dsputil_mmi.c
Go to the documentation of this file.
1 /*
2  * MMI optimized DSP utils
3  * Copyright (c) 2000, 2001 Fabrice Bellard
4  *
5  * MMI optimization by Leon van Stuivenberg
6  * clear_blocks_mmi() by BroadQ
7  *
8  * This file is part of Libav.
9  *
10  * Libav is free software; you can redistribute it and/or
11  * modify it under the terms of the GNU Lesser General Public
12  * License as published by the Free Software Foundation; either
13  * version 2.1 of the License, or (at your option) any later version.
14  *
15  * Libav is distributed in the hope that it will be useful,
16  * but WITHOUT ANY WARRANTY; without even the implied warranty of
17  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
18  * Lesser General Public License for more details.
19  *
20  * You should have received a copy of the GNU Lesser General Public
21  * License along with Libav; if not, write to the Free Software
22  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
23  */
24 
25 #include "libavcodec/dsputil.h"
26 #include "mmi.h"
27 
28 void ff_mmi_idct_put(uint8_t *dest, int line_size, DCTELEM *block);
29 void ff_mmi_idct_add(uint8_t *dest, int line_size, DCTELEM *block);
31 
32 static void clear_blocks_mmi(DCTELEM * blocks)
33 {
34  __asm__ volatile(
35  ".set noreorder \n"
36  "addiu $9, %0, 768 \n"
37  "nop \n"
38  "1: \n"
39  "sq $0, 0(%0) \n"
40  "move $8, %0 \n"
41  "addi %0, %0, 64 \n"
42  "sq $0, 16($8) \n"
43  "slt $10, %0, $9 \n"
44  "sq $0, 32($8) \n"
45  "bnez $10, 1b \n"
46  "sq $0, 48($8) \n"
47  ".set reorder \n"
48  : "+r" (blocks) :: "$8", "$9", "memory" );
49 }
50 
51 
52 static void get_pixels_mmi(DCTELEM *block, const uint8_t *pixels, int line_size)
53 {
54  __asm__ volatile(
55  ".set push \n\t"
56  ".set mips3 \n\t"
57  "ld $8, 0(%0) \n\t"
58  "add %0, %0, %2 \n\t"
59  "ld $9, 0(%0) \n\t"
60  "add %0, %0, %2 \n\t"
61  "ld $10, 0(%0) \n\t"
62  "pextlb $8, $0, $8 \n\t"
63  "sq $8, 0(%1) \n\t"
64  "add %0, %0, %2 \n\t"
65  "ld $8, 0(%0) \n\t"
66  "pextlb $9, $0, $9 \n\t"
67  "sq $9, 16(%1) \n\t"
68  "add %0, %0, %2 \n\t"
69  "ld $9, 0(%0) \n\t"
70  "pextlb $10, $0, $10 \n\t"
71  "sq $10, 32(%1) \n\t"
72  "add %0, %0, %2 \n\t"
73  "ld $10, 0(%0) \n\t"
74  "pextlb $8, $0, $8 \n\t"
75  "sq $8, 48(%1) \n\t"
76  "add %0, %0, %2 \n\t"
77  "ld $8, 0(%0) \n\t"
78  "pextlb $9, $0, $9 \n\t"
79  "sq $9, 64(%1) \n\t"
80  "add %0, %0, %2 \n\t"
81  "ld $9, 0(%0) \n\t"
82  "pextlb $10, $0, $10 \n\t"
83  "sq $10, 80(%1) \n\t"
84  "pextlb $8, $0, $8 \n\t"
85  "sq $8, 96(%1) \n\t"
86  "pextlb $9, $0, $9 \n\t"
87  "sq $9, 112(%1) \n\t"
88  ".set pop \n\t"
89  : "+r" (pixels) : "r" (block), "r" (line_size) : "$8", "$9", "$10", "memory" );
90 }
91 
92 
93 static void put_pixels8_mmi(uint8_t *block, const uint8_t *pixels, int line_size, int h)
94 {
95  __asm__ volatile(
96  ".set push \n\t"
97  ".set mips3 \n\t"
98  "1: \n\t"
99  "ldr $8, 0(%1) \n\t"
100  "addiu %2, %2, -1 \n\t"
101  "ldl $8, 7(%1) \n\t"
102  "add %1, %1, %3 \n\t"
103  "sd $8, 0(%0) \n\t"
104  "add %0, %0, %3 \n\t"
105  "bgtz %2, 1b \n\t"
106  ".set pop \n\t"
107  : "+r" (block), "+r" (pixels), "+r" (h) : "r" (line_size)
108  : "$8", "memory" );
109 }
110 
111 
112 static void put_pixels16_mmi(uint8_t *block, const uint8_t *pixels, int line_size, int h)
113 {
114  __asm__ volatile (
115  ".set push \n\t"
116  ".set mips3 \n\t"
117  "1: \n\t"
118  "ldr $8, 0(%1) \n\t"
119  "add $11, %1, %3 \n\t"
120  "ldl $8, 7(%1) \n\t"
121  "add $10, %0, %3 \n\t"
122  "ldr $9, 8(%1) \n\t"
123  "ldl $9, 15(%1) \n\t"
124  "ldr $12, 0($11) \n\t"
125  "add %1, $11, %3 \n\t"
126  "ldl $12, 7($11) \n\t"
127  "pcpyld $8, $9, $8 \n\t"
128  "sq $8, 0(%0) \n\t"
129  "ldr $13, 8($11) \n\t"
130  "addiu %2, %2, -2 \n\t"
131  "ldl $13, 15($11) \n\t"
132  "add %0, $10, %3 \n\t"
133  "pcpyld $12, $13, $12 \n\t"
134  "sq $12, 0($10) \n\t"
135  "bgtz %2, 1b \n\t"
136  ".set pop \n\t"
137  : "+r" (block), "+r" (pixels), "+r" (h) : "r" (line_size)
138  : "$8", "$9", "$10", "$11", "$12", "$13", "memory" );
139 }
140 
141 
143 {
144  const int idct_algo= avctx->idct_algo;
145  const int high_bit_depth = avctx->bits_per_raw_sample > 8;
146 
147  if (!high_bit_depth) {
149 
150  c->put_pixels_tab[1][0] = put_pixels8_mmi;
152 
153  c->put_pixels_tab[0][0] = put_pixels16_mmi;
155 
157  }
158 
159  if (avctx->bits_per_raw_sample <= 8 &&
160  (idct_algo==FF_IDCT_AUTO || idct_algo==FF_IDCT_PS2)) {
163  c->idct = ff_mmi_idct;
165  }
166 }
167