2 * MMI optimized DSP utils
3 * Copyright (c) 2000, 2001 Fabrice Bellard
5 * MMI optimization by Leon van Stuivenberg
6 * clear_blocks_mmi() by BroadQ
8 * This file is part of Libav.
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.
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.
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
25 #include "libavcodec/dsputil.h"
28 static void clear_blocks_mmi(DCTELEM * blocks)
32 "addiu $9, %0, 768 \n"
44 : "+r" (blocks) :: "$8", "$9", "memory" );
48 static void get_pixels_mmi(DCTELEM *block, const uint8_t *pixels, int line_size)
58 "pextlb $8, $0, $8 \n\t"
62 "pextlb $9, $0, $9 \n\t"
66 "pextlb $10, $0, $10 \n\t"
70 "pextlb $8, $0, $8 \n\t"
74 "pextlb $9, $0, $9 \n\t"
78 "pextlb $10, $0, $10 \n\t"
80 "pextlb $8, $0, $8 \n\t"
82 "pextlb $9, $0, $9 \n\t"
85 : "+r" (pixels) : "r" (block), "r" (line_size) : "$8", "$9", "$10", "memory" );
89 static void put_pixels8_mmi(uint8_t *block, const uint8_t *pixels, int line_size, int h)
96 "addiu %2, %2, -1 \n\t"
100 "add %0, %0, %3 \n\t"
103 : "+r" (block), "+r" (pixels), "+r" (h) : "r" (line_size)
108 static void put_pixels16_mmi(uint8_t *block, const uint8_t *pixels, int line_size, int h)
115 "add $11, %1, %3 \n\t"
117 "add $10, %0, %3 \n\t"
119 "ldl $9, 15(%1) \n\t"
120 "ldr $12, 0($11) \n\t"
121 "add %1, $11, %3 \n\t"
122 "ldl $12, 7($11) \n\t"
123 "pcpyld $8, $9, $8 \n\t"
125 "ldr $13, 8($11) \n\t"
126 "addiu %2, %2, -2 \n\t"
127 "ldl $13, 15($11) \n\t"
128 "add %0, $10, %3 \n\t"
129 "pcpyld $12, $13, $12 \n\t"
130 "sq $12, 0($10) \n\t"
133 : "+r" (block), "+r" (pixels), "+r" (h) : "r" (line_size)
134 : "$8", "$9", "$10", "$11", "$12", "$13", "memory" );
138 void ff_dsputil_init_mmi(DSPContext* c, AVCodecContext *avctx)
140 const int idct_algo= avctx->idct_algo;
141 const int high_bit_depth = avctx->bits_per_raw_sample > 8;
143 if (!high_bit_depth) {
144 c->clear_blocks = clear_blocks_mmi;
146 c->put_pixels_tab[1][0] = put_pixels8_mmi;
147 c->put_no_rnd_pixels_tab[1][0] = put_pixels8_mmi;
149 c->put_pixels_tab[0][0] = put_pixels16_mmi;
150 c->put_no_rnd_pixels_tab[0][0] = put_pixels16_mmi;
152 c->get_pixels = get_pixels_mmi;
155 if (avctx->bits_per_raw_sample <= 8 &&
156 (idct_algo == FF_IDCT_AUTO || idct_algo == FF_IDCT_MMI)) {
157 c->idct_put= ff_mmi_idct_put;
158 c->idct_add= ff_mmi_idct_add;
159 c->idct = ff_mmi_idct;
160 c->idct_permutation_type= FF_LIBMPEG2_IDCT_PERM;