dsputil.h

播放H264文件的播放器

static _inline void emms(void)
{
    __asm__ volatile ("emms;":::"memory");
}


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

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

#elif ARCH_ARM

extern int mm_flags;

#if HAVE_NEON
#   define DECLARE_ALIGNED_8(t, v) DECLARE_ALIGNED(16, t, v)
#   define STRIDE_ALIGN 16
#endif

#elif ARCH_PPC

extern int mm_flags;

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

#elif 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 */
    FFTComplex *tmp_buf;
    void (*fft_permute)(struct FFTContext *s, FFTComplex *z);
    void (*fft_calc)(struct FFTContext *s, FFTComplex *z);
    void (*imdct_calc)(struct MDCTContext *s, FFTSample *output, const FFTSample *input);
    void (*imdct_half)(struct MDCTContext *s, FFTSample *output, const FFTSample *input);
} FFTContext;

extern FFTSample* ff_cos_tabs[13];

/**
 * Sets up a complex FFT.
 * @param nbits           log2 of the length of the input array
 * @param inverse         if 0 perform the forward transform, if 1 perform the inverse
 */
int ff_fft_init(FFTContext *s, int nbits, int inverse);
void ff_fft_permute_c(FFTContext *s, FFTComplex *z);
void ff_fft_permute_sse(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);

/**
 * Do the permutation needed BEFORE calling ff_fft_calc().
 */
static _inline void ff_fft_permute(FFTContext *s, FFTComplex *z)
{
    s->fft_permute(s, z);
}
/**
 * Do a complex FFT with the parameters defined in ff_fft_init(). The
 * input data must be permuted before. No 1.0/sqrt(n) normalization is done.
 */
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;

static _inline void ff_imdct_calc(MDCTContext *s, FFTSample *output, const FFTSample *input)
{
    s->fft.imdct_calc(s, output, input);
}
static _inline void ff_imdct_half(MDCTContext *s, FFTSample *output, const FFTSample *input)
{
    s->fft.imdct_half(s, output, input);
}

/**
 * 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);
extern float ff_sine_128 [ 128];
extern float ff_sine_256 [ 256];
extern float ff_sine_512 [ 512];
extern float ff_sine_1024[1024];
extern float ff_sine_2048[2048];
extern float ff_sine_4096[4096];
extern float *ff_sine_windows[6];

int ff_mdct_init(MDCTContext *s, int nbits, int inverse);
void ff_imdct_calc_c(MDCTContext *s, FFTSample *output, const FFTSample *input);
void ff_imdct_half_c(MDCTContext *s, FFTSample *output, const FFTSample *input);
void ff_imdct_calc_3dn(MDCTContext *s, FFTSample *output, const FFTSample *input);
void ff_imdct_half_3dn(MDCTContext *s, FFTSample *output, const FFTSample *input);
void ff_imdct_calc_3dn2(MDCTContext *s, FFTSample *output, const FFTSample *input);
void ff_imdct_half_3dn2(MDCTContext *s, FFTSample *output, const FFTSample *input);
void ff_imdct_calc_sse(MDCTContext *s, FFTSample *output, const FFTSample *input);
void ff_imdct_half_sse(MDCTContext *s, FFTSample *output, const FFTSample *input);
void ff_mdct_calc(MDCTContext *s, FFTSample *out, const FFTSample *input);
void ff_mdct_end(MDCTContext *s);

/* Real Discrete Fourier Transform */

enum RDFTransformType {
    RDFT,
    IRDFT,
    RIDFT,
    IRIDFT,
};

typedef struct {
    int nbits;
    int inverse;
    int sign_convention;

    /* pre/post rotation tables */
    FFTSample *tcos;
    FFTSample *tsin;
    FFTContext fft;
} RDFTContext;

/**
 * Sets up a real FFT.
 * @param nbits           log2 of the length of the input array
 * @param trans           the type of transform
 */
int ff_rdft_init(RDFTContext *s, int nbits, enum RDFTransformType trans);
void ff_rdft_calc(RDFTContext *s, FFTSample *data);
void ff_rdft_end(RDFTContext *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   ));
		memcpy(dst, src, sizeof(uint16_t));
        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   ));
		memcpy(dst, src, sizeof(uint32_t));
        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 ));
		memcpy(dst, src, 2*sizeof(uint32_t));
        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 ));
		memcpy(dst, src, 2*sizeof(uint32_t));
        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));
		memcpy(dst, src, 4*sizeof(uint32_t));
        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));
		memcpy(dst, src, 4*sizeof(uint32_t));
        dst[16]= src[16];
        dst+=dstStride;
        src+=srcStride;
    }
}

#endif /* AVCODEC_DSPUTIL_H */