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
|
@/*****************************************************************************
@*
@* 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_inter_pred_chroma_vert_neon_w16inp_w16out_neon.s
@*
@* @brief
@* contains function definitions for inter prediction interpolation.
@* functions are coded using neon intrinsics and can be compiled using
@* rvct
@*
@* @author
@* yogeswaran rs / parthiban
@*
@* @par list of functions:
@*
@*
@* @remarks
@* none
@*
@*******************************************************************************
@*/
@/**
@/**
@*******************************************************************************
@*
@* @brief
@* chroma interprediction filter for 16bit vertical input and output.
@*
@* @par description:
@* applies a vertical filter with coefficients pointed to by 'pi1_coeff' to
@* the elements pointed by 'pu1_src' and writes to the location pointed by
@* 'pu1_dst' input is 16 bits the filter output is downshifted by 6 and
@* 8192 is subtracted to store it as a 16 bit number the output is used as
@* a input to weighted prediction assumptions : the function is optimized
@* considering the fact width and height are multiple of 2.
@*
@* @param[in] pi2_src
@* word16 pointer to the source
@*
@* @param[out] pi2_dst
@* word16 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 filter coefficients
@*
@* @param[in] ht
@* integer height of the array
@*
@* @param[in] wd
@* integer width of the array
@*
@* @returns
@*
@* @remarks
@* none
@*
@*******************************************************************************
@*/
@void ihevc_inter_pred_chroma_vert_w16inp_w16out(word16 *pi2_src,
@ word16 *pi2_dst,
@ word32 src_strd,
@ word32 dst_strd,
@ word8 *pi1_coeff,
@ word32 ht,
@ word32 wd)
@**************variables vs registers*****************************************
@r0 => *pu1_src
@r1 => *pi2_dst
@r2 => src_strd
@r3 => dst_strd
.equ coeff_offset, 104
.equ ht_offset, 108
.equ wd_offset, 112
.text
.align 4
.globl ihevc_inter_pred_chroma_vert_w16inp_w16out_a9q
.type ihevc_inter_pred_chroma_vert_w16inp_w16out_a9q, %function
ihevc_inter_pred_chroma_vert_w16inp_w16out_a9q:
stmfd sp!, {r4-r12, r14} @stack stores the values of the arguments
vpush {d8 - d15}
ldr r4, [sp,#coeff_offset] @loads pi1_coeff
ldr r6, [sp,#wd_offset] @wd
lsl r2,r2,#1 @src_strd = 2* src_strd
ldr r5,[sp,#ht_offset] @loads ht
vld1.8 {d0},[r4] @loads pi1_coeff
sub r4,r0,r2 @pu1_src - src_strd
vmovl.s8 q0,d0 @long the value
tst r6,#3 @checks wd == 2
