clarify *_offsets.asm differences
[libvpx.git] / vp8 / encoder / dct.c
blobb5a11ae34f9b8bfe3197117fdd7b9389017ade3d
1 /*
2 * Copyright (c) 2010 The WebM project authors. All Rights Reserved.
4 * Use of this source code is governed by a BSD-style license
5 * that can be found in the LICENSE file in the root of the source
6 * tree. An additional intellectual property rights grant can be found
7 * in the file PATENTS. All contributing project authors may
8 * be found in the AUTHORS file in the root of the source tree.
9 */
12 #include <math.h>
14 void vp8_short_fdct4x4_c(short *input, short *output, int pitch)
16 int i;
17 int a1, b1, c1, d1;
18 short *ip = input;
19 short *op = output;
21 for (i = 0; i < 4; i++)
23 a1 = ((ip[0] + ip[3])<<3);
24 b1 = ((ip[1] + ip[2])<<3);
25 c1 = ((ip[1] - ip[2])<<3);
26 d1 = ((ip[0] - ip[3])<<3);
28 op[0] = a1 + b1;
29 op[2] = a1 - b1;
31 op[1] = (c1 * 2217 + d1 * 5352 + 14500)>>12;
32 op[3] = (d1 * 2217 - c1 * 5352 + 7500)>>12;
34 ip += pitch / 2;
35 op += 4;
38 ip = output;
39 op = output;
40 for (i = 0; i < 4; i++)
42 a1 = ip[0] + ip[12];
43 b1 = ip[4] + ip[8];
44 c1 = ip[4] - ip[8];
45 d1 = ip[0] - ip[12];
47 op[0] = ( a1 + b1 + 7)>>4;
48 op[8] = ( a1 - b1 + 7)>>4;
50 op[4] =((c1 * 2217 + d1 * 5352 + 12000)>>16) + (d1!=0);
51 op[12] = (d1 * 2217 - c1 * 5352 + 51000)>>16;
53 ip++;
54 op++;
58 void vp8_short_fdct8x4_c(short *input, short *output, int pitch)
60 vp8_short_fdct4x4_c(input, output, pitch);
61 vp8_short_fdct4x4_c(input + 4, output + 16, pitch);
64 void vp8_short_walsh4x4_c(short *input, short *output, int pitch)
66 int i;
67 int a1, b1, c1, d1;
68 int a2, b2, c2, d2;
69 short *ip = input;
70 short *op = output;
73 for (i = 0; i < 4; i++)
75 a1 = ((ip[0] + ip[2])<<2);
76 d1 = ((ip[1] + ip[3])<<2);
77 c1 = ((ip[1] - ip[3])<<2);
78 b1 = ((ip[0] - ip[2])<<2);
80 op[0] = a1 + d1 + (a1!=0);
81 op[1] = b1 + c1;
82 op[2] = b1 - c1;
83 op[3] = a1 - d1;
84 ip += pitch / 2;
85 op += 4;
88 ip = output;
89 op = output;
91 for (i = 0; i < 4; i++)
93 a1 = ip[0] + ip[8];
94 d1 = ip[4] + ip[12];
95 c1 = ip[4] - ip[12];
96 b1 = ip[0] - ip[8];
98 a2 = a1 + d1;
99 b2 = b1 + c1;
100 c2 = b1 - c1;
101 d2 = a1 - d1;
103 a2 += a2<0;
104 b2 += b2<0;
105 c2 += c2<0;
106 d2 += d2<0;
108 op[0] = (a2+3) >> 3;
109 op[4] = (b2+3) >> 3;
110 op[8] = (c2+3) >> 3;
111 op[12]= (d2+3) >> 3;
113 ip++;
114 op++;