Chromium Code Reviews
chromiumcodereview-hr@appspot.gserviceaccount.com (chromiumcodereview-hr) | Please choose your nickname with Settings | Help | Chromium Project | Gerrit Changes | Sign out
(206)

Issues Search
    My Issues | Starred     Open | Closed | All

Side by Side Diff: third_party/libwebp/dsp/enc_mips32.c

Issue 1422493004: libwebp: update to 0.4.4 (Closed) Base URL: https://chromium.googlesource.com/chromium/src.git@master
Patch Set: Created 5 years, 1 month ago
Use n/p to move between diff chunks; N/P to move between comments. Draft comments are only viewable by you.
Jump to:
View unified diff | Download patch
« no previous file with comments | « third_party/libwebp/dsp/dsp.h ('k') | third_party/libwebp/dsp/lossless_mips32.c » ('j') | no next file with comments »
Toggle Intra-line Diffs ('i') | Expand Comments ('e') | Collapse Comments ('c') | Show Comments Hide Comments ('s')
OLDNEW
1 // Copyright 2014 Google Inc. All Rights Reserved. 1 // Copyright 2014 Google Inc. All Rights Reserved.
2 // 2 //
3 // Use of this source code is governed by a BSD-style license 3 // Use of this source code is governed by a BSD-style license
4 // that can be found in the COPYING file in the root of the source 4 // that can be found in the COPYING file in the root of the source
5 // tree. An additional intellectual property rights grant can be found 5 // tree. An additional intellectual property rights grant can be found
6 // in the file PATENTS. All contributing project authors may 6 // in the file PATENTS. All contributing project authors may
7 // be found in the AUTHORS file in the root of the source tree. 7 // be found in the AUTHORS file in the root of the source tree.
8 // ----------------------------------------------------------------------------- 8 // -----------------------------------------------------------------------------
9 // 9 //
10 // MIPS version of speed-critical encoding functions. 10 // MIPS version of speed-critical encoding functions.
(...skipping 16 matching lines...) Expand all
27 static const int kC1 = 20091 + (1 << 16); 27 static const int kC1 = 20091 + (1 << 16);
28 static const int kC2 = 35468; 28 static const int kC2 = 35468;
29 29
30 // macro for one vertical pass in ITransformOne 30 // macro for one vertical pass in ITransformOne
31 // MUL macro inlined 31 // MUL macro inlined
32 // temp0..temp15 holds tmp[0]..tmp[15] 32 // temp0..temp15 holds tmp[0]..tmp[15]
33 // A..D - offsets in bytes to load from in buffer 33 // A..D - offsets in bytes to load from in buffer
34 // TEMP0..TEMP3 - registers for corresponding tmp elements 34 // TEMP0..TEMP3 - registers for corresponding tmp elements
35 // TEMP4..TEMP5 - temporary registers 35 // TEMP4..TEMP5 - temporary registers
36 #define VERTICAL_PASS(A, B, C, D, TEMP4, TEMP0, TEMP1, TEMP2, TEMP3) \ 36 #define VERTICAL_PASS(A, B, C, D, TEMP4, TEMP0, TEMP1, TEMP2, TEMP3) \
37 "lh %[temp16], "#A"(%[temp20]) \n\t" \ 37 "lh %[temp16], " #A "(%[temp20]) \n\t" \
38 "lh %[temp18], "#B"(%[temp20]) \n\t" \ 38 "lh %[temp18], " #B "(%[temp20]) \n\t" \
39 "lh %[temp17], "#C"(%[temp20]) \n\t" \ 39 "lh %[temp17], " #C "(%[temp20]) \n\t" \
40 "lh %[temp19], "#D"(%[temp20]) \n\t" \ 40 "lh %[temp19], " #D "(%[temp20]) \n\t" \
41 "addu %["#TEMP4"], %[temp16], %[temp18] \n\t" \ 41 "addu %[" #TEMP4 "], %[temp16], %[temp18] \n\t" \
42 "subu %[temp16], %[temp16], %[temp18] \n\t" \ 42 "subu %[temp16], %[temp16], %[temp18] \n\t" \
43 "mul %["#TEMP0"], %[temp17], %[kC2] \n\t" \ 43 "mul %[" #TEMP0 "], %[temp17], %[kC2] \n\t" \
44 "mul %[temp18], %[temp19], %[kC1] \n\t" \ 44 "mul %[temp18], %[temp19], %[kC1] \n\t" \
45 "mul %[temp17], %[temp17], %[kC1] \n\t" \ 45 "mul %[temp17], %[temp17], %[kC1] \n\t" \
46 "mul %[temp19], %[temp19], %[kC2] \n\t" \ 46 "mul %[temp19], %[temp19], %[kC2] \n\t" \
47 "sra %["#TEMP0"], %["#TEMP0"], 16 \n\n" \ 47 "sra %[" #TEMP0 "], %[" #TEMP0 "], 16 \n\n" \
48 "sra %[temp18], %[temp18], 16 \n\n" \ 48 "sra %[temp18], %[temp18], 16 \n\n" \
49 "sra %[temp17], %[temp17], 16 \n\n" \ 49 "sra %[temp17], %[temp17], 16 \n\n" \
50 "sra %[temp19], %[temp19], 16 \n\n" \ 50 "sra %[temp19], %[temp19], 16 \n\n" \
51 "subu %["#TEMP2"], %["#TEMP0"], %[temp18] \n\t" \ 51 "subu %[" #TEMP2 "], %[" #TEMP0 "], %[temp18] \n\t" \
52 "addu %["#TEMP3"], %[temp17], %[temp19] \n\t" \ 52 "addu %[" #TEMP3 "], %[temp17], %[temp19] \n\t" \
53 "addu %["#TEMP0"], %["#TEMP4"], %["#TEMP3"] \n\t" \ 53 "addu %[" #TEMP0 "], %[" #TEMP4 "], %[" #TEMP3 "] \n\t" \
54 "addu %["#TEMP1"], %[temp16], %["#TEMP2"] \n\t" \ 54 "addu %[" #TEMP1 "], %[temp16], %[" #TEMP2 "] \n\t" \
55 "subu %["#TEMP2"], %[temp16], %["#TEMP2"] \n\t" \ 55 "subu %[" #TEMP2 "], %[temp16], %[" #TEMP2 "] \n\t" \
56 "subu %["#TEMP3"], %["#TEMP4"], %["#TEMP3"] \n\t" 56 "subu %[" #TEMP3 "], %[" #TEMP4 "], %[" #TEMP3 "] \n\t"
57 57
