov x11, #2 //mov #2 to x11 (to be used to add to 2dc_val & 3dc_val)
mov x9, #0
mov v17.s[0], w11
mov v17.s[1], w9
clz w5,w4
add x6, x0, x4 //&src[nt]
sub x20, x5, #32 //log2nt
neg x5, x20
add x7, x0, x4, lsl #1 //&src[2nt]
add x8, x7, #1 //&src[2nt+1]
mvn x5, x5
add x5, x5, #1
dup v7.2s,w5
ldrb w14, [x8]
sxtw x14,w14
shl d7, d7,#32
sub x9, x7, #1 //&src[2nt-1]
sshr d7, d7,#32
mov x7, x8 //x7 also stores 2nt+1
ldrb w12, [x9]
sxtw x12,w12
add x14, x14, x12 //src[2nt+1] + src[2nt-1]
add x14, x14, x11 //src[2nt+1] + src[2nt-1] + 2
cmp x4, #4
beq dc_4
mov x10, x4 //nt
add_loop:
ld1 {v0.8b},[x6],#8 //load from src[nt]
mov x5, #0 //
ld1 {v1.8b},[x8],#8 //load from src[2nt+1]
uaddlp v2.4h, v0.8b
mov v6.s[0], w4
mov v6.s[1], w5 //store nt to accumulate
uaddlp v3.4h, v1.8b
ld1 {v0.8b},[x6],#8 //load from src[nt] (extra load for 8)
ld1 {v1.8b},[x8],#8 //load from src[2nt+1] (extra load for 8)
add v4.4h, v2.4h , v3.4h
uaddlp v5.2s, v4.4h
uadalp v6.1d, v5.2s //accumulate all inp into d6 (end for nt==8)
subs x10, x10,#8
beq epil_add_loop
core_loop_add:
uaddlp v2.4h, v0.8b
subs x10, x10,#8
uaddlp v3.4h, v1.8b
add v4.4h, v2.4h , v3.4h
ld1 {v0.8b},[x6],#8 //load from src[nt] (extra load for 16)
uaddlp v5.2s, v4.4h
ld1 {v1.8b},[x8],#8 //load from src[2nt+1] (extra load for 16)
uadalp v6.1d, v5.2s //accumulate all inp into d6
bne core_loop_add
epil_add_loop:
sshl d18, d6, d7 //(dc_val) shr by log2nt+1
cmp x4, #32
mov v28.s[0], w14
mov v28.s[1], w5 //src[2nt+1]+2+src[2nt-1] moved to d28
mov x20,#128
csel x6, x20, x6,eq
dup v16.8b, v18.b[0] //dc_val
shl d25, d18,#1 //2*dc
beq prologue_cpy_32
add d27, d25 , d28 //src[2nt+1]+2+src[2nt-1]+2dc_val
mov x20,#0
csel x6, x20, x6,ne //nt
ushr v29.4h, v27.4h,#2 //final dst[0]'s value in d15[0]
csel x10, x4, x10,ne
add d23, d25 , d18 //3*dc
sub x12, x3, x3, lsl #3 //-7*strd
add d23, d23 , d17 //3*dc + 2
add x12, x12, #8 //offset after one 8x8 block (-7*strd + 8)
dup v24.8h, v23.h[0] //3*dc + 2 (moved to all lanes)
sub x0, x3, x4 //strd - nt
prologue_col:
//0th column and 0-7 rows done here
//x8 and x9 (2nt+1+col 2nt-1-row)
mov x8, x7 //&src[2nt+1]
add x0, x0, #8 //strd - nt + 8
ld1 {v0.8b},[x8],#8 //col 1::7 load (prol)
sub x9, x9, #7 //&src[2nt-1-row]
ld1 {v1.8b},[x9] //row 7::1 (0 also) load (prol)
sub x9, x9, #8
uxtl v20.8h, v0.8b
ld1 {v6.8b},[x8] //col 8::15 load (prol extra)
add v20.8h, v20.8h , v24.8h //col 1::7 add 3dc+2 (prol)
uxtl v22.8h, v1.8b
sqshrun v2.8b, v20.8h,#2 //columns shx2 movn (prol)
uxtl v26.8h, v6.8b
add v22.8h, v22.8h , v24.8h //row 1::7 add 3dc+2 (prol)
movi d19, #0x00000000000000ff //
sqshrun v3.8b, v22.8h,#2 //rows shx2 movn (prol)
bsl v19.8b, v29.8b , v2.8b //first row with dst[0]
add v26.8h, v26.8h , v24.8h //col 8::15 add 3dc+2 (prol extra)
