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
|
/*
* Copyright (c) 2010 The VP8 project authors. All Rights Reserved.
*
* Use of this source code is governed by a BSD-style license and patent
* grant that can be found in the LICENSE file in the root of the source
* tree. All contributing project authors may be found in the AUTHORS
* file in the root of the source tree.
*/
#include "memory.h"
#include "preproc.h"
#include "pragmas.h"
/****************************************************************************
* Macros
****************************************************************************/
#define FRAMECOUNT 7
#define ROUNDUP32(X) ( ( ( (unsigned long) X ) + 31 )&( 0xFFFFFFE0 ) )
/****************************************************************************
* Imports
****************************************************************************/
extern void vpx_get_processor_flags(int *mmx_enabled, int *xmm_enabled, int *wmt_enabled);
/****************************************************************************
* Exported Global Variables
****************************************************************************/
void (*temp_filter)(pre_proc_instance *ppi, unsigned char *s, unsigned char *d, int bytes, int strength);
/****************************************************************************
*
* ROUTINE : temp_filter_wmt
*
* INPUTS : pre_proc_instance *ppi : Pointer to pre-processor instance.
* unsigned char *s : Pointer to source frame.
* unsigned char *d : Pointer to destination frame.
* int bytes : Number of bytes to filter.
* int strength : Strength of filter to apply.
*
* OUTPUTS : None.
*
* RETURNS : void
*
* FUNCTION : Performs a closesness adjusted temporarl blur
*
* SPECIAL NOTES : Destination frame can be same as source frame.
*
****************************************************************************/
void temp_filter_wmt
(
pre_proc_instance *ppi,
unsigned char *s,
unsigned char *d,
int bytes,
int strength
)
{
int byte = 0;
unsigned char *frameptr = ppi->frame_buffer;
__declspec(align(16)) unsigned short threes[] = { 3, 3, 3, 3, 3, 3, 3, 3};
__declspec(align(16)) unsigned short sixteens[] = {16, 16, 16, 16, 16, 16, 16, 16};
if (ppi->frame == 0)
{
do
{
int i;
int frame = 0;
do
{
for (i = 0; i < 8; i++)
{
*frameptr = s[byte+i];
++frameptr;
}
++frame;
}
while (frame < FRAMECOUNT);
for (i = 0; i < 8; i++)
d[byte+i] = s[byte+i];
byte += 8;
}
while (byte < bytes);
}
else
{
int i;
int offset2 = (ppi->frame % FRAMECOUNT);
do
{
__declspec(align(16)) unsigned short counts[8];
__declspec(align(16)) unsigned short sums[8];
__asm
{
mov eax, offset2
mov edi, s // source pixels
pxor xmm1, xmm1 // accumulator
pxor xmm7, xmm7
mov esi, frameptr // accumulator
pxor xmm2, xmm2 // count
movq xmm3, QWORD PTR [edi]
movq QWORD PTR [esi+8*eax], xmm3
punpcklbw xmm3, xmm2 // xmm3 source pixels
mov ecx, FRAMECOUNT
next_frame:
movq xmm4, QWORD PTR [esi] // get frame buffer values
punpcklbw xmm4, xmm7 // xmm4 frame buffer pixels
movdqa xmm6, xmm4 // save the pixel values
psubsw xmm4, xmm3 // subtracted pixel values
pmullw xmm4, xmm4 // square xmm4
movd xmm5, strength
psrlw xmm4, xmm5 // should be strength
pmullw xmm4, threes // 3 * modifier
movdqa xmm5, sixteens // 16s
psubusw xmm5, xmm4 // 16 - modifiers
movdqa xmm4, xmm5 // save the modifiers
pmullw xmm4, xmm6 // multiplier values
paddusw xmm1, xmm4 // accumulator
paddusw xmm2, xmm5 // count
add esi, 8 // next frame
dec ecx // next set of eight pixels
jnz next_frame
movdqa counts, xmm2
psrlw xmm2, 1 // divide count by 2 for rounding
paddusw xmm1, xmm2 // rounding added in
mov frameptr, esi
movdqa sums, xmm1
}
for (i = 0; i < 8; i++)
{
int blurvalue = sums[i] * ppi->fixed_divide[counts[i]];
blurvalue >>= 16;
d[i] = blurvalue;
}
s += 8;
d += 8;
byte += 8;
}
while (byte < bytes);
}
++ppi->frame;
__asm emms
}
/****************************************************************************
*
* ROUTINE : temp_filter_mmx
*
* INPUTS : pre_proc_instance *ppi : Pointer to pre-processor instance.
* unsigned char *s : Pointer to source frame.
* unsigned char *d : Pointer to destination frame.
* int bytes : Number of bytes to filter.
* int strength : Strength of filter to apply.
*
* OUTPUTS : None.
*
* RETURNS : void
*
* FUNCTION : Performs a closesness adjusted temporarl blur
*
* SPECIAL NOTES : Destination frame can be same as source frame.
*
****************************************************************************/
void temp_filter_mmx
(
pre_proc_instance *ppi,
unsigned char *s,
unsigned char *d,
int bytes,
int strength
)
{
int byte = 0;
unsigned char *frameptr = ppi->frame_buffer;
__declspec(align(16)) unsigned short threes[] = { 3, 3, 3, 3};
__declspec(align(16)) unsigned short sixteens[] = {16, 16, 16, 16};
if (ppi->frame == 0)
{
do
{
int i;
int frame = 0;
do
{
for (i = 0; i < 4; i++)
{
*frameptr = s[byte+i];
++frameptr;
}
++frame;
}
while (frame < FRAMECOUNT);
for (i = 0; i < 4; i++)
d[byte+i] = s[byte+i];
byte += 4;
}
while (byte < bytes);
}
else
{
int i;
int offset2 = (ppi->frame % FRAMECOUNT);
do
{
__declspec(align(16)) unsigned short counts[8];
__declspec(align(16)) unsigned short sums[8];
__asm
{
mov eax, offset2
mov edi, s // source pixels
pxor mm1, mm1 // accumulator
pxor mm7, mm7
mov esi, frameptr // accumulator
pxor mm2, mm2 // count
movd mm3, DWORD PTR [edi]
movd DWORD PTR [esi+4*eax], mm3
punpcklbw mm3, mm2 // mm3 source pixels
mov ecx, FRAMECOUNT
next_frame:
movd mm4, DWORD PTR [esi] // get frame buffer values
punpcklbw mm4, mm7 // mm4 frame buffer pixels
movq mm6, mm4 // save the pixel values
psubsw mm4, mm3 // subtracted pixel values
pmullw mm4, mm4 // square mm4
movd mm5, strength
psrlw mm4, mm5 // should be strength
pmullw mm4, threes // 3 * modifier
movq mm5, sixteens // 16s
psubusw mm5, mm4 // 16 - modifiers
movq mm4, mm5 // save the modifiers
pmullw mm4, mm6 // multiplier values
paddusw mm1, mm4 // accumulator
paddusw mm2, mm5 // count
add esi, 4 // next frame
dec ecx // next set of eight pixels
jnz next_frame
movq counts, mm2
psrlw mm2, 1 // divide count by 2 for rounding
paddusw mm1, mm2 // rounding added in
mov frameptr, esi
movq sums, mm1
}
for (i = 0; i < 4; i++)
{
int blurvalue = sums[i] * ppi->fixed_divide[counts[i]];
blurvalue >>= 16;
d[i] = blurvalue;
}
s += 4;
d += 4;
byte += 4;
}
while (byte < bytes);
}
++ppi->frame;
__asm emms
}
|