* absolutely necessary to call emms_c() between dsp & float/double code
*/
-#ifndef FFMPEG_DSPUTIL_H
-#define FFMPEG_DSPUTIL_H
+#ifndef AVCODEC_DSPUTIL_H
+#define AVCODEC_DSPUTIL_H
#include "avcodec.h"
void ff_h264_idct_dc_add_c(uint8_t *dst, DCTELEM *block, int stride);
void ff_h264_lowres_idct_add_c(uint8_t *dst, int stride, DCTELEM *block);
void ff_h264_lowres_idct_put_c(uint8_t *dst, int stride, DCTELEM *block);
+void ff_h264_idct_add16_c(uint8_t *dst, const int *blockoffset, DCTELEM *block, int stride, const uint8_t nnzc[6*8]);
+void ff_h264_idct_add16intra_c(uint8_t *dst, const int *blockoffset, DCTELEM *block, int stride, const uint8_t nnzc[6*8]);
+void ff_h264_idct8_add4_c(uint8_t *dst, const int *blockoffset, DCTELEM *block, int stride, const uint8_t nnzc[6*8]);
+void ff_h264_idct_add8_c(uint8_t **dest, const int *blockoffset, DCTELEM *block, int stride, const uint8_t nnzc[6*8]);
void ff_vector_fmul_add_add_c(float *dst, const float *src0, const float *src1,
const float *src2, int src3, int blocksize, int step);
void ff_vp3_idct_put_c(uint8_t *dest/*align 8*/, int line_size, DCTELEM *block/*align 16*/);
void ff_vp3_idct_add_c(uint8_t *dest/*align 8*/, int line_size, DCTELEM *block/*align 16*/);
+void ff_vp3_v_loop_filter_c(uint8_t *src, int stride, int *bounding_values);
+void ff_vp3_h_loop_filter_c(uint8_t *src, int stride, int *bounding_values);
+
/* 1/2^n downscaling functions from imgconvert.c */
void ff_img_copy_plane(uint8_t *dst, int dst_wrap, const uint8_t *src, int src_wrap, int width, int height);
void ff_shrink22(uint8_t *dst, int dst_wrap, const uint8_t *src, int src_wrap, int width, int height);
const uint8_t *scantable;
uint8_t permutated[64];
uint8_t raster_end[64];
-#ifdef ARCH_POWERPC
+#ifdef ARCH_PPC
/** Used by dct_quantize_altivec to find last-non-zero */
DECLARE_ALIGNED(16, uint8_t, inverse[64]);
#endif
*/
void (*gmc )(uint8_t *dst/*align 8*/, uint8_t *src/*align 1*/, int stride, int h, int ox, int oy,
int dxx, int dxy, int dyx, int dyy, int shift, int r, int width, int height);
+ void (*clear_block)(DCTELEM *block/*align 16*/);
void (*clear_blocks)(DCTELEM *blocks/*align 16*/);
int (*pix_sum)(uint8_t * pix, int line_size);
int (*pix_norm1)(uint8_t * pix, int line_size);
void (*h264_v_loop_filter_luma)(uint8_t *pix/*align 16*/, int stride, int alpha, int beta, int8_t *tc0);
void (*h264_h_loop_filter_luma)(uint8_t *pix/*align 4 */, int stride, int alpha, int beta, int8_t *tc0);
/* v/h_loop_filter_luma_intra: align 16 */
+ void (*h264_v_loop_filter_luma_intra)(uint8_t *pix, int stride, int alpha, int beta);
+ void (*h264_h_loop_filter_luma_intra)(uint8_t *pix, int stride, int alpha, int beta);
void (*h264_v_loop_filter_chroma)(uint8_t *pix/*align 8*/, int stride, int alpha, int beta, int8_t *tc0);
void (*h264_h_loop_filter_chroma)(uint8_t *pix/*align 4*/, int stride, int alpha, int beta, int8_t *tc0);
void (*h264_v_loop_filter_chroma_intra)(uint8_t *pix/*align 8*/, int stride, int alpha, int beta);
void (*x8_v_loop_filter)(uint8_t *src, int stride, int qscale);
void (*x8_h_loop_filter)(uint8_t *src, int stride, int qscale);
+ void (*vp3_v_loop_filter)(uint8_t *src, int stride, int *bounding_values);
+ void (*vp3_h_loop_filter)(uint8_t *src, int stride, int *bounding_values);
+
/* assume len is a multiple of 4, and arrays are 16-byte aligned */
void (*vorbis_inverse_coupling)(float *mag, float *ang, int blocksize);
void (*ac3_downmix)(float (*samples)[256], float (*matrix)[2], int out_ch, int in_ch, int len);
#define EDGE_WIDTH 16
/* h264 functions */
- void (*h264_idct_add)(uint8_t *dst, DCTELEM *block, int stride);
- void (*h264_idct8_add)(uint8_t *dst, DCTELEM *block, int stride);
- void (*h264_idct_dc_add)(uint8_t *dst, DCTELEM *block, int stride);
- void (*h264_idct8_dc_add)(uint8_t *dst, DCTELEM *block, int stride);
+ /* NOTE!!! if you implement any of h264_idct8_add, h264_idct8_add4 then you must implement all of them
+ NOTE!!! if you implement any of h264_idct_add, h264_idct_add16, h264_idct_add16intra, h264_idct_add8 then you must implement all of them
+ The reason for above, is that no 2 out of one list may use a different permutation.
