Mercurial > libavcodec.hg

comparison vp3dsp.c @ 3503:074c9f3d8e62 libavcodec

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
remove a few useless casts and avoid the useless t1/t2 variables
author michael
date Wed, 19 Jul 2006 23:17:18 +0000
parents 0b546eab515d
children 40d83cd39561
comparison
equal deleted inserted replaced
3502:47c9a225c4d4 3503:074c9f3d8e62
33 #define xC4S4 46341 33 #define xC4S4 46341
34 #define xC5S3 36410 34 #define xC5S3 36410
35 #define xC6S2 25080 35 #define xC6S2 25080
36 #define xC7S1 12785 36 #define xC7S1 12785
37 37
38 #define M(a,b) (((a) * (b))>>16)
39
38 static always_inline void idct(uint8_t *dst, int stride, int16_t *input, int type) 40 static always_inline void idct(uint8_t *dst, int stride, int16_t *input, int type)
39 { 41 {
40 int16_t *ip = input; 42 int16_t *ip = input;
41 uint8_t *cm = cropTbl + MAX_NEG_CROP; 43 uint8_t *cm = cropTbl + MAX_NEG_CROP;
42 44
43 int A_, B_, C_, D_, _Ad, _Bd, _Cd, _Dd, E_, F_, G_, H_; 45 int A_, B_, C_, D_, _Ad, _Bd, _Cd, _Dd, E_, F_, G_, H_;
44 int _Ed, _Gd, _Add, _Bdd, _Fd, _Hd; 46 int _Ed, _Gd, _Add, _Bdd, _Fd, _Hd;
45 int t1, t2;
46 47
47 int i; 48 int i;
48 49
49 /* Inverse DCT on the rows now */ 50 /* Inverse DCT on the rows now */
50 for (i = 0; i < 8; i++) { 51 for (i = 0; i < 8; i++) {
51 /* Check for non-zero values */ 52 /* Check for non-zero values */
52 if ( ip[0] | ip[1] | ip[2] | ip[3] | ip[4] | ip[5] | ip[6] | ip[7] ) { 53 if ( ip[0] | ip[1] | ip[2] | ip[3] | ip[4] | ip[5] | ip[6] | ip[7] ) {
53 t1 = (int32_t)(xC1S7 * ip[1]); 54 A_ = M(xC1S7, ip[1]) + M(xC7S1, ip[7]);
54 t2 = (int32_t)(xC7S1 * ip[7]); 55 B_ = M(xC7S1, ip[1]) - M(xC1S7, ip[7]);
55 t1 >>= 16; 56 C_ = M(xC3S5, ip[3]) + M(xC5S3, ip[5]);
56 t2 >>= 16; 57 D_ = M(xC3S5, ip[5]) - M(xC5S3, ip[3]);
57 A_ = t1 + t2; 58
58 59 _Ad = M(xC4S4, (A_ - C_));
59 t1 = (int32_t)(xC7S1 * ip[1]); 60 _Bd = M(xC4S4, (B_ - D_));
60 t2 = (int32_t)(xC1S7 * ip[7]);
61 t1 >>= 16;
62 t2 >>= 16;
63 B_ = t1 - t2;
64
65 t1 = (int32_t)(xC3S5 * ip[3]);
66 t2 = (int32_t)(xC5S3 * ip[5]);
67 t1 >>= 16;
68 t2 >>= 16;
69 C_ = t1 + t2;
70
71 t1 = (int32_t)(xC3S5 * ip[5]);
72 t2 = (int32_t)(xC5S3 * ip[3]);
73 t1 >>= 16;
74 t2 >>= 16;
75 D_ = t1 - t2;
76
77
78 t1 = (int32_t)(xC4S4 * (A_ - C_));
79 t1 >>= 16;
80 _Ad = t1;
81
82 t1 = (int32_t)(xC4S4 * (B_ - D_));
83 t1 >>= 16;
84 _Bd = t1;
85
86 61
87 _Cd = A_ + C_; 62 _Cd = A_ + C_;
88 _Dd = B_ + D_; 63 _Dd = B_ + D_;
89 64
90 t1 = (int32_t)(xC4S4 * (ip[0] + ip[4])); 65 E_ = M(xC4S4, (ip[0] + ip[4]));
91 t1 >>= 16; 66 F_ = M(xC4S4, (ip[0] - ip[4]));
92 E_ = t1; 67
93 68 G_ = M(xC2S6, ip[2]) + M(xC6S2, ip[6]);
94 t1 = (int32_t)(xC4S4 * (ip[0] - ip[4])); 69 H_ = M(xC6S2, ip[2]) - M(xC2S6, ip[6]);
95 t1 >>= 16;
96 F_ = t1;
97
98 t1 = (int32_t)(xC2S6 * ip[2]);