58 // macro for one horizontal pass in ITransformOne 58 // macro for one horizontal pass in ITransformOne
59 // MUL and STORE macros inlined 59 // MUL and STORE macros inlined
60 // a = clip_8b(a) is replaced with: a = max(a, 0); a = min(a, 255) 60 // a = clip_8b(a) is replaced with: a = max(a, 0); a = min(a, 255)
61 // temp0..temp15 holds tmp[0]..tmp[15] 61 // temp0..temp15 holds tmp[0]..tmp[15]
62 // A..D - offsets in bytes to load from ref and store to dst buffer 62 // A..D - offsets in bytes to load from ref and store to dst buffer
63 // TEMP0, TEMP4, TEMP8 and TEMP12 - registers for corresponding tmp elements 63 // TEMP0, TEMP4, TEMP8 and TEMP12 - registers for corresponding tmp elements
64 #define HORIZONTAL_PASS(A, B, C, D, TEMP0, TEMP4, TEMP8, TEMP12) \ 64 #define HORIZONTAL_PASS(A, B, C, D, TEMP0, TEMP4, TEMP8, TEMP12) \
65 "addiu %["#TEMP0"], %["#TEMP0"], 4 \n\t" \ 65 "addiu %[" #TEMP0 "], %[" #TEMP0 "], 4 \n\t" \
66 "addu %[temp16], %["#TEMP0"], %["#TEMP8"] \n\t" \ 66 "addu %[temp16], %[" #TEMP0 "], %[" #TEMP8 "] \n\t" \
67 "subu %[temp17], %["#TEMP0"], %["#TEMP8"] \n\t" \ 67 "subu %[temp17], %[" #TEMP0 "], %[" #TEMP8 "] \n\t" \
68 "mul %["#TEMP0"], %["#TEMP4"], %[kC2] \n\t" \ 68 "mul %[" #TEMP0 "], %[" #TEMP4 "], %[kC2] \n\t" \
69 "mul %["#TEMP8"], %["#TEMP12"], %[kC1] \n\t" \ 69 "mul %[" #TEMP8 "], %[" #TEMP12 "], %[kC1] \n\t" \
70 "mul %["#TEMP4"], %["#TEMP4"], %[kC1] \n\t" \ 70 "mul %[" #TEMP4 "], %[" #TEMP4 "], %[kC1] \n\t" \
71 "mul %["#TEMP12"], %["#TEMP12"], %[kC2] \n\t" \ 71 "mul %[" #TEMP12 "], %[" #TEMP12 "], %[kC2] \n\t" \
72 "sra %["#TEMP0"], %["#TEMP0"], 16 \n\t" \ 72 "sra %[" #TEMP0 "], %[" #TEMP0 "], 16 \n\t" \
73 "sra %["#TEMP8"], %["#TEMP8"], 16 \n\t" \ 73 "sra %[" #TEMP8 "], %[" #TEMP8 "], 16 \n\t" \
74 "sra %["#TEMP4"], %["#TEMP4"], 16 \n\t" \ 74 "sra %[" #TEMP4 "], %[" #TEMP4 "], 16 \n\t" \
75 "sra %["#TEMP12"], %["#TEMP12"], 16 \n\t" \ 75 "sra %[" #TEMP12 "], %[" #TEMP12 "], 16 \n\t" \
76 "subu %[temp18], %["#TEMP0"], %["#TEMP8"] \n\t" \ 76 "subu %[temp18], %[" #TEMP0 "], %[" #TEMP8 "] \n\t" \
77 "addu %[temp19], %["#TEMP4"], %["#TEMP12"] \n\t" \ 77 "addu %[temp19], %[" #TEMP4 "], %[" #TEMP12 "] \n\t" \
78 "addu %["#TEMP0"], %[temp16], %[temp19] \n\t" \ 78 "addu %[" #TEMP0 "], %[temp16], %[temp19] \n\t" \
79 "addu %["#TEMP4"], %[temp17], %[temp18] \n\t" \ 79 "addu %[" #TEMP4 "], %[temp17], %[temp18] \n\t" \
80 "subu %["#TEMP8"], %[temp17], %[temp18] \n\t" \ 80 "subu %[" #TEMP8 "], %[temp17], %[temp18] \n\t" \
81 "subu %["#TEMP12"], %[temp16], %[temp19] \n\t" \ 81 "subu %[" #TEMP12 "], %[temp16], %[temp19] \n\t" \
82 "lw %[temp20], 0(%[args]) \n\t" \ 82 "lw %[temp20], 0(%[args]) \n\t" \
83 "sra %["#TEMP0"], %["#TEMP0"], 3 \n\t" \ 83 "sra %[" #TEMP0 "], %[" #TEMP0 "], 3 \n\t" \
84 "sra %["#TEMP4"], %["#TEMP4"], 3 \n\t" \ 84 "sra %[" #TEMP4 "], %[" #TEMP4 "], 3 \n\t" \
85 "sra %["#TEMP8"], %["#TEMP8"], 3 \n\t" \ 85 "sra %[" #TEMP8 "], %[" #TEMP8 "], 3 \n\t" \
86 "sra %["#TEMP12"], %["#TEMP12"], 3 \n\t" \ 86 "sra %[" #TEMP12 "], %[" #TEMP12 "], 3 \n\t" \
87 "lbu %[temp16], "#A"(%[temp20]) \n\t" \ 87 "lbu %[temp16], " #A "(%[temp20]) \n\t" \
88 "lbu %[temp17], "#B"(%[temp20]) \n\t" \ 88 "lbu %[temp17], " #B "(%[temp20]) \n\t" \
89 "lbu %[temp18], "#C"(%[temp20]) \n\t" \ 89 "lbu %[temp18], " #C "(%[temp20]) \n\t" \
90 "lbu %[temp19], "#D"(%[temp20]) \n\t" \ 90 "lbu %[temp19], " #D "(%[temp20]) \n\t" \
91 "addu %["#TEMP0"], %[temp16], %["#TEMP0"] \n\t" \ 91 "addu %[" #TEMP0 "], %[temp16], %[" #TEMP0 "] \n\t" \
92 "addu %["#TEMP4"], %[temp17], %["#TEMP4"] \n\t" \ 92 "addu %[" #TEMP4 "], %[temp17], %[" #TEMP4 "] \n\t" \
93 "addu %["#TEMP8"], %[temp18], %["#TEMP8"] \n\t" \ 93 "addu %[" #TEMP8 "], %[temp18], %[" #TEMP8 "] \n\t" \
94 "addu %["#TEMP12"], %[temp19], %["#TEMP12"] \n\t" \ 94 "addu %[" #TEMP12 "], %[temp19], %[" #TEMP12 "] \n\t" \
95 "slt %[temp16], %["#TEMP0"], $zero \n\t" \ 95 "slt %[temp16], %[" #TEMP0 "], $zero \n\t" \
96 "slt %[temp17], %["#TEMP4"], $zero \n\t" \ 96 "slt %[temp17], %[" #TEMP4 "], $zero \n\t" \
97 "slt %[temp18], %["#TEMP8"], $zero \n\t" \ 97 "slt %[temp18], %[" #TEMP8 "], $zero \n\t" \
98 "slt %[temp19], %["#TEMP12"], $zero \n\t" \ 98 "slt %[temp19], %[" #TEMP12 "], $zero \n\t" \
99 "movn %["#TEMP0"], $zero, %[temp16] \n\t" \ 99 "movn %[" #TEMP0 "], $zero, %[temp16] \n\t" \
100 "movn %["#TEMP4"], $zero, %[temp17] \n\t" \ 100 "movn %[" #TEMP4 "], $zero, %[temp17] \n\t" \
101 "movn %["#TEMP8"], $zero, %[temp18] \n\t" \ 101 "movn %[" #TEMP8 "], $zero, %[temp18] \n\t" \
102 "movn %["#TEMP12"], $zero, %[temp19] \n\t" \ 102 "movn %[" #TEMP12 "], $zero, %[temp19] \n\t" \
103 "addiu %[temp20], $zero, 255 \n\t" \ 103 "addiu %[temp20], $zero, 255 \n\t" \
104 "slt %[temp16], %["#TEMP0"], %[temp20] \n\t" \ 104 "slt %[temp16], %[" #TEMP0 "], %[temp20] \n\t" \
105 "slt %[temp17], %["#TEMP4"], %[temp20] \n\t" \ 105 "slt %[temp17], %[" #TEMP4 "], %[temp20] \n\t" \