+ */
+ void (*h264_idct_add)(uint8_t *dst/*align 4*/, DCTELEM *block/*align 16*/, int stride);
+ void (*h264_idct8_add)(uint8_t *dst/*align 8*/, DCTELEM *block/*align 16*/, int stride);
+ void (*h264_idct_dc_add)(uint8_t *dst/*align 4*/, DCTELEM *block/*align 16*/, int stride);
+ void (*h264_idct8_dc_add)(uint8_t *dst/*align 8*/, DCTELEM *block/*align 16*/, int stride);
void (*h264_dct)(DCTELEM block[4][4]);
+ void (*h264_idct_add16)(uint8_t *dst/*align 16*/, const int *blockoffset, DCTELEM *block/*align 16*/, int stride, const uint8_t nnzc[6*8]);
+ void (*h264_idct8_add4)(uint8_t *dst/*align 16*/, const int *blockoffset, DCTELEM *block/*align 16*/, int stride, const uint8_t nnzc[6*8]);
+ void (*h264_idct_add8)(uint8_t **dst/*align 16*/, const int *blockoffset, DCTELEM *block/*align 16*/, int stride, const uint8_t nnzc[6*8]);
+ void (*h264_idct_add16intra)(uint8_t *dst/*align 16*/, const int *blockoffset, DCTELEM *block/*align 16*/, int stride, const uint8_t nnzc[6*8]);
/* snow wavelet */
void (*vertical_compose97i)(IDWTELEM *b0, IDWTELEM *b1, IDWTELEM *b2, IDWTELEM *b3, IDWTELEM *b4, IDWTELEM *b5, int width);
* @param shift number of bits to discard from product
*/
int32_t (*scalarproduct_int16)(int16_t *v1, int16_t *v2/*align 16*/, int len, int shift);
+
+ /* rv30 functions */
+ qpel_mc_func put_rv30_tpel_pixels_tab[4][16];
+ qpel_mc_func avg_rv30_tpel_pixels_tab[4][16];
+
+ /* rv40 functions */
+ qpel_mc_func put_rv40_qpel_pixels_tab[4][16];
+ qpel_mc_func avg_rv40_qpel_pixels_tab[4][16];
+ h264_chroma_mc_func put_rv40_chroma_pixels_tab[3];
+ h264_chroma_mc_func avg_rv40_chroma_pixels_tab[3];
} DSPContext;
void dsputil_static_init(void);
int mm_support(void);
void dsputil_init_alpha(DSPContext* c, AVCodecContext *avctx);
-void dsputil_init_armv4l(DSPContext* c, AVCodecContext *avctx);
+void dsputil_init_arm(DSPContext* c, AVCodecContext *avctx);
void dsputil_init_bfin(DSPContext* c, AVCodecContext *avctx);
void dsputil_init_mlib(DSPContext* c, AVCodecContext *avctx);
void dsputil_init_mmi(DSPContext* c, AVCodecContext *avctx);
#undef emms_c
-#define MM_MMX 0x0001 /* standard MMX */
-#define MM_3DNOW 0x0004 /* AMD 3DNOW */
-#define MM_MMXEXT 0x0002 /* SSE integer functions or AMD MMX ext */
-#define MM_SSE 0x0008 /* SSE functions */
-#define MM_SSE2 0x0010 /* PIV SSE2 functions */
-#define MM_3DNOWEXT 0x0020 /* AMD 3DNowExt */
-#define MM_SSE3 0x0040 /* Prescott SSE3 functions */
-#define MM_SSSE3 0x0080 /* Conroe SSSE3 functions */
-
extern int mm_flags;
void add_pixels_clamped_mmx(const DCTELEM *block, uint8_t *pixels, int line_size);
static inline void emms(void)
{
- asm volatile ("emms;":::"memory");
+ __asm__ volatile ("emms;":::"memory");
}
#define emms_c() \
{\
- if (mm_flags & MM_MMX)\
+ if (mm_flags & FF_MM_MMX)\
emms();\
}
void dsputil_init_pix_mmx(DSPContext* c, AVCodecContext *avctx);
-#elif defined(ARCH_ARMV4L)
-
-#define MM_IWMMXT 0x0100 /* XScale IWMMXT */
+#elif defined(ARCH_ARM)
extern int mm_flags;
-#elif defined(ARCH_POWERPC)
+#ifdef HAVE_NEON
+# define DECLARE_ALIGNED_8(t, v) DECLARE_ALIGNED(16, t, v)
+# define STRIDE_ALIGN 16
+#endif
-#define MM_ALTIVEC 0x0001 /* standard AltiVec */
+#elif defined(ARCH_PPC)
extern int mm_flags;
}
}
-#endif /* FFMPEG_DSPUTIL_H */
+#endif /* AVCODEC_DSPUTIL_H */