rev64 v3.8b, v3.8b
st1 {v19.8b},[x2], x3 //store row 0 (prol)
sshr d3, d3,#8 //row 0 shift (prol) (first value to be ignored)
movi d20, #0x00000000000000ff //byte mask row 1 (prol)
loop_again_col_row:
bsl v20.8b, v3.8b , v16.8b //row 1 (prol)
movi d21, #0x00000000000000ff //byte mask row 2 (prol)
sshr d3, d3,#8 //row 1 shift (prol)
st1 {v20.8b},[x2], x3 //store row 1 (prol)
sqshrun v4.8b, v26.8h,#2 //columns shx2 movn (prol extra)
bsl v21.8b, v3.8b , v16.8b //row 2 (prol)
movi d20, #0x00000000000000ff //byte mask row 3 (prol)
sshr d3, d3,#8 //row 2 shift (prol)
st1 {v21.8b},[x2], x3 //store row 2 (prol)
bsl v20.8b, v3.8b , v16.8b //row 3 (prol)
movi d21, #0x00000000000000ff //byte mask row 4 (prol)
sshr d3, d3,#8 //row 3 shift (prol)
st1 {v20.8b},[x2], x3 //store row 3 (prol)
bsl v21.8b, v3.8b , v16.8b //row 4 (prol)
movi d20, #0x00000000000000ff //byte mask row 5 (prol)
sshr d3, d3,#8 //row 4 shift (prol)
st1 {v21.8b},[x2], x3 //store row 4 (prol)
bsl v20.8b, v3.8b , v16.8b //row 5 (prol)
movi d21, #0x00000000000000ff //byte mask row 6 (prol)
sshr d3, d3,#8 //row 5 shift (prol)
st1 {v20.8b},[x2], x3 //store row 5 (prol)
ld1 {v1.8b},[x9] //row 8::15 load (prol extra)
bsl v21.8b, v3.8b , v16.8b //row 6 (prol)
uxtl v22.8h, v1.8b
movi d20, #0x00000000000000ff //byte mask row 7 (prol)
sshr d3, d3,#8 //row 6 shift (prol)
st1 {v21.8b},[x2], x3 //store row 6 (prol)
bsl v20.8b, v3.8b , v16.8b //row 7 (prol)
add v22.8h, v22.8h , v24.8h //row 8::15 add 3dc+2 (prol extra)
sshr d3, d3,#8 //row 7 shift (prol)
st1 {v20.8b},[x2], x12 //store row 7 (prol)
subs x10, x10, #8 //counter for cols
beq end_func
blt copy_16
movi d20, #0x00000000000000ff //byte mask row 9 (prol)
sqshrun v3.8b, v22.8h,#2 //rows shx2 movn (prol)
rev64 v3.8b, v3.8b
st1 {v4.8b},[x2], x3 //store 2nd col (for 16x16)
st1 {v16.8b},[x2], x3
st1 {v16.8b},[x2], x3
st1 {v16.8b},[x2], x3
st1 {v16.8b},[x2], x3
st1 {v16.8b},[x2], x3
st1 {v16.8b},[x2], x3
st1 {v16.8b},[x2], x0 //go to next row for 16
bsl v20.8b, v3.8b , v16.8b //row 9 (prol)
subs x10, x10, #8
st1 {v20.8b},[x2], x3 //store row 9 (prol)
sshr d3, d3,#8 //row 9 shift (prol)
movi d20, #0x00000000000000ff //byte mask row 9 (prol)
b loop_again_col_row
copy_16:
st1 {v16.8b},[x2], x3
st1 {v16.8b},[x2], x3
st1 {v16.8b},[x2], x3
st1 {v16.8b},[x2], x3
st1 {v16.8b},[x2], x3
st1 {v16.8b},[x2], x3
st1 {v16.8b},[x2], x3
st1 {v16.8b},[x2]
b end_func
prologue_cpy_32:
mov x9, #128
//sub x7, x3, #-24
add x5, x2, x3
add x8, x5, x3
add x10, x8, x3
dup v20.16b, v16.b[0]
lsl x6, x3, #2
sub x6, x6, #16
st1 {v20.16b}, [x2],#16
st1 {v20.16b}, [x5],#16
st1 {v20.16b}, [x8],#16
st1 {v20.16b}, [x10],#16
st1 {v20.16b}, [x2], x6
st1 {v20.16b}, [x5], x6
st1 {v20.16b}, [x8], x6
st1 {v20.16b}, [x10], x6
sub x9, x9, #32 //32x32 prol/epil counter dec