106 "slt %[temp18], %["#TEMP8"], %[temp20] \n\t" \ 106 "slt %[temp18], %[" #TEMP8 "], %[temp20] \n\t" \
107 "slt %[temp19], %["#TEMP12"], %[temp20] \n\t" \ 107 "slt %[temp19], %[" #TEMP12 "], %[temp20] \n\t" \
108 "movz %["#TEMP0"], %[temp20], %[temp16] \n\t" \ 108 "movz %[" #TEMP0 "], %[temp20], %[temp16] \n\t" \
109 "movz %["#TEMP4"], %[temp20], %[temp17] \n\t" \ 109 "movz %[" #TEMP4 "], %[temp20], %[temp17] \n\t" \
110 "lw %[temp16], 8(%[args]) \n\t" \ 110 "lw %[temp16], 8(%[args]) \n\t" \
111 "movz %["#TEMP8"], %[temp20], %[temp18] \n\t" \ 111 "movz %[" #TEMP8 "], %[temp20], %[temp18] \n\t" \
112 "movz %["#TEMP12"], %[temp20], %[temp19] \n\t" \ 112 "movz %[" #TEMP12 "], %[temp20], %[temp19] \n\t" \
113 "sb %["#TEMP0"], "#A"(%[temp16]) \n\t" \ 113 "sb %[" #TEMP0 "], " #A "(%[temp16]) \n\t" \
114 "sb %["#TEMP4"], "#B"(%[temp16]) \n\t" \ 114 "sb %[" #TEMP4 "], " #B "(%[temp16]) \n\t" \
115 "sb %["#TEMP8"], "#C"(%[temp16]) \n\t" \ 115 "sb %[" #TEMP8 "], " #C "(%[temp16]) \n\t" \
116 "sb %["#TEMP12"], "#D"(%[temp16]) \n\t" 116 "sb %[" #TEMP12 "], " #D "(%[temp16]) \n\t"
117 117
118 // Does one or two inverse transforms. 118 // Does one or two inverse transforms.
119 static WEBP_INLINE void ITransformOne(const uint8_t* ref, const int16_t* in, 119 static WEBP_INLINE void ITransformOne(const uint8_t* ref, const int16_t* in,
120 uint8_t* dst) { 120 uint8_t* dst) {
121 int temp0, temp1, temp2, temp3, temp4, temp5, temp6; 121 int temp0, temp1, temp2, temp3, temp4, temp5, temp6;
122 int temp7, temp8, temp9, temp10, temp11, temp12, temp13; 122 int temp7, temp8, temp9, temp10, temp11, temp12, temp13;
123 int temp14, temp15, temp16, temp17, temp18, temp19, temp20; 123 int temp14, temp15, temp16, temp17, temp18, temp19, temp20;
124 const int* args[3] = {(const int*)ref, (const int*)in, (const int*)dst}; 124 const int* args[3] = {(const int*)ref, (const int*)in, (const int*)dst};
125 125
126 __asm__ volatile( 126 __asm__ volatile(
(...skipping 30 matching lines...) Expand all
157 157
158 #undef VERTICAL_PASS 158 #undef VERTICAL_PASS
159 #undef HORIZONTAL_PASS 159 #undef HORIZONTAL_PASS
160 160
161 // macro for one pass through for loop in QuantizeBlock 161 // macro for one pass through for loop in QuantizeBlock
162 // QUANTDIV macro inlined 162 // QUANTDIV macro inlined
163 // J - offset in bytes (kZigzag[n] * 2) 163 // J - offset in bytes (kZigzag[n] * 2)
164 // K - offset in bytes (kZigzag[n] * 4) 164 // K - offset in bytes (kZigzag[n] * 4)
165 // N - offset in bytes (n * 2) 165 // N - offset in bytes (n * 2)
166 #define QUANTIZE_ONE(J, K, N) \ 166 #define QUANTIZE_ONE(J, K, N) \
167 "lh %[temp0], "#J"(%[ppin]) \n\t" \ 167 "lh %[temp0], " #J "(%[ppin]) \n\t" \
168 "lhu %[temp1], "#J"(%[ppsharpen]) \n\t" \ 168 "lhu %[temp1], " #J "(%[ppsharpen]) \n\t" \
169 "lw %[temp2], "#K"(%[ppzthresh]) \n\t" \ 169 "lw %[temp2], " #K "(%[ppzthresh]) \n\t" \
170 "sra %[sign], %[temp0], 15 \n\t" \ 170 "sra %[sign], %[temp0], 15 \n\t" \
171 "xor %[coeff], %[temp0], %[sign] \n\t" \ 171 "xor %[coeff], %[temp0], %[sign] \n\t" \
172 "subu %[coeff], %[coeff], %[sign] \n\t" \ 172 "subu %[coeff], %[coeff], %[sign] \n\t" \
173 "addu %[coeff], %[coeff], %[temp1] \n\t" \ 173 "addu %[coeff], %[coeff], %[temp1] \n\t" \
174 "slt %[temp4], %[temp2], %[coeff] \n\t" \ 174 "slt %[temp4], %[temp2], %[coeff] \n\t" \
175 "addiu %[temp5], $zero, 0 \n\t" \ 175 "addiu %[temp5], $zero, 0 \n\t" \
176 "addiu %[level], $zero, 0 \n\t" \ 176 "addiu %[level], $zero, 0 \n\t" \
177 "beqz %[temp4], 2f \n\t" \ 177 "beqz %[temp4], 2f \n\t" \
178 "lhu %[temp1], "#J"(%[ppiq]) \n\t" \ 178 "lhu %[temp1], " #J "(%[ppiq]) \n\t" \
179 "lw %[temp2], "#K"(%[ppbias]) \n\t" \ 179 "lw %[temp2], " #K "(%[ppbias]) \n\t" \
180 "lhu %[temp3], "#J"(%[ppq]) \n\t" \ 180 "lhu %[temp3], " #J "(%[ppq]) \n\t" \
181 "mul %[level], %[coeff], %[temp1] \n\t" \ 181 "mul %[level], %[coeff], %[temp1] \n\t" \
182 "addu %[level], %[level], %[temp2] \n\t" \ 182 "addu %[level], %[level], %[temp2] \n\t" \
183 "sra %[level], %[level], 17 \n\t" \ 183 "sra %[level], %[level], 17 \n\t" \
184 "slt %[temp4], %[max_level], %[level] \n\t" \ 184 "slt %[temp4], %[max_level], %[level] \n\t" \
185 "movn %[level], %[max_level], %[temp4] \n\t" \ 185 "movn %[level], %[max_level], %[temp4] \n\t" \
186 "xor %[level], %[level], %[sign] \n\t" \ 186 "xor %[level], %[level], %[sign] \n\t" \
187 "subu %[level], %[level], %[sign] \n\t" \ 187 "subu %[level], %[level], %[sign] \n\t" \
188 "mul %[temp5], %[level], %[temp3] \n\t" \ 188 "mul %[temp5], %[level], %[temp3] \n\t" \
189 "2: \n\t" \ 189 "2: \n\t" \
190 "sh %[temp5], "#J"(%[ppin]) \n\t" \ 190 "sh %[temp5], " #J "(%[ppin]) \n\t" \
191 "sh %[level], "#N"(%[pout]) \n\t" 191 "sh %[level], " #N "(%[pout]) \n\t"
192 192
193 static int QuantizeBlock(int16_t in[16], int16_t out[16], 193 static int QuantizeBlock(int16_t in[16], int16_t out[16],
