* 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_vector_fmul_add_add_c(float *dst, const float *src0, const float *src1,
const float *src2, int src3, int blocksize, int step);
-void ff_float_to_int16_c(int16_t *dst, const float *src, int len);
+void ff_vector_fmul_window_c(float *dst, const float *src0, const float *src1,
+ const float *win, float add_bias, int len);
+void ff_float_to_int16_c(int16_t *dst, const float *src, long len);
+void ff_float_to_int16_interleave_c(int16_t *dst, const float **src, long len, int channels);
/* encoding scans */
extern const uint8_t ff_alternate_horizontal_scan[64];
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);
void (*add_png_paeth_prediction)(uint8_t *dst, uint8_t *src, uint8_t *top, int w, int bpp);
void (*bswap_buf)(uint32_t *dst, const uint32_t *src, int w);
- void (*h264_v_loop_filter_luma)(uint8_t *pix, int stride, int alpha, int beta, int8_t *tc0);
- void (*h264_h_loop_filter_luma)(uint8_t *pix, int stride, int alpha, int beta, int8_t *tc0);
- void (*h264_v_loop_filter_chroma)(uint8_t *pix, int stride, int alpha, int beta, int8_t *tc0);
- void (*h264_h_loop_filter_chroma)(uint8_t *pix, int stride, int alpha, int beta, int8_t *tc0);
- void (*h264_v_loop_filter_chroma_intra)(uint8_t *pix, int stride, int alpha, int beta);
- void (*h264_h_loop_filter_chroma_intra)(uint8_t *pix, int stride, int alpha, int beta);
+ 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_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 (*h264_h_loop_filter_chroma_intra)(uint8_t *pix/*align 8*/, int stride, int alpha, int beta);
// h264_loop_filter_strength: simd only. the C version is inlined in h264.c
void (*h264_loop_filter_strength)(int16_t bS[2][4][4], uint8_t nnz[40], int8_t ref[2][40], int16_t mv[2][40][2],
- int bidir, int edges, int step, int mask_mv0, int mask_mv1);
+ int bidir, int edges, int step, int mask_mv0, int mask_mv1, int field);
void (*h263_v_loop_filter)(uint8_t *src, int stride, int qscale);
void (*h263_h_loop_filter)(uint8_t *src, int stride, int qscale);
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);
/* no alignment needed */
void (*flac_compute_autocorr)(const int32_t *data, int len, int lag, double *autoc);
/* assume len is a multiple of 8, and arrays are 16-byte aligned */
void (*vector_fmul_reverse)(float *dst, const float *src0, const float *src1, int len);
/* assume len is a multiple of 8, and src arrays are 16-byte aligned */
void (*vector_fmul_add_add)(float *dst, const float *src0, const float *src1, const float *src2, int src3, int len, int step);
+ /* assume len is a multiple of 4, and arrays are 16-byte aligned */
+ void (*vector_fmul_window)(float *dst, const float *src0, const float *src1, const float *win, float add_bias, int len);
+ /* assume len is a multiple of 8, and arrays are 16-byte aligned */
+ void (*int32_to_float_fmul_scalar)(float *dst, const int *src, float mul, int len);
/* C version: convert floats from the range [384.0,386.0] to ints in [-32768,32767]
* simd versions: convert floats from [-32768.0,32767.0] without rescaling and arrays are 16byte aligned */
- void (*float_to_int16)(int16_t *dst, const float *src, int len);
+ void (*float_to_int16)(int16_t *dst, const float *src, long len);
+ void (*float_to_int16_interleave)(int16_t *dst, const float **src, long len, int channels);
/* (I)DCT */
void (*fdct)(DCTELEM *block/* align 16*/);
#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);
+ 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]);
/* snow wavelet */
/* ape functions */
/**
* Add contents of the second vector to the first one.
- * @param len length of vectors, should be multiple of 8
+ * @param len length of vectors, should be multiple of 16
*/
void (*add_int16)(int16_t *v1/*align 16*/, int16_t *v2, int len);
/**
* Add contents of the second vector to the first one.
- * @param len length of vectors, should be multiple of 8
+ * @param len length of vectors, should be multiple of 16
*/
void (*sub_int16)(int16_t *v1/*align 16*/, int16_t *v2, int len);
/**
* Calculate scalar product of two vectors.
- * @param len length of vectors, should be multiple of 8
+ * @param len length of vectors, should be multiple of 16
* @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);
#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();\
}
#elif defined(ARCH_ARMV4L)
-#define MM_IWMMXT 0x0100 /* XScale IWMMXT */
-
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_POWERPC)
extern int mm_flags;
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, FFTSample *tmp);
+ void (*imdct_calc)(struct MDCTContext *s, FFTSample *output, const FFTSample *input);
+ void (*imdct_half)(struct MDCTContext *s, FFTSample *output, const FFTSample *input);
} FFTContext;
int ff_fft_init(FFTContext *s, int nbits, int inverse);
-void ff_fft_permute(FFTContext *s, FFTComplex *z);
+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);
+static inline void ff_fft_permute(FFTContext *s, FFTComplex *z)
+{
+ s->fft_permute(s, z);
+}
static inline void ff_fft_calc(FFTContext *s, FFTComplex *z)
{
s->fft_calc(s, z);
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 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_windows[5];
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_calc_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_mdct_calc(MDCTContext *s, FFTSample *out,
- const FFTSample *input, FFTSample *tmp);
+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);
#define WRAPPER8_16(name8, name16)\
}
}
-#endif /* FFMPEG_DSPUTIL_H */
+#endif /* AVCODEC_DSPUTIL_H */
projects / ffmpeg / blobdiff