dsputil.h

ffmpeg移植到symbian的全部源代码

   one or more MultiMedia extension */
int mm_support(void);

void dsputil_init_alpha(DSPContext* c, AVCodecContext *avctx);
void dsputil_init_armv4l(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);
void dsputil_init_mmx(DSPContext* c, AVCodecContext *avctx);
void dsputil_init_ppc(DSPContext* c, AVCodecContext *avctx);
void dsputil_init_sh4(DSPContext* c, AVCodecContext *avctx);
void dsputil_init_vis(DSPContext* c, AVCodecContext *avctx);

#define DECLARE_ALIGNED_16(t, v) DECLARE_ALIGNED(16, t, v)

#if defined(HAVE_MMX)

#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);
void put_pixels_clamped_mmx(const DCTELEM *block, uint8_t *pixels, int line_size);
void put_signed_pixels_clamped_mmx(const DCTELEM *block, uint8_t *pixels, int line_size);

static inline void emms(void)
{
    asm volatile ("emms;":::"memory");
}


#define emms_c() \
{\
    if (mm_flags & MM_MMX)\
        emms();\
}

void dsputil_init_pix_mmx(DSPContext* c, AVCodecContext *avctx);

#elif defined(ARCH_ARMV4L)

#define MM_IWMMXT    0x0100 /* XScale IWMMXT */

extern int mm_flags;

#elif defined(ARCH_POWERPC)

#define MM_ALTIVEC    0x0001 /* standard AltiVec */

extern int mm_flags;

#define DECLARE_ALIGNED_8(t, v) DECLARE_ALIGNED(16, t, v)
#define STRIDE_ALIGN 16

#elif defined(HAVE_MMI)

#define DECLARE_ALIGNED_8(t, v) DECLARE_ALIGNED(16, t, v)
#define STRIDE_ALIGN 16

#else

#define mm_flags 0
#define mm_support() 0

#endif

#ifndef DECLARE_ALIGNED_8
#   define DECLARE_ALIGNED_8(t, v) DECLARE_ALIGNED(8, t, v)
#endif

#ifndef STRIDE_ALIGN
#   define STRIDE_ALIGN 8
#endif

/* PSNR */
void get_psnr(uint8_t *orig_image[3], uint8_t *coded_image[3],
              int orig_linesize[3], int coded_linesize,
              AVCodecContext *avctx);

/* FFT computation */

/* NOTE: soon integer code will be added, so you must use the
   FFTSample type */
typedef float FFTSample;

struct MDCTContext;

typedef struct FFTComplex {
    FFTSample re, im;
} FFTComplex;

typedef struct FFTContext {
    int nbits;
    int inverse;
    uint16_t *revtab;
    FFTComplex *exptab;
    FFTComplex *exptab1; /* only used by SSE code */
    void (*fft_calc)(struct FFTContext *s, FFTComplex *z);
    void (*imdct_calc)(struct MDCTContext *s, FFTSample *output,
                       const FFTSample *input, FFTSample *tmp);
    void (*imdct_half)(struct MDCTContext *s, FFTSample *output,
                       const FFTSample *input, FFTSample *tmp);
} FFTContext;

int ff_fft_init(FFTContext *s, int nbits, int inverse);
void ff_fft_permute(FFTContext *s, FFTComplex *z);
void ff_fft_calc_c(FFTContext *s, FFTComplex *z);
void ff_fft_calc_sse(FFTContext *s, FFTComplex *z);
void ff_fft_calc_3dn(FFTContext *s, FFTComplex *z);
void ff_fft_calc_3dn2(FFTContext *s, FFTComplex *z);
void ff_fft_calc_altivec(FFTContext *s, FFTComplex *z);

static inline void ff_fft_calc(FFTContext *s, FFTComplex *z)
{
    s->fft_calc(s, z);
}
void ff_fft_end(FFTContext *s);

/* MDCT computation */

typedef struct MDCTContext {
    int n;  /* size of MDCT (i.e. number of input data * 2) */
    int nbits; /* n = 2^nbits */
    /* pre/post rotation tables */
    FFTSample *tcos;
    FFTSample *tsin;
    FFTContext fft;
} MDCTContext;

/**
 * Generate a Kaiser-Bessel Derived Window.
 * @param   window  pointer to half window
 * @param   alpha   determines window shape
 * @param   n       size of half window
 */
void ff_kbd_window_init(float *window, float alpha, int n);