194 const VP8Matrix* const mtx) { 194 const VP8Matrix* const mtx) {
195 int temp0, temp1, temp2, temp3, temp4, temp5; 195 int temp0, temp1, temp2, temp3, temp4, temp5;
196 int sign, coeff, level, i; 196 int sign, coeff, level, i;
197 int max_level = MAX_LEVEL; 197 int max_level = MAX_LEVEL;
198 198
199 int16_t* ppin = &in[0]; 199 int16_t* ppin = &in[0];
200 int16_t* pout = &out[0]; 200 int16_t* pout = &out[0];
201 const uint16_t* ppsharpen = &mtx->sharpen_[0]; 201 const uint16_t* ppsharpen = &mtx->sharpen_[0];
(...skipping 40 matching lines...) Expand 10 before | Expand all | Expand 10 after
242 } 242 }
243 243
244 #undef QUANTIZE_ONE 244 #undef QUANTIZE_ONE
245 245
246 // macro for one horizontal pass in Disto4x4 (TTransform) 246 // macro for one horizontal pass in Disto4x4 (TTransform)
247 // two calls of function TTransform are merged into single one 247 // two calls of function TTransform are merged into single one
248 // A..D - offsets in bytes to load from a and b buffers 248 // A..D - offsets in bytes to load from a and b buffers
249 // E..H - offsets in bytes to store first results to tmp buffer 249 // E..H - offsets in bytes to store first results to tmp buffer
250 // E1..H1 - offsets in bytes to store second results to tmp buffer 250 // E1..H1 - offsets in bytes to store second results to tmp buffer
251 #define HORIZONTAL_PASS(A, B, C, D, E, F, G, H, E1, F1, G1, H1) \ 251 #define HORIZONTAL_PASS(A, B, C, D, E, F, G, H, E1, F1, G1, H1) \
252 "lbu %[temp0], "#A"(%[a]) \n\t" \ 252 "lbu %[temp0], " #A "(%[a]) \n\t" \
253 "lbu %[temp1], "#B"(%[a]) \n\t" \ 253 "lbu %[temp1], " #B "(%[a]) \n\t" \
254 "lbu %[temp2], "#C"(%[a]) \n\t" \ 254 "lbu %[temp2], " #C "(%[a]) \n\t" \
255 "lbu %[temp3], "#D"(%[a]) \n\t" \ 255 "lbu %[temp3], " #D "(%[a]) \n\t" \
256 "lbu %[temp4], "#A"(%[b]) \n\t" \ 256 "lbu %[temp4], " #A "(%[b]) \n\t" \
257 "lbu %[temp5], "#B"(%[b]) \n\t" \ 257 "lbu %[temp5], " #B "(%[b]) \n\t" \
258 "lbu %[temp6], "#C"(%[b]) \n\t" \ 258 "lbu %[temp6], " #C "(%[b]) \n\t" \
259 "lbu %[temp7], "#D"(%[b]) \n\t" \ 259 "lbu %[temp7], " #D "(%[b]) \n\t" \
260 "addu %[temp8], %[temp0], %[temp2] \n\t" \ 260 "addu %[temp8], %[temp0], %[temp2] \n\t" \
261 "subu %[temp0], %[temp0], %[temp2] \n\t" \ 261 "subu %[temp0], %[temp0], %[temp2] \n\t" \
262 "addu %[temp2], %[temp1], %[temp3] \n\t" \ 262 "addu %[temp2], %[temp1], %[temp3] \n\t" \
263 "subu %[temp1], %[temp1], %[temp3] \n\t" \ 263 "subu %[temp1], %[temp1], %[temp3] \n\t" \
264 "addu %[temp3], %[temp4], %[temp6] \n\t" \ 264 "addu %[temp3], %[temp4], %[temp6] \n\t" \
265 "subu %[temp4], %[temp4], %[temp6] \n\t" \ 265 "subu %[temp4], %[temp4], %[temp6] \n\t" \
266 "addu %[temp6], %[temp5], %[temp7] \n\t" \ 266 "addu %[temp6], %[temp5], %[temp7] \n\t" \
267 "subu %[temp5], %[temp5], %[temp7] \n\t" \ 267 "subu %[temp5], %[temp5], %[temp7] \n\t" \
268 "addu %[temp7], %[temp8], %[temp2] \n\t" \ 268 "addu %[temp7], %[temp8], %[temp2] \n\t" \
269 "subu %[temp2], %[temp8], %[temp2] \n\t" \ 269 "subu %[temp2], %[temp8], %[temp2] \n\t" \
270 "addu %[temp8], %[temp0], %[temp1] \n\t" \ 270 "addu %[temp8], %[temp0], %[temp1] \n\t" \
271 "subu %[temp0], %[temp0], %[temp1] \n\t" \ 271 "subu %[temp0], %[temp0], %[temp1] \n\t" \
272 "addu %[temp1], %[temp3], %[temp6] \n\t" \ 272 "addu %[temp1], %[temp3], %[temp6] \n\t" \
273 "subu %[temp3], %[temp3], %[temp6] \n\t" \ 273 "subu %[temp3], %[temp3], %[temp6] \n\t" \
274 "addu %[temp6], %[temp4], %[temp5] \n\t" \ 274 "addu %[temp6], %[temp4], %[temp5] \n\t" \
275 "subu %[temp4], %[temp4], %[temp5] \n\t" \ 275 "subu %[temp4], %[temp4], %[temp5] \n\t" \
276 "sw %[temp7], "#E"(%[tmp]) \n\t" \ 276 "sw %[temp7], " #E "(%[tmp]) \n\t" \
277 "sw %[temp2], "#H"(%[tmp]) \n\t" \ 277 "sw %[temp2], " #H "(%[tmp]) \n\t" \
278 "sw %[temp8], "#F"(%[tmp]) \n\t" \ 278 "sw %[temp8], " #F "(%[tmp]) \n\t" \
279 "sw %[temp0], "#G"(%[tmp]) \n\t" \ 279 "sw %[temp0], " #G "(%[tmp]) \n\t" \
280 "sw %[temp1], "#E1"(%[tmp]) \n\t" \ 280 "sw %[temp1], " #E1 "(%[tmp]) \n\t" \
281 "sw %[temp3], "#H1"(%[tmp]) \n\t" \ 281 "sw %[temp3], " #H1 "(%[tmp]) \n\t" \
282 "sw %[temp6], "#F1"(%[tmp]) \n\t" \ 282 "sw %[temp6], " #F1 "(%[tmp]) \n\t" \
283 "sw %[temp4], "#G1"(%[tmp]) \n\t" 283 "sw %[temp4], " #G1 "(%[tmp]) \n\t"
284 284
285 // macro for one vertical pass in Disto4x4 (TTransform) 285 // macro for one vertical pass in Disto4x4 (TTransform)
286 // two calls of function TTransform are merged into single one 286 // two calls of function TTransform are merged into single one
287 // since only one accu is available in mips32r1 instruction set 287 // since only one accu is available in mips32r1 instruction set
288 // first is done second call of function TTransform and after 288 // first is done second call of function TTransform and after
289 // that first one. 289 // that first one.