vdup.16 d12,d0[0] @coeff_0
vdup.16 d13,d0[1] @coeff_1
vdup.16 d14,d0[2] @coeff_2
vdup.16 d15,d0[3] @coeff_3
bgt core_loop_ht_2 @jumps to loop handles wd 2
tst r5,#3 @checks ht == mul of 4
beq core_loop_ht_4 @jumps to loop handles ht mul of 4
core_loop_ht_2:
lsl r7,r2,#1 @2*src_strd
lsl r3,r3,#1 @2*dst_strd
lsl r9,r6,#2 @4*wd
sub r6,r3,r6,lsl #1 @2*dst_strd - 2*wd
sub r8,r7,r9 @2*src_strd - 4*wd
mov r12,r9 @4wd
inner_loop_ht_2:
add r0,r4,r2 @increments pi2_src
vld1.16 {d0},[r4]! @loads pu1_src
vmull.s16 q0,d0,d12 @vmull_s16(src_tmp1, coeff_0)
subs r12,r12,#8 @2wd + 8
vld1.16 {d2},[r0],r2 @loads pi2_src
vmull.s16 q4,d2,d12 @vmull_s16(src_tmp2, coeff_0)
vld1.16 {d3},[r0],r2 @loads pi2_src
vmlal.s16 q0,d2,d13
vld1.16 {d6},[r0],r2
vmlal.s16 q4,d3,d13
vld1.16 {d2},[r0]
add r7,r1,r3 @pu1_dst + dst_strd
vmlal.s16 q0,d3,d14
vmlal.s16 q4,d6,d14
vmlal.s16 q0,d6,d15
vmlal.s16 q4,d2,d15
vqshrn.s32 d0,q0,#6 @right shift
vqshrn.s32 d30,q4,#6 @right shift
vst1.32 {d0},[r1]! @stores the loaded value
vst1.32 {d30},[r7] @stores the loaded value
bgt inner_loop_ht_2 @inner loop -again
@inner loop ends
subs r5,r5,#2 @increments ht
add r1,r1,r6,lsl #1 @pu1_dst += 2*dst_strd - 2*wd
mov r12,r9 @4wd
add r4,r4,r8 @pi1_src_tmp1 += 2*src_strd - 4*wd
bgt inner_loop_ht_2 @loop again
b end_loops @jumps to end
core_loop_ht_4:
lsl r7,r2,#2 @2*src_strd
lsl r10,r3,#2 @2*dst_strd
mov r11,r6,lsr #1 @divide by 2
sub lr,r10,r6,lsl #1 @2*dst_strd - 2*wd
sub r8,r7,r6,lsl #2 @2*src_strd - 4*wd
mul r12,r5,r11 @multiply height by width
sub r12,#4 @subtract by one for epilog
mov r11,r6,lsl #1 @2*wd
lsl r3,r3,#1 @2*dst_strd
prolog:
add r0,r4,r2 @increments pi2_src
vld1.16 {d0},[r4]! @loads pu1_src
vld1.16 {d1},[r0],r2 @loads pi2_src
subs r11,r11,#4
vld1.16 {d2},[r0],r2 @loads pi2_src
vmull.s16 q15,d0,d12 @vmull_s16(src_tmp1, coeff_0)
vld1.16 {d3},[r0],r2
vmlal.s16 q15,d1,d13
vmlal.s16 q15,d2,d14
add r9,r1,r3 @pu1_dst + dst_strd
vmlal.s16 q15,d3,d15
vld1.16 {d4},[r0],r2
vmull.s16 q14,d1,d12 @vmull_s16(src_tmp2, coeff_0)
addle r4,r4,r8
movle r11,r6,lsl #1
vmlal.s16 q14,d2,d13
vmlal.s16 q14,d3,d14
vld1.s16 {d5},[r0],r2
vmlal.s16 q14,d4,d15
vqshrn.s32 d30,q15,#6 @right shift
vld1.s16 {d6},[r0],r2
vmull.s16 q13,d2,d12 @vmull_s16(src_tmp2, coeff_0)
vmlal.s16 q13,d3,d13
vmlal.s16 q13,d4,d14
add r0,r4,r2
vld1.16 {d0},[r4]! @loads pu1_src
vmlal.s16 q13,d5,d15
vqshrn.s32 d28,q14,#6 @right shift
vld1.16 {d1},[r0],r2 @loads pi2_src
vmull.s16 q12,d3,d12 @vmull_s16(src_tmp2, coeff_0)
vst1.32 {d30},[r1]! @stores the loaded value