/**
 * Generate a sine window.
 * @param   window  pointer to half window
 * @param   n       size of half window
 */
void ff_sine_window_init(float *window, int n);

int ff_mdct_init(MDCTContext *s, int nbits, int inverse);
void ff_imdct_calc(MDCTContext *s, FFTSample *output,
                const FFTSample *input, FFTSample *tmp);
void ff_imdct_half(MDCTContext *s, FFTSample *output,
                   const FFTSample *input, FFTSample *tmp);
void ff_imdct_calc_3dn2(MDCTContext *s, FFTSample *output,
                        const FFTSample *input, FFTSample *tmp);
void ff_imdct_half_3dn2(MDCTContext *s, FFTSample *output,
                        const FFTSample *input, FFTSample *tmp);
void ff_imdct_calc_sse(MDCTContext *s, FFTSample *output,
                       const FFTSample *input, FFTSample *tmp);
void ff_imdct_half_sse(MDCTContext *s, FFTSample *output,
                       const FFTSample *input, FFTSample *tmp);
void ff_mdct_calc(MDCTContext *s, FFTSample *out,
               const FFTSample *input, FFTSample *tmp);
void ff_mdct_end(MDCTContext *s);

#define WRAPPER8_16(name8, name16)\
static int name16(void /*MpegEncContext*/ *s, uint8_t *dst, uint8_t *src, int stride, int h){\
    return name8(s, dst           , src           , stride, h)\
          +name8(s, dst+8         , src+8         , stride, h);\
}

#define WRAPPER8_16_SQ(name8, name16)\
static int name16(void /*MpegEncContext*/ *s, uint8_t *dst, uint8_t *src, int stride, int h){\
    int score=0;\
    score +=name8(s, dst           , src           , stride, 8);\
    score +=name8(s, dst+8         , src+8         , stride, 8);\
    if(h==16){\
        dst += 8*stride;\
        src += 8*stride;\
        score +=name8(s, dst           , src           , stride, 8);\
        score +=name8(s, dst+8         , src+8         , stride, 8);\
    }\
    return score;\
}


static inline void copy_block2(uint8_t *dst, uint8_t *src, int dstStride, int srcStride, int h)
{
    int i;
    for(i=0; i<h; i++)
    {
        AV_WN16(dst   , AV_RN16(src   ));
        dst+=dstStride;
        src+=srcStride;
    }
}

static inline void copy_block4(uint8_t *dst, uint8_t *src, int dstStride, int srcStride, int h)
{
    int i;
    for(i=0; i<h; i++)
    {
        AV_WN32(dst   , AV_RN32(src   ));
        dst+=dstStride;
        src+=srcStride;
    }
}

static inline void copy_block8(uint8_t *dst, uint8_t *src, int dstStride, int srcStride, int h)
{
    int i;
    for(i=0; i<h; i++)
    {
        AV_WN32(dst   , AV_RN32(src   ));
        AV_WN32(dst+4 , AV_RN32(src+4 ));
        dst+=dstStride;
        src+=srcStride;
    }
}

static inline void copy_block9(uint8_t *dst, uint8_t *src, int dstStride, int srcStride, int h)
{
    int i;
    for(i=0; i<h; i++)
    {
        AV_WN32(dst   , AV_RN32(src   ));
        AV_WN32(dst+4 , AV_RN32(src+4 ));
        dst[8]= src[8];
        dst+=dstStride;
        src+=srcStride;
    }
}

static inline void copy_block16(uint8_t *dst, uint8_t *src, int dstStride, int srcStride, int h)
{
    int i;
    for(i=0; i<h; i++)
    {
        AV_WN32(dst   , AV_RN32(src   ));
        AV_WN32(dst+4 , AV_RN32(src+4 ));
        AV_WN32(dst+8 , AV_RN32(src+8 ));
        AV_WN32(dst+12, AV_RN32(src+12));
        dst+=dstStride;
        src+=srcStride;
    }
}

static inline void copy_block17(uint8_t *dst, uint8_t *src, int dstStride, int srcStride, int h)
{
    int i;
    for(i=0; i<h; i++)
    {
        AV_WN32(dst   , AV_RN32(src   ));
        AV_WN32(dst+4 , AV_RN32(src+4 ));
        AV_WN32(dst+8 , AV_RN32(src+8 ));
        AV_WN32(dst+12, AV_RN32(src+12));
        dst[16]= src[16];
        dst+=dstStride;
        src+=srcStride;
    }
}

#endif /* FFMPEG_DSPUTIL_H */