290 // const int sum1 = TTransform(a, w); 290 // const int sum1 = TTransform(a, w);
291 // const int sum2 = TTransform(b, w); 291 // const int sum2 = TTransform(b, w);
292 // return abs(sum2 - sum1) >> 5; 292 // return abs(sum2 - sum1) >> 5;
293 // (sum2 - sum1) is calculated with madds (sub2) and msubs (sub1) 293 // (sum2 - sum1) is calculated with madds (sub2) and msubs (sub1)
294 // A..D - offsets in bytes to load first results from tmp buffer 294 // A..D - offsets in bytes to load first results from tmp buffer
295 // A1..D1 - offsets in bytes to load second results from tmp buffer 295 // A1..D1 - offsets in bytes to load second results from tmp buffer
296 // E..H - offsets in bytes to load from w buffer 296 // E..H - offsets in bytes to load from w buffer
297 #define VERTICAL_PASS(A, B, C, D, A1, B1, C1, D1, E, F, G, H) \ 297 #define VERTICAL_PASS(A, B, C, D, A1, B1, C1, D1, E, F, G, H) \
298 "lw %[temp0], "#A1"(%[tmp]) \n\t" \ 298 "lw %[temp0], " #A1 "(%[tmp]) \n\t" \
299 "lw %[temp1], "#C1"(%[tmp]) \n\t" \ 299 "lw %[temp1], " #C1 "(%[tmp]) \n\t" \
300 "lw %[temp2], "#B1"(%[tmp]) \n\t" \ 300 "lw %[temp2], " #B1 "(%[tmp]) \n\t" \
301 "lw %[temp3], "#D1"(%[tmp]) \n\t" \ 301 "lw %[temp3], " #D1 "(%[tmp]) \n\t" \
302 "addu %[temp8], %[temp0], %[temp1] \n\t" \ 302 "addu %[temp8], %[temp0], %[temp1] \n\t" \
303 "subu %[temp0], %[temp0], %[temp1] \n\t" \ 303 "subu %[temp0], %[temp0], %[temp1] \n\t" \
304 "addu %[temp1], %[temp2], %[temp3] \n\t" \ 304 "addu %[temp1], %[temp2], %[temp3] \n\t" \
305 "subu %[temp2], %[temp2], %[temp3] \n\t" \ 305 "subu %[temp2], %[temp2], %[temp3] \n\t" \
306 "addu %[temp3], %[temp8], %[temp1] \n\t" \ 306 "addu %[temp3], %[temp8], %[temp1] \n\t" \
307 "subu %[temp8], %[temp8], %[temp1] \n\t" \ 307 "subu %[temp8], %[temp8], %[temp1] \n\t" \
308 "addu %[temp1], %[temp0], %[temp2] \n\t" \ 308 "addu %[temp1], %[temp0], %[temp2] \n\t" \
309 "subu %[temp0], %[temp0], %[temp2] \n\t" \ 309 "subu %[temp0], %[temp0], %[temp2] \n\t" \
310 "sra %[temp4], %[temp3], 31 \n\t" \ 310 "sra %[temp4], %[temp3], 31 \n\t" \
311 "sra %[temp5], %[temp1], 31 \n\t" \ 311 "sra %[temp5], %[temp1], 31 \n\t" \
312 "sra %[temp6], %[temp0], 31 \n\t" \ 312 "sra %[temp6], %[temp0], 31 \n\t" \
313 "sra %[temp7], %[temp8], 31 \n\t" \ 313 "sra %[temp7], %[temp8], 31 \n\t" \
314 "xor %[temp3], %[temp3], %[temp4] \n\t" \ 314 "xor %[temp3], %[temp3], %[temp4] \n\t" \
315 "xor %[temp1], %[temp1], %[temp5] \n\t" \ 315 "xor %[temp1], %[temp1], %[temp5] \n\t" \
316 "xor %[temp0], %[temp0], %[temp6] \n\t" \ 316 "xor %[temp0], %[temp0], %[temp6] \n\t" \
317 "xor %[temp8], %[temp8], %[temp7] \n\t" \ 317 "xor %[temp8], %[temp8], %[temp7] \n\t" \
318 "subu %[temp3], %[temp3], %[temp4] \n\t" \ 318 "subu %[temp3], %[temp3], %[temp4] \n\t" \
319 "subu %[temp1], %[temp1], %[temp5] \n\t" \ 319 "subu %[temp1], %[temp1], %[temp5] \n\t" \
320 "subu %[temp0], %[temp0], %[temp6] \n\t" \ 320 "subu %[temp0], %[temp0], %[temp6] \n\t" \
321 "subu %[temp8], %[temp8], %[temp7] \n\t" \ 321 "subu %[temp8], %[temp8], %[temp7] \n\t" \
322 "lhu %[temp4], "#E"(%[w]) \n\t" \ 322 "lhu %[temp4], " #E "(%[w]) \n\t" \
323 "lhu %[temp5], "#F"(%[w]) \n\t" \ 323 "lhu %[temp5], " #F "(%[w]) \n\t" \
324 "lhu %[temp6], "#G"(%[w]) \n\t" \ 324 "lhu %[temp6], " #G "(%[w]) \n\t" \
325 "lhu %[temp7], "#H"(%[w]) \n\t" \ 325 "lhu %[temp7], " #H "(%[w]) \n\t" \
326 "madd %[temp4], %[temp3] \n\t" \ 326 "madd %[temp4], %[temp3] \n\t" \
327 "madd %[temp5], %[temp1] \n\t" \ 327 "madd %[temp5], %[temp1] \n\t" \
328 "madd %[temp6], %[temp0] \n\t" \ 328 "madd %[temp6], %[temp0] \n\t" \
329 "madd %[temp7], %[temp8] \n\t" \ 329 "madd %[temp7], %[temp8] \n\t" \
330 "lw %[temp0], "#A"(%[tmp]) \n\t" \ 330 "lw %[temp0], " #A "(%[tmp]) \n\t" \
331 "lw %[temp1], "#C"(%[tmp]) \n\t" \ 331 "lw %[temp1], " #C "(%[tmp]) \n\t" \
332 "lw %[temp2], "#B"(%[tmp]) \n\t" \ 332 "lw %[temp2], " #B "(%[tmp]) \n\t" \