vmlal.s16 q12,d4,d13
vld1.16 {d2},[r0],r2 @loads pi2_src
vmlal.s16 q12,d5,d14
vld1.16 {d3},[r0],r2
vmlal.s16 q12,d6,d15
addle r1,r1,lr,lsl #1
vqshrn.s32 d26,q13,#6 @right shift
subs r12,r12,#4
beq epilog @jumps to epilog
kernel_4:
vmull.s16 q15,d0,d12 @vmull_s16(src_tmp1, coeff_0)
subs r11,r11,#4
vmlal.s16 q15,d1,d13
vst1.32 {d28},[r9],r3 @stores the loaded value
vmlal.s16 q15,d2,d14
vmlal.s16 q15,d3,d15
vqshrn.s32 d24,q12,#6 @right shift
vld1.16 {d4},[r0],r2
vmull.s16 q14,d1,d12 @vmull_s16(src_tmp2, coeff_0)
vmlal.s16 q14,d2,d13
vmlal.s16 q14,d3,d14
vmlal.s16 q14,d4,d15
vst1.32 {d26},[r9],r3 @stores the loaded value
addle r4,r4,r8
movle r11,r6,lsl #1
vqshrn.s32 d30,q15,#6 @right shift
vld1.s16 {d5},[r0],r2
vmull.s16 q13,d2,d12 @vmull_s16(src_tmp2, coeff_0)
vld1.s16 {d6},[r0],r2
vmlal.s16 q13,d3,d13
vst1.32 {d24},[r9] @stores the loaded value
add r0,r4,r2
vmlal.s16 q13,d4,d14
vld1.16 {d0},[r4]! @loads pu1_src
vmlal.s16 q13,d5,d15
vqshrn.s32 d28,q14,#6 @right shift
vld1.16 {d1},[r0],r2 @loads pi2_src
vmull.s16 q12,d3,d12 @vmull_s16(src_tmp2, coeff_0)
vld1.16 {d2},[r0],r2 @loads pi2_src
vmlal.s16 q12,d4,d13
add r9,r1,r3 @pu1_dst + dst_strd
vld1.16 {d3},[r0],r2
vmlal.s16 q12,d5,d14
vst1.32 {d30},[r1]! @stores the loaded value
vmlal.s16 q12,d6,d15
vqshrn.s32 d26,q13,#6 @right shift
addle r1,r1,lr,lsl #1
subs r12,r12,#4
bgt kernel_4 @jumps to kernel_4
epilog:
vmull.s16 q15,d0,d12 @vmull_s16(src_tmp1, coeff_0)
vst1.32 {d28},[r9],r3 @stores the loaded value
vmlal.s16 q15,d1,d13
vmlal.s16 q15,d2,d14
vmlal.s16 q15,d3,d15
vqshrn.s32 d24,q12,#6 @right shift
vmull.s16 q14,d1,d12 @vmull_s16(src_tmp2, coeff_0)
vld1.16 {d4},[r0],r2
vmlal.s16 q14,d2,d13
vst1.32 {d26},[r9],r3 @stores the loaded value
vmlal.s16 q14,d3,d14
vmlal.s16 q14,d4,d15
vqshrn.s32 d30,q15,#6 @right shift
vmull.s16 q13,d2,d12 @vmull_s16(src_tmp2, coeff_0)
vld1.s16 {d5},[r0],r2
vmlal.s16 q13,d3,d13
vmlal.s16 q13,d4,d14
vmlal.s16 q13,d5,d15
vqshrn.s32 d28,q14,#6 @right shift
vst1.32 {d24},[r9] @stores the loaded value
vmull.s16 q12,d3,d12 @vmull_s16(src_tmp2, coeff_0)
vmlal.s16 q12,d4,d13
add r9,r1,r3 @pu1_dst + dst_strd
vld1.s16 {d6},[r0],r2
vmlal.s16 q12,d5,d14
vmlal.s16 q12,d6,d15
vst1.32 {d30},[r1]! @stores the loaded value
vqshrn.s32 d26,q13,#6 @right shift
vst1.32 {d28},[r9],r3 @stores the loaded value
vqshrn.s32 d24,q12,#6 @right shift
vst1.32 {d26},[r9],r3 @stores the loaded value
vst1.32 {d24},[r9] @stores the loaded value
end_loops:
vpop {d8 - d15}
ldmfd sp!,{r4-r12,r15} @reload the registers from sp
|