4 * Copyright (C) 2007 Marc Hoffman <marc.hoffman@analog.com>
5 * Copyright (c) 2006 Michael Benjamin <michael.benjamin@analog.com>
7 * This file is part of FFmpeg.
9 * FFmpeg is free software; you can redistribute it and/or
10 * modify it under the terms of the GNU Lesser General Public
11 * License as published by the Free Software Foundation; either
12 * version 2.1 of the License, or (at your option) any later version.
14 * FFmpeg is distributed in the hope that it will be useful,
15 * but WITHOUT ANY WARRANTY; without even the implied warranty of
16 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
17 * Lesser General Public License for more details.
19 * You should have received a copy of the GNU Lesser General Public
20 * License along with FFmpeg; if not, write to the Free Software
21 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
24 #include "libavutil/attributes.h"
25 #include "libavcodec/avcodec.h"
26 #include "libavcodec/dsputil.h"
27 #include "dsputil_bfin.h"
31 static void bfin_idct_add (uint8_t *dest, int line_size, int16_t *block)
34 ff_bfin_add_pixels_clamped (block, dest, line_size);
37 static void bfin_idct_put (uint8_t *dest, int line_size, int16_t *block)
40 ff_bfin_put_pixels_clamped (block, dest, line_size);
44 static void bfin_clear_blocks (int16_t *blocks)
46 // This is just a simple memset.
51 "LSETUP(clear_blocks_blkfn_lab,clear_blocks_blkfn_lab)LC0=P0;"
52 "clear_blocks_blkfn_lab:"
54 ::"a" (blocks):"P0","I0","R0");
57 static int bfin_pix_abs16 (void *c, uint8_t *blk1, uint8_t *blk2, int line_size, int h)
59 return ff_bfin_z_sad16x16 (blk1,blk2,line_size,line_size,h);
62 static int bfin_vsad_intra16 (void *c, uint8_t *blk1, uint8_t *dummy, int stride, int h) {
63 return ff_bfin_z_sad16x16 (blk1,blk1+stride,stride<<1,stride<<1,h);
66 static int bfin_vsad (void *c, uint8_t *blk1, uint8_t *blk2, int stride, int h) {
67 return ff_bfin_z_sad16x16 (blk1,blk1+stride,stride<<1,stride<<1,h)
68 + ff_bfin_z_sad16x16 (blk2,blk2+stride,stride<<1,stride<<1,h);
71 static uint8_t vtmp_blk[256] attribute_l1_data_b;
73 static int bfin_pix_abs16_x2 (void *c, uint8_t *blk1, uint8_t *blk2, int line_size, int h)
75 ff_bfin_put_pixels16uc (vtmp_blk, blk2, blk2+1, 16, line_size, h);
76 return ff_bfin_z_sad16x16 (blk1, vtmp_blk, line_size, 16, h);
79 static int bfin_pix_abs16_y2 (void *c, uint8_t *blk1, uint8_t *blk2, int line_size, int h)
81 ff_bfin_put_pixels16uc (vtmp_blk, blk2, blk2+line_size, 16, line_size, h);
82 return ff_bfin_z_sad16x16 (blk1, vtmp_blk, line_size, 16, h);
85 static int bfin_pix_abs16_xy2 (void *c, uint8_t *blk1, uint8_t *blk2, int line_size, int h)
87 ff_bfin_z_put_pixels16_xy2 (vtmp_blk, blk2, 16, line_size, h);
88 return ff_bfin_z_sad16x16 (blk1, vtmp_blk, line_size, 16, h);
91 static int bfin_pix_abs8 (void *c, uint8_t *blk1, uint8_t *blk2, int line_size, int h)
93 return ff_bfin_z_sad8x8 (blk1,blk2,line_size,line_size, h);
96 static int bfin_pix_abs8_x2 (void *c, uint8_t *blk1, uint8_t *blk2, int line_size, int h)
98 ff_bfin_put_pixels8uc (vtmp_blk, blk2, blk2+1, 8, line_size, h);
99 return ff_bfin_z_sad8x8 (blk1, vtmp_blk, line_size, 8, h);
102 static int bfin_pix_abs8_y2 (void *c, uint8_t *blk1, uint8_t *blk2, int line_size, int h)
104 ff_bfin_put_pixels8uc (vtmp_blk, blk2, blk2+line_size, 8, line_size, h);
105 return ff_bfin_z_sad8x8 (blk1, vtmp_blk, line_size, 8, h);
108 static int bfin_pix_abs8_xy2 (void *c, uint8_t *blk1, uint8_t *blk2, int line_size, int h)
110 ff_bfin_z_put_pixels8_xy2 (vtmp_blk, blk2, 8, line_size, h);
111 return ff_bfin_z_sad8x8 (blk1, vtmp_blk, line_size, 8, h);
117 start on 2/11 100 frames of 352x240@25 compiled with no optimization -g debugging
119 6.360s ~ 1.58x off with -O2
120 5.740s ~ 1.43x off with idcts
122 2.64s 2/20 same sman.mp4 decode only
126 av_cold void ff_dsputil_init_bfin(DSPContext *c, AVCodecContext *avctx)
128 const int high_bit_depth = avctx->bits_per_raw_sample > 8;
130 c->diff_pixels = ff_bfin_diff_pixels;
131 c->put_pixels_clamped = ff_bfin_put_pixels_clamped;
132 c->add_pixels_clamped = ff_bfin_add_pixels_clamped;
135 c->get_pixels = ff_bfin_get_pixels;
136 c->clear_blocks = bfin_clear_blocks;
137 c->pix_sum = ff_bfin_pix_sum;
138 c->pix_norm1 = ff_bfin_pix_norm1;
140 c->sad[0] = bfin_pix_abs16;
141 c->sad[1] = bfin_pix_abs8;
143 /* c->vsad[0] = bfin_vsad; */
144 /* c->vsad[4] = bfin_vsad_intra16; */
146 /* TODO [0] 16 [1] 8 */
147 c->pix_abs[0][0] = bfin_pix_abs16;
148 c->pix_abs[0][1] = bfin_pix_abs16_x2;
149 c->pix_abs[0][2] = bfin_pix_abs16_y2;
150 c->pix_abs[0][3] = bfin_pix_abs16_xy2;
152 c->pix_abs[1][0] = bfin_pix_abs8;
153 c->pix_abs[1][1] = bfin_pix_abs8_x2;
154 c->pix_abs[1][2] = bfin_pix_abs8_y2;
155 c->pix_abs[1][3] = bfin_pix_abs8_xy2;
158 c->sse[0] = ff_bfin_sse16;
159 c->sse[1] = ff_bfin_sse8;
160 c->sse[2] = ff_bfin_sse4;
162 if (avctx->bits_per_raw_sample <= 8) {
163 if (avctx->dct_algo == FF_DCT_AUTO)
164 c->fdct = ff_bfin_fdct;
166 if (avctx->idct_algo == FF_IDCT_AUTO) {
167 c->idct_permutation_type = FF_NO_IDCT_PERM;
168 c->idct = ff_bfin_idct;
169 c->idct_add = bfin_idct_add;
170 c->idct_put = bfin_idct_put;