333 "lw %[temp3], "#D"(%[tmp]) \n\t" \ 333 "lw %[temp3], " #D "(%[tmp]) \n\t" \
334 "addu %[temp8], %[temp0], %[temp1] \n\t" \ 334 "addu %[temp8], %[temp0], %[temp1] \n\t" \
335 "subu %[temp0], %[temp0], %[temp1] \n\t" \ 335 "subu %[temp0], %[temp0], %[temp1] \n\t" \
336 "addu %[temp1], %[temp2], %[temp3] \n\t" \ 336 "addu %[temp1], %[temp2], %[temp3] \n\t" \
337 "subu %[temp2], %[temp2], %[temp3] \n\t" \ 337 "subu %[temp2], %[temp2], %[temp3] \n\t" \
338 "addu %[temp3], %[temp8], %[temp1] \n\t" \ 338 "addu %[temp3], %[temp8], %[temp1] \n\t" \
339 "subu %[temp1], %[temp8], %[temp1] \n\t" \ 339 "subu %[temp1], %[temp8], %[temp1] \n\t" \
340 "addu %[temp8], %[temp0], %[temp2] \n\t" \ 340 "addu %[temp8], %[temp0], %[temp2] \n\t" \
341 "subu %[temp0], %[temp0], %[temp2] \n\t" \ 341 "subu %[temp0], %[temp0], %[temp2] \n\t" \
342 "sra %[temp2], %[temp3], 31 \n\t" \ 342 "sra %[temp2], %[temp3], 31 \n\t" \
343 "xor %[temp3], %[temp3], %[temp2] \n\t" \ 343 "xor %[temp3], %[temp3], %[temp2] \n\t" \
(...skipping 56 matching lines...) Expand 10 before | Expand all | Expand 10 after
400 D += Disto4x4(a + x + y, b + x + y, w); 400 D += Disto4x4(a + x + y, b + x + y, w);
401 } 401 }
402 } 402 }
403 return D; 403 return D;
404 } 404 }
405 405
406 // macro for one horizontal pass in FTransform 406 // macro for one horizontal pass in FTransform
407 // temp0..temp15 holds tmp[0]..tmp[15] 407 // temp0..temp15 holds tmp[0]..tmp[15]
408 // A..D - offsets in bytes to load from src and ref buffers 408 // A..D - offsets in bytes to load from src and ref buffers
409 // TEMP0..TEMP3 - registers for corresponding tmp elements 409 // TEMP0..TEMP3 - registers for corresponding tmp elements
410 #define HORIZONTAL_PASS(A, B, C, D, TEMP0, TEMP1, TEMP2, TEMP3) \ 410 #define HORIZONTAL_PASS(A, B, C, D, TEMP0, TEMP1, TEMP2, TEMP3) \
411 "lw %["#TEMP1"], 0(%[args]) \n\t" \ 411 "lw %[" #TEMP1 "], 0(%[args]) \n\t" \
412 "lw %["#TEMP2"], 4(%[args]) \n\t" \ 412 "lw %[" #TEMP2 "], 4(%[args]) \n\t" \
413 "lbu %[temp16], "#A"(%["#TEMP1"]) \n\t" \ 413 "lbu %[temp16], " #A "(%[" #TEMP1 "]) \n\t" \
414 "lbu %[temp17], "#A"(%["#TEMP2"]) \n\t" \ 414 "lbu %[temp17], " #A "(%[" #TEMP2 "]) \n\t" \
415 "lbu %[temp18], "#B"(%["#TEMP1"]) \n\t" \ 415 "lbu %[temp18], " #B "(%[" #TEMP1 "]) \n\t" \
416 "lbu %[temp19], "#B"(%["#TEMP2"]) \n\t" \ 416 "lbu %[temp19], " #B "(%[" #TEMP2 "]) \n\t" \
417 "subu %[temp20], %[temp16], %[temp17] \n\t" \ 417 "subu %[temp20], %[temp16], %[temp17] \n\t" \
418 "lbu %[temp16], "#C"(%["#TEMP1"]) \n\t" \ 418 "lbu %[temp16], " #C "(%[" #TEMP1 "]) \n\t" \
419 "lbu %[temp17], "#C"(%["#TEMP2"]) \n\t" \ 419 "lbu %[temp17], " #C "(%[" #TEMP2 "]) \n\t" \
420 "subu %["#TEMP0"], %[temp18], %[temp19] \n\t" \ 420 "subu %[" #TEMP0 "], %[temp18], %[temp19] \n\t" \
421 "lbu %[temp18], "#D"(%["#TEMP1"]) \n\t" \ 421 "lbu %[temp18], " #D "(%[" #TEMP1 "]) \n\t" \
422 "lbu %[temp19], "#D"(%["#TEMP2"]) \n\t" \ 422 "lbu %[temp19], " #D "(%[" #TEMP2 "]) \n\t" \
423 "subu %["#TEMP1"], %[temp16], %[temp17] \n\t" \ 423 "subu %[" #TEMP1 "], %[temp16], %[temp17] \n\t" \
424 "subu %["#TEMP2"], %[temp18], %[temp19] \n\t" \ 424 "subu %[" #TEMP2 "], %[temp18], %[temp19] \n\t" \
425 "addu %["#TEMP3"], %[temp20], %["#TEMP2"] \n\t" \ 425 "addu %[" #TEMP3 "], %[temp20], %[" #TEMP2 "] \n\t" \
426 "subu %["#TEMP2"], %[temp20], %["#TEMP2"] \n\t" \ 426 "subu %[" #TEMP2 "], %[temp20], %[" #TEMP2 "] \n\t" \
427 "addu %[temp20], %["#TEMP0"], %["#TEMP1"] \n\t" \ 427 "addu %[temp20], %[" #TEMP0 "], %[" #TEMP1 "] \n\t" \
428 "subu %["#TEMP0"], %["#TEMP0"], %["#TEMP1"] \n\t" \ 428 "subu %[" #TEMP0 "], %[" #TEMP0 "], %[" #TEMP1 "] \n\t" \
429 "mul %[temp16], %["#TEMP2"], %[c5352] \n\t" \ 429 "mul %[temp16], %[" #TEMP2 "], %[c5352] \n\t" \
430 "mul %[temp17], %["#TEMP2"], %[c2217] \n\t" \ 430 "mul %[temp17], %[" #TEMP2 "], %[c2217] \n\t" \
431 "mul %[temp18], %["#TEMP0"], %[c5352] \n\t" \ 431 "mul %[temp18], %[" #TEMP0 "], %[c5352] \n\t" \
