ps2 idct patch by (Leon van Stuivenberg <leonvs at iae dot nl>)
[libav.git] / libavcodec / ps2 / dsputil_mmi.c
1 /*
2 * MMI optimized DSP utils
3 * Copyright (c) 2000, 2001 Fabrice Bellard.
4 *
5 * This library is free software; you can redistribute it and/or
6 * modify it under the terms of the GNU Lesser General Public
7 * License as published by the Free Software Foundation; either
8 * version 2 of the License, or (at your option) any later version.
9 *
10 * This library is distributed in the hope that it will be useful,
11 * but WITHOUT ANY WARRANTY; without even the implied warranty of
12 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
13 * Lesser General Public License for more details.
14 *
15 * You should have received a copy of the GNU Lesser General Public
16 * License along with this library; if not, write to the Free Software
17 * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
18 *
19 * MMI optimization by Leon van Stuivenberg <leonvs@iae.nl>
20 */
21
22 #include "../dsputil.h"
23
24 void ff_mmi_idct(DCTELEM * block);
25
26 #include "mmi.h"
27
28
29 static void clear_blocks_mmi(DCTELEM * blocks)
30 {
31 /* $4 = blocks */
32 int i;
33 for (i = 0; i < 6; i++) {
34 sq($0, 0, $4);
35 sq($0, 16, $4);
36 sq($0, 32, $4);
37 sq($0, 48, $4);
38 sq($0, 64, $4);
39 sq($0, 80, $4);
40 sq($0, 96, $4);
41 sq($0, 112, $4);
42 __asm__ __volatile__("addi $4, $4, 128");
43 }
44 }
45
46
47 static void put_pixels_clamped_mmi(const DCTELEM * block, UINT8 * pixels,
48 int line_size)
49 {
50 /* $4 = block, $5 = pixels, $6 = line_size */
51 __asm__ __volatile__("li $11, 255":::"$11");
52 lq($4, 0, $12);
53 pcpyld($11, $11, $11);
54 pcpyh($11, $11);
55
56 #define PUT(rs) \
57 ppacb($0, $##rs, $##rs); \
58 sd3(rs, 0, 5); \
59 __asm__ __volatile__ ("add $5, $5, $6");
60
61 pminh($12, $11, $12);
62 pmaxh($12, $0, $12);
63 lq($4, 16, $13);
64 PUT(12);
65
66 pminh($13, $11, $13);
67 pmaxh($13, $0, $13);
68 lq($4, 32, $12);
69 PUT(13);
70
71 pminh($12, $11, $12);
72 pmaxh($12, $0, $12);
73 lq($4, 48, $13);
74 PUT(12);
75
76 pminh($13, $11, $13);
77 pmaxh($13, $0, $13);
78 lq($4, 64, $12);
79 PUT(13);
80
81 pminh($12, $11, $12);
82 pmaxh($12, $0, $12);
83 lq($4, 80, $13);
84 PUT(12);
85
86 pminh($13, $11, $13);
87 pmaxh($13, $0, $13);
88 lq($4, 96, $12);
89 PUT(13);
90
91 pminh($12, $11, $12);
92 pmaxh($12, $0, $12);
93 lq($4, 112, $13);
94 PUT(12);
95
96 pminh($13, $11, $13);
97 pmaxh($13, $0, $13);
98 PUT(13);
99 }
100
101 /* todo
102 static void add_pixels_clamped_mmi(const DCTELEM * block, UINT8 * pixels,
103 int line_size)
104 {
105 }
106 */
107
108
109 void dsputil_init_mmi(void)
110 {
111 put_pixels_clamped = put_pixels_clamped_mmi;
112 //add_pixels_clamped = add_pixels_clamped_mmi;
113 clear_blocks = clear_blocks_mmi;
114 ff_idct = ff_mmi_idct;
115 }