3 * Copyright (c) 2000, 2001, 2002 Fabrice Bellard.
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.
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.
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
27 typedef short DCTELEM;
29 void fdct_ifast (DCTELEM *data);
30 void ff_jpeg_fdct_islow (DCTELEM *data);
32 void j_rev_dct (DCTELEM *data);
34 void ff_fdct_mmx(DCTELEM *block);
37 extern const UINT8 ff_alternate_horizontal_scan[64];
38 extern const UINT8 ff_alternate_vertical_scan[64];
39 extern const UINT8 ff_zigzag_direct[64];
41 /* pixel operations */
42 #define MAX_NEG_CROP 384
45 extern UINT32 squareTbl[512];
46 extern UINT8 cropTbl[256 + 2 * MAX_NEG_CROP];
48 void dsputil_init(void);
50 /* minimum alignment rules ;)
51 if u notice errors in the align stuff, need more alignment for some asm code for some cpu
52 or need to use a function with less aligned data then send a mail to the ffmpeg-dev list, ...
54 !warning these alignments might not match reallity, (missing attribute((align)) stuff somewhere possible)
55 i (michael) didnt check them, these are just the alignents which i think could be reached easily ...
57 !future video codecs might need functions with less strict alignment
60 /* pixel ops : interface with DCT */
61 extern void (*get_pixels)(DCTELEM *block/*align 16*/, const UINT8 *pixels/*align 8*/, int line_size);
62 extern void (*diff_pixels)(DCTELEM *block/*align 16*/, const UINT8 *s1/*align 8*/, const UINT8 *s2/*align 8*/, int stride);
63 extern void (*put_pixels_clamped)(const DCTELEM *block/*align 16*/, UINT8 *pixels/*align 8*/, int line_size);
64 extern void (*add_pixels_clamped)(const DCTELEM *block/*align 16*/, UINT8 *pixels/*align 8*/, int line_size);
65 extern void (*gmc1)(UINT8 *dst/*align 8*/, UINT8 *src/*align 1*/, int srcStride, int h, int x16, int y16, int rounder);
66 extern void (*clear_blocks)(DCTELEM *blocks/*align 16*/);
67 extern int (*pix_sum)(UINT8 * pix, int line_size);
68 extern int (*pix_norm1)(UINT8 * pix, int line_size);
72 void get_pixels_c(DCTELEM *block, const UINT8 *pixels, int line_size);
73 void diff_pixels_c(DCTELEM *block, const UINT8 *s1, const UINT8 *s2, int stride);
74 void put_pixels_clamped_c(const DCTELEM *block, UINT8 *pixels, int line_size);
75 void add_pixels_clamped_c(const DCTELEM *block, UINT8 *pixels, int line_size);
76 void clear_blocks_c(DCTELEM *blocks);
78 /* add and put pixel (decoding) */
79 // blocksizes for op_pixels_func are 8x4,8x8 16x8 16x16
80 typedef void (*op_pixels_func)(UINT8 *block/*align width (8 or 16)*/, const UINT8 *pixels/*align 1*/, int line_size, int h);
81 typedef void (*qpel_mc_func)(UINT8 *dst/*align width (8 or 16)*/, UINT8 *src/*align 1*/, int stride);
83 extern op_pixels_func put_pixels_tab[2][4];
84 extern op_pixels_func avg_pixels_tab[2][4];
85 extern op_pixels_func put_no_rnd_pixels_tab[2][4];
86 extern op_pixels_func avg_no_rnd_pixels_tab[2][4];
87 extern qpel_mc_func put_qpel_pixels_tab[2][16];
88 extern qpel_mc_func avg_qpel_pixels_tab[2][16];