432 "mul %[temp19], %["#TEMP0"], %[c2217] \n\t" \ 432 "mul %[temp19], %[" #TEMP0 "], %[c2217] \n\t" \
433 "addu %["#TEMP1"], %["#TEMP3"], %[temp20] \n\t" \ 433 "addu %[" #TEMP1 "], %[" #TEMP3 "], %[temp20] \n\t" \
434 "subu %[temp20], %["#TEMP3"], %[temp20] \n\t" \ 434 "subu %[temp20], %[" #TEMP3 "], %[temp20] \n\t" \
435 "sll %["#TEMP0"], %["#TEMP1"], 3 \n\t" \ 435 "sll %[" #TEMP0 "], %[" #TEMP1 "], 3 \n\t" \
436 "sll %["#TEMP2"], %[temp20], 3 \n\t" \ 436 "sll %[" #TEMP2 "], %[temp20], 3 \n\t" \
437 "addiu %[temp16], %[temp16], 1812 \n\t" \ 437 "addiu %[temp16], %[temp16], 1812 \n\t" \
438 "addiu %[temp17], %[temp17], 937 \n\t" \ 438 "addiu %[temp17], %[temp17], 937 \n\t" \
439 "addu %[temp16], %[temp16], %[temp19] \n\t" \ 439 "addu %[temp16], %[temp16], %[temp19] \n\t" \
440 "subu %[temp17], %[temp17], %[temp18] \n\t" \ 440 "subu %[temp17], %[temp17], %[temp18] \n\t" \
441 "sra %["#TEMP1"], %[temp16], 9 \n\t" \ 441 "sra %[" #TEMP1 "], %[temp16], 9 \n\t" \
442 "sra %["#TEMP3"], %[temp17], 9 \n\t" 442 "sra %[" #TEMP3 "], %[temp17], 9 \n\t"
443 443
444 // macro for one vertical pass in FTransform 444 // macro for one vertical pass in FTransform
445 // temp0..temp15 holds tmp[0]..tmp[15] 445 // temp0..temp15 holds tmp[0]..tmp[15]
446 // A..D - offsets in bytes to store to out buffer 446 // A..D - offsets in bytes to store to out buffer
447 // TEMP0, TEMP4, TEMP8 and TEMP12 - registers for corresponding tmp elements 447 // TEMP0, TEMP4, TEMP8 and TEMP12 - registers for corresponding tmp elements
448 #define VERTICAL_PASS(A, B, C, D, TEMP0, TEMP4, TEMP8, TEMP12) \ 448 #define VERTICAL_PASS(A, B, C, D, TEMP0, TEMP4, TEMP8, TEMP12) \
449 "addu %[temp16], %["#TEMP0"], %["#TEMP12"] \n\t" \ 449 "addu %[temp16], %[" #TEMP0 "], %[" #TEMP12 "] \n\t" \
450 "subu %[temp19], %["#TEMP0"], %["#TEMP12"] \n\t" \ 450 "subu %[temp19], %[" #TEMP0 "], %[" #TEMP12 "] \n\t" \
451 "addu %[temp17], %["#TEMP4"], %["#TEMP8"] \n\t" \ 451 "addu %[temp17], %[" #TEMP4 "], %[" #TEMP8 "] \n\t" \
452 "subu %[temp18], %["#TEMP4"], %["#TEMP8"] \n\t" \ 452 "subu %[temp18], %[" #TEMP4 "], %[" #TEMP8 "] \n\t" \
453 "mul %["#TEMP8"], %[temp19], %[c2217] \n\t" \ 453 "mul %[" #TEMP8 "], %[temp19], %[c2217] \n\t" \
454 "mul %["#TEMP12"], %[temp18], %[c2217] \n\t" \ 454 "mul %[" #TEMP12 "], %[temp18], %[c2217] \n\t" \
455 "mul %["#TEMP4"], %[temp19], %[c5352] \n\t" \ 455 "mul %[" #TEMP4 "], %[temp19], %[c5352] \n\t" \
456 "mul %[temp18], %[temp18], %[c5352] \n\t" \ 456 "mul %[temp18], %[temp18], %[c5352] \n\t" \
457 "addiu %[temp16], %[temp16], 7 \n\t" \ 457 "addiu %[temp16], %[temp16], 7 \n\t" \
458 "addu %["#TEMP0"], %[temp16], %[temp17] \n\t" \ 458 "addu %[" #TEMP0 "], %[temp16], %[temp17] \n\t" \
459 "sra %["#TEMP0"], %["#TEMP0"], 4 \n\t" \ 459 "sra %[" #TEMP0 "], %[" #TEMP0 "], 4 \n\t" \
460 "addu %["#TEMP12"], %["#TEMP12"], %["#TEMP4"] \n\t" \ 460 "addu %[" #TEMP12 "], %[" #TEMP12 "], %[" #TEMP4 "] \n\t" \
461 "subu %["#TEMP4"], %[temp16], %[temp17] \n\t" \ 461 "subu %[" #TEMP4 "], %[temp16], %[temp17] \n\t" \
462 "sra %["#TEMP4"], %["#TEMP4"], 4 \n\t" \ 462 "sra %[" #TEMP4 "], %[" #TEMP4 "], 4 \n\t" \
463 "addiu %["#TEMP8"], %["#TEMP8"], 30000 \n\t" \ 463 "addiu %[" #TEMP8 "], %[" #TEMP8 "], 30000 \n\t" \
464 "addiu %["#TEMP12"], %["#TEMP12"], 12000 \n\t" \ 464 "addiu %[" #TEMP12 "], %[" #TEMP12 "], 12000 \n\t" \
465 "addiu %["#TEMP8"], %["#TEMP8"], 21000 \n\t" \ 465 "addiu %[" #TEMP8 "], %[" #TEMP8 "], 21000 \n\t" \
466 "subu %["#TEMP8"], %["#TEMP8"], %[temp18] \n\t" \ 466 "subu %[" #TEMP8 "], %[" #TEMP8 "], %[temp18] \n\t" \
467 "sra %["#TEMP12"], %["#TEMP12"], 16 \n\t" \ 467 "sra %[" #TEMP12 "], %[" #TEMP12 "], 16 \n\t" \
468 "sra %["#TEMP8"], %["#TEMP8"], 16 \n\t" \ 468 "sra %[" #TEMP8 "], %[" #TEMP8 "], 16 \n\t" \
469 "addiu %[temp16], %["#TEMP12"], 1 \n\t" \ 469 "addiu %[temp16], %[" #TEMP12 "], 1 \n\t" \
470 "movn %["#TEMP12"], %[temp16], %[temp19] \n\t" \ 470 "movn %[" #TEMP12 "], %[temp16], %[temp19] \n\t" \