kernel_copy:
st1 {v20.16b}, [x2],#16
st1 {v20.16b}, [x5],#16
st1 {v20.16b}, [x8],#16
st1 {v20.16b}, [x10],#16
st1 {v20.16b}, [x2], x6
st1 {v20.16b}, [x5], x6
st1 {v20.16b}, [x8], x6
st1 {v20.16b}, [x10], x6
subs x9, x9, #32
st1 {v20.16b}, [x2],#16
st1 {v20.16b}, [x5],#16
st1 {v20.16b}, [x8],#16
st1 {v20.16b}, [x10],#16
st1 {v20.16b}, [x2], x6
st1 {v20.16b}, [x5], x6
st1 {v20.16b}, [x8], x6
st1 {v20.16b}, [x10], x6
bne kernel_copy
epilogue_copy:
st1 {v20.16b}, [x2],#16
st1 {v20.16b}, [x5],#16
st1 {v20.16b}, [x8],#16
st1 {v20.16b}, [x10],#16
st1 {v20.16b}, [x2]
st1 {v20.16b}, [x5]
st1 {v20.16b}, [x8]
st1 {v20.16b}, [x10]
b end_func
dc_4:
ld1 {v0.8b},[x6],#8 //load from src[nt]
ld1 {v1.8b},[x8],#8 //load from src[2nt+1]
uaddlp v2.4h, v0.8b
mov x5, #0 //
mov v6.s[0], w4
mov v6.s[1], w5 //store nt to accumulate
uaddlp v3.4h, v1.8b
add v4.4h, v2.4h , v3.4h
uaddlp v5.2s, v4.4h
movi d30, #0x00000000ffffffff
and v5.8b, v5.8b , v30.8b
mov v28.s[0], w14
mov v28.s[1], w5 //src[2nt+1]+2+src[2nt-1] moved to d28
add d6, d6 , d5 //accumulate all inp into d6 (end for nt==8)
sshl d18, d6, d7 //(dc_val) shr by log2nt+1
mov x8, x7 //&src[2nt+1]
shl d25, d18,#1 //2*dc
sub x9, x9, #3 //&src[2nt-1-row]
dup v16.8b, v18.b[0] //dc_val
add d27, d25 , d28 //src[2nt+1]+2+src[2nt-1]+2dc_val
ushr v29.4h, v27.4h,#2 //final dst[0]'s value in d15[0]
sub x12, x3, x3, lsl #2 //-3*strd
add d23, d25 , d18 //3*dc
add d23, d23 , d17 //3*dc + 2
add x12, x12, #4 //offset after one 4x4 block (-3*strd + 4)
dup v24.8h, v23.h[0] //3*dc + 2 (moved to all lanes)
sub x0, x3, x4 //strd - nt
ld1 {v0.8b},[x8] //col 1::3 load (prol)
ld1 {v1.8b},[x9] //row 3::1 (0 also) load (prol)
uxtl v20.8h, v0.8b
uxtl v22.8h, v1.8b
add v20.8h, v20.8h , v24.8h //col 1::7 add 3dc+2 (prol)
add v22.8h, v22.8h , v24.8h //row 1::7 add 3dc+2 (prol)
movi d19, #0x00000000000000ff //
sqshrun v2.8b, v20.8h,#2 //columns shx2 movn (prol)
movi d20, #0x00000000000000ff //byte mask row 1 (prol)
sqshrun v3.8b, v22.8h,#2 //rows shx2 movn (prol)
bsl v19.8b, v29.8b , v2.8b //first row with dst[0]
rev64 v3.8b, v3.8b
st1 {v19.s}[0],[x2], x3 //store row 0 (prol)
sshr d3, d3,#40 //row 0 shift (prol) (first value to be ignored)
movi d21, #0x00000000000000ff //byte mask row 2 (prol)
bsl v20.8b, v3.8b , v16.8b //row 1 (prol)
sshr d3, d3,#8 //row 1 shift (prol)
st1 {v20.s}[0],[x2], x3 //store row 1 (prol)
bsl v21.8b, v3.8b , v16.8b //row 2 (prol)
movi d20, #0x00000000000000ff //byte mask row 3 (prol)
sshr d3, d3,#8 //row 2 shift (prol)
st1 {v21.s}[0],[x2], x3 //store row 2 (prol)
bsl v20.8b, v3.8b , v16.8b //row 3 (prol)
st1 {v20.s}[0],[x2] //store row 3 (prol)
epilogue_end:
end_func:
// ldmfd sp!,{x4-x12,x15} //reload the registers from sp
ldp x19, x20,[sp],#16
ret
|