89 extern qpel_mc_func put_no_rnd_qpel_pixels_tab[2][16];
90 extern qpel_mc_func avg_no_rnd_qpel_pixels_tab[2][16];
92 #define CALL_2X_PIXELS(a, b, n)\
93 static void a(uint8_t *block, const uint8_t *pixels, int line_size, int h){\
94 b(block , pixels , line_size, h);\
95 b(block+n, pixels+n, line_size, h);\
98 /* motion estimation */
100 typedef int (*op_pixels_abs_func)(UINT8 *blk1/*align width (8 or 16)*/, UINT8 *blk2/*align 1*/, int line_size);
102 extern op_pixels_abs_func pix_abs16x16;
103 extern op_pixels_abs_func pix_abs16x16_x2;
104 extern op_pixels_abs_func pix_abs16x16_y2;
105 extern op_pixels_abs_func pix_abs16x16_xy2;
106 extern op_pixels_abs_func pix_abs8x8;
107 extern op_pixels_abs_func pix_abs8x8_x2;
108 extern op_pixels_abs_func pix_abs8x8_y2;
109 extern op_pixels_abs_func pix_abs8x8_xy2;
111 int pix_abs16x16_c(UINT8 *blk1, UINT8 *blk2, int lx);
112 int pix_abs16x16_x2_c(UINT8 *blk1, UINT8 *blk2, int lx);
113 int pix_abs16x16_y2_c(UINT8 *blk1, UINT8 *blk2, int lx);
114 int pix_abs16x16_xy2_c(UINT8 *blk1, UINT8 *blk2, int lx);
116 void block_permute(INT16 *block, UINT8 *permutation);
118 #if defined(HAVE_MMX)
120 #define MM_MMX 0x0001 /* standard MMX */
121 #define MM_3DNOW 0x0004 /* AMD 3DNOW */
122 #define MM_MMXEXT 0x0002 /* SSE integer functions or AMD MMX ext */
123 #define MM_SSE 0x0008 /* SSE functions */
124 #define MM_SSE2 0x0010 /* PIV SSE2 functions */
128 int mm_support(void);
130 static inline void emms(void)
132 __asm __volatile ("emms;":::"memory");
137 if (mm_flags & MM_MMX)\
141 #define __align8 __attribute__ ((aligned (8)))
143 void dsputil_init_mmx(void);
144 void dsputil_set_bit_exact_mmx(void);
146 #elif defined(ARCH_ARMV4L)
150 /* This is to use 4 bytes read to the IDCT pointers for some 'zero'
152 #define __align8 __attribute__ ((aligned (4)))
154 void dsputil_init_armv4l(void);
156 #elif defined(HAVE_MLIB)
160 /* SPARC/VIS IDCT needs 8-byte aligned DCT blocks */
161 #define __align8 __attribute__ ((aligned (8)))
163 void dsputil_init_mlib(void);
165 #elif defined(ARCH_ALPHA)
168 #define __align8 __attribute__ ((aligned (8)))
170 void dsputil_init_alpha(void);
172 #elif defined(ARCH_POWERPC)
175 #define __align8 __attribute__ ((aligned (16)))
177 void dsputil_init_ppc(void);
179 #elif defined(HAVE_MMI)
183 #define __align8 __attribute__ ((aligned (16)))
185 void dsputil_init_mmi(void);
197 struct unaligned_64 { uint64_t l; } __attribute__((packed));
198 struct unaligned_32 { uint32_t l; } __attribute__((packed));
200 #define LD32(a) (((const struct unaligned_32 *) (a))->l)
201 #define LD64(a) (((const struct unaligned_64 *) (a))->l)
203 #define ST32(a, b) (((struct unaligned_32 *) (a))->l) = (b)
207 #define LD32(a) (*((uint32_t*)(a)))
208 #define LD64(a) (*((uint64_t*)(a)))
210 #define ST32(a, b) *((uint32_t*)(a)) = (b)
212 #endif /* !__GNUC__ */
215 void get_psnr(UINT8 *orig_image[3], UINT8 *coded_image[3],
216 int orig_linesize[3], int coded_linesize,
217 AVCodecContext *avctx);