471 "sh %["#TEMP0"], "#A"(%[temp20]) \n\t" \ 471 "sh %[" #TEMP0 "], " #A "(%[temp20]) \n\t" \
472 "sh %["#TEMP4"], "#C"(%[temp20]) \n\t" \ 472 "sh %[" #TEMP4 "], " #C "(%[temp20]) \n\t" \
473 "sh %["#TEMP8"], "#D"(%[temp20]) \n\t" \ 473 "sh %[" #TEMP8 "], " #D "(%[temp20]) \n\t" \
474 "sh %["#TEMP12"], "#B"(%[temp20]) \n\t" 474 "sh %[" #TEMP12 "], " #B "(%[temp20]) \n\t"
475 475
476 static void FTransform(const uint8_t* src, const uint8_t* ref, int16_t* out) { 476 static void FTransform(const uint8_t* src, const uint8_t* ref, int16_t* out) {
477 int temp0, temp1, temp2, temp3, temp4, temp5, temp6, temp7, temp8; 477 int temp0, temp1, temp2, temp3, temp4, temp5, temp6, temp7, temp8;
478 int temp9, temp10, temp11, temp12, temp13, temp14, temp15, temp16; 478 int temp9, temp10, temp11, temp12, temp13, temp14, temp15, temp16;
479 int temp17, temp18, temp19, temp20; 479 int temp17, temp18, temp19, temp20;
480 const int c2217 = 2217; 480 const int c2217 = 2217;
481 const int c5352 = 5352; 481 const int c5352 = 5352;
482 const int* const args[3] = 482 const int* const args[3] =
483 { (const int*)src, (const int*)ref, (const int*)out }; 483 { (const int*)src, (const int*)ref, (const int*)out };
484 484
(...skipping 130 matching lines...) Expand 10 before | Expand all | Expand 10 after
615 const int b = VP8EncBands[n + 1]; 615 const int b = VP8EncBands[n + 1];
616 const int ctx = (v == 1) ? 1 : 2; 616 const int ctx = (v == 1) ? 1 : 2;
617 const int last_p0 = res->prob[b][ctx][0]; 617 const int last_p0 = res->prob[b][ctx][0];
618 cost += VP8BitCost(0, last_p0); 618 cost += VP8BitCost(0, last_p0);
619 } 619 }
620 } 620 }
621 return cost; 621 return cost;
622 } 622 }
623 623
624 #define GET_SSE_INNER(A, B, C, D) \ 624 #define GET_SSE_INNER(A, B, C, D) \
625 "lbu %[temp0], "#A"(%[a]) \n\t" \ 625 "lbu %[temp0], " #A "(%[a]) \n\t" \
626 "lbu %[temp1], "#A"(%[b]) \n\t" \ 626 "lbu %[temp1], " #A "(%[b]) \n\t" \
627 "lbu %[temp2], "#B"(%[a]) \n\t" \ 627 "lbu %[temp2], " #B "(%[a]) \n\t" \
628 "lbu %[temp3], "#B"(%[b]) \n\t" \ 628 "lbu %[temp3], " #B "(%[b]) \n\t" \
629 "lbu %[temp4], "#C"(%[a]) \n\t" \ 629 "lbu %[temp4], " #C "(%[a]) \n\t" \
630 "lbu %[temp5], "#C"(%[b]) \n\t" \ 630 "lbu %[temp5], " #C "(%[b]) \n\t" \
631 "lbu %[temp6], "#D"(%[a]) \n\t" \ 631 "lbu %[temp6], " #D "(%[a]) \n\t" \
632 "lbu %[temp7], "#D"(%[b]) \n\t" \ 632 "lbu %[temp7], " #D "(%[b]) \n\t" \
633 "subu %[temp0], %[temp0], %[temp1] \n\t" \ 633 "subu %[temp0], %[temp0], %[temp1] \n\t" \
634 "subu %[temp2], %[temp2], %[temp3] \n\t" \ 634 "subu %[temp2], %[temp2], %[temp3] \n\t" \
635 "subu %[temp4], %[temp4], %[temp5] \n\t" \ 635 "subu %[temp4], %[temp4], %[temp5] \n\t" \
636 "subu %[temp6], %[temp6], %[temp7] \n\t" \ 636 "subu %[temp6], %[temp6], %[temp7] \n\t" \
637 "madd %[temp0], %[temp0] \n\t" \ 637 "madd %[temp0], %[temp0] \n\t" \
638 "madd %[temp2], %[temp2] \n\t" \ 638 "madd %[temp2], %[temp2] \n\t" \
639 "madd %[temp4], %[temp4] \n\t" \ 639 "madd %[temp4], %[temp4] \n\t" \
640 "madd %[temp6], %[temp6] \n\t" 640 "madd %[temp6], %[temp6] \n\t"
641 641
642 #define GET_SSE(A, B, C, D) \ 642 #define GET_SSE(A, B, C, D) \
(...skipping 124 matching lines...) Expand 10 before | Expand all | Expand 10 after
767 VP8TDisto16x16 = Disto16x16; 767 VP8TDisto16x16 = Disto16x16;
768 VP8FTransform = FTransform; 768 VP8FTransform = FTransform;
769 #if !defined(WORK_AROUND_GCC) 769 #if !defined(WORK_AROUND_GCC)
770 VP8SSE16x16 = SSE16x16; 770 VP8SSE16x16 = SSE16x16;
771 VP8SSE8x8 = SSE8x8; 771 VP8SSE8x8 = SSE8x8;
772 VP8SSE16x8 = SSE16x8; 772 VP8SSE16x8 = SSE16x8;
773 VP8SSE4x4 = SSE4x4; 773 VP8SSE4x4 = SSE4x4;
774 #endif 774 #endif
775 #endif // WEBP_USE_MIPS32 775 #endif // WEBP_USE_MIPS32
776 } 776 }
OLDNEW
« no previous file with comments | « third_party/libwebp/dsp/dsp.h ('k') | third_party/libwebp/dsp/lossless_mips32.c » ('j') | no next file with comments »

Powered by Google App Engine
This is Rietveld 408576698