99 t2 = (int32_t)(xC6S2 * ip[6]);
100 t1 >>= 16;
101 t2 >>= 16;
102 G_ = t1 + t2;
103
104 t1 = (int32_t)(xC6S2 * ip[2]);
105 t2 = (int32_t)(xC2S6 * ip[6]);
106 t1 >>= 16;
107 t2 >>= 16;
108 H_ = t1 - t2;
109
110 70
111 _Ed = E_ - G_; 71 _Ed = E_ - G_;
112 _Gd = E_ + G_; 72 _Gd = E_ + G_;
113 73
114 _Add = F_ + _Ad; 74 _Add = F_ + _Ad;
127 ip[3] = _Ed + _Dd ; 87 ip[3] = _Ed + _Dd ;
128 ip[4] = _Ed - _Dd ; 88 ip[4] = _Ed - _Dd ;
129 89
130 ip[5] = _Fd + _Bdd; 90 ip[5] = _Fd + _Bdd;
131 ip[6] = _Fd - _Bdd; 91 ip[6] = _Fd - _Bdd;
132
133 } 92 }
134 93
135 ip += 8; /* next row */ 94 ip += 8; /* next row */
136 } 95 }
137 96
140 for ( i = 0; i < 8; i++) { 99 for ( i = 0; i < 8; i++) {
141 /* Check for non-zero values (bitwise or faster than ||) */ 100 /* Check for non-zero values (bitwise or faster than ||) */
142 if ( ip[1 * 8] | ip[2 * 8] | ip[3 * 8] | 101 if ( ip[1 * 8] | ip[2 * 8] | ip[3 * 8] |
143 ip[4 * 8] | ip[5 * 8] | ip[6 * 8] | ip[7 * 8] ) { 102 ip[4 * 8] | ip[5 * 8] | ip[6 * 8] | ip[7 * 8] ) {
144 103
145 t1 = (int32_t)(xC1S7 * ip[1*8]); 104 A_ = M(xC1S7, ip[1*8]) + M(xC7S1, ip[7*8]);
146 t2 = (int32_t)(xC7S1 * ip[7*8]); 105 B_ = M(xC7S1, ip[1*8]) - M(xC1S7, ip[7*8]);
147 t1 >>= 16; 106 C_ = M(xC3S5, ip[3*8]) + M(xC5S3, ip[5*8]);
148 t2 >>= 16; 107 D_ = M(xC3S5, ip[5*8]) - M(xC5S3, ip[3*8]);
149 A_ = t1 + t2; 108
150 109 _Ad = M(xC4S4, (A_ - C_));
151 t1 = (int32_t)(xC7S1 * ip[1*8]); 110 _Bd = M(xC4S4, (B_ - D_));
152 t2 = (int32_t)(xC1S7 * ip[7*8]);
153 t1 >>= 16;
154 t2 >>= 16;
155 B_ = t1 - t2;
156
157 t1 = (int32_t)(xC3S5 * ip[3*8]);
158 t2 = (int32_t)(xC5S3 * ip[5*8]);
159 t1 >>= 16;
160 t2 >>= 16;
161 C_ = t1 + t2;
162
163 t1 = (int32_t)(xC3S5 * ip[5*8]);
164 t2 = (int32_t)(xC5S3 * ip[3*8]);
165 t1 >>= 16;
166 t2 >>= 16;
167 D_ = t1 - t2;
168
169
170 t1 = (int32_t)(xC4S4 * (A_ - C_));
171 t1 >>= 16;
172 _Ad = t1;
173
174 t1 = (int32_t)(xC4S4 * (B_ - D_));
175 t1 >>= 16;
176 _Bd = t1;
177
178 111
179 _Cd = A_ + C_; 112 _Cd = A_ + C_;
180 _Dd = B_ + D_; 113 _Dd = B_ + D_;
181 114
182 t1 = (int32_t)(xC4S4 * (ip[0*8] + ip[4*8])); 115 E_ = M(xC4S4, (ip[0*8] + ip[4*8]));
183 t1 >>= 16; 116 F_ = M(xC4S4, (ip[0*8] - ip[4*8]));
184 E_ = t1; 117
185 118 G_ = M(xC2S6, ip[2*8]) + M(xC6S2, ip[6*8]);
186 t1 = (int32_t)(xC4S4 * (ip[0*8] - ip[4*8])); 119 H_ = M(xC6S2, ip[2*8]) - M(xC2S6, ip[6*8]);
187 t1 >>= 16;
188 F_ = t1;
189
190 t1 = (int32_t)(xC2S6 * ip[2*8]);
191 t2 = (int32_t)(xC6S2 * ip[6*8]);
192 t1 >>= 16;
193 t2 >>= 16;
194 G_ = t1 + t2;
195
196 t1 = (int32_t)(xC6S2 * ip[2*8]);
197 t2 = (int32_t)(xC2S6 * ip[6*8]);
198 t1 >>= 16;
199 t2 >>= 16;
200 H_ = t1 - t2;
201
202 120
203 _Ed = E_ - G_; 121 _Ed = E_ - G_;
204 _Gd = E_ + G_; 122 _Gd = E_ + G_;
205 123
206 _Add = F_ + _Ad; 124 _Add = F_ + _Ad;