X-Git-Url: https://git.sesse.net/?a=blobdiff_plain;f=libavcodec%2Fdsputil.h;h=8fe9be77984be4dde66315c3ea7f7d71a6ee12be;hb=059715a41cf67834c1ce153caa245c394abb82e0;hp=f8711d296032305e80af6c1676d438607c4cc488;hpb=10acc47995831802501d0dbb9aab4ba7f91cbcfb;p=ffmpeg diff --git a/libavcodec/dsputil.h b/libavcodec/dsputil.h index f8711d29603..8fe9be77984 100644 --- a/libavcodec/dsputil.h +++ b/libavcodec/dsputil.h @@ -1,6 +1,7 @@ /* * DSP utils * Copyright (c) 2000, 2001, 2002 Fabrice Bellard. + * Copyright (c) 2002-2004 Michael Niedermayer * * This library is free software; you can redistribute it and/or * modify it under the terms of the GNU Lesser General Public @@ -14,7 +15,7 @@ * * You should have received a copy of the GNU Lesser General Public * License along with this library; if not, write to the Free Software - * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA */ /** @@ -34,15 +35,28 @@ //#define DEBUG /* dct code */ typedef short DCTELEM; +typedef int DWTELEM; void fdct_ifast (DCTELEM *data); +void fdct_ifast248 (DCTELEM *data); void ff_jpeg_fdct_islow (DCTELEM *data); void ff_fdct248_islow (DCTELEM *data); void j_rev_dct (DCTELEM *data); +void j_rev_dct4 (DCTELEM *data); +void j_rev_dct2 (DCTELEM *data); +void j_rev_dct1 (DCTELEM *data); void ff_fdct_mmx(DCTELEM *block); void ff_fdct_mmx2(DCTELEM *block); +void ff_fdct_sse2(DCTELEM *block); + +void ff_h264_idct8_add_c(uint8_t *dst, DCTELEM *block, int stride); +void ff_h264_idct_add_c(uint8_t *dst, DCTELEM *block, int stride); +void ff_h264_idct8_dc_add_c(uint8_t *dst, DCTELEM *block, int stride); +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); /* encoding scans */ extern const uint8_t ff_alternate_horizontal_scan[64]; @@ -51,12 +65,16 @@ extern const uint8_t ff_zigzag_direct[64]; extern const uint8_t ff_zigzag248_direct[64]; /* pixel operations */ -#define MAX_NEG_CROP 384 +#define MAX_NEG_CROP 1024 /* temporary */ extern uint32_t squareTbl[512]; extern uint8_t cropTbl[256 + 2 * MAX_NEG_CROP]; +/* VP3 DSP functions */ +void ff_vp3_idct_c(DCTELEM *block/* align 16*/); +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*/); /* minimum alignment rules ;) if u notice errors in the align stuff, need more alignment for some asm code for some cpu @@ -78,10 +96,13 @@ void clear_blocks_c(DCTELEM *blocks); /* add and put pixel (decoding) */ // blocksizes for op_pixels_func are 8x4,8x8 16x8 16x16 +//h for op_pixels_func is limited to {width/2, width} but never larger than 16 and never smaller then 4 typedef void (*op_pixels_func)(uint8_t *block/*align width (8 or 16)*/, const uint8_t *pixels/*align 1*/, int line_size, int h); typedef void (*tpel_mc_func)(uint8_t *block/*align width (8 or 16)*/, const uint8_t *pixels/*align 1*/, int line_size, int w, int h); typedef void (*qpel_mc_func)(uint8_t *dst/*align width (8 or 16)*/, uint8_t *src/*align 1*/, int stride); typedef void (*h264_chroma_mc_func)(uint8_t *dst/*align 8*/, uint8_t *src/*align 1*/, int srcStride, int h, int x, int y); +typedef void (*h264_weight_func)(uint8_t *block, int stride, int log2_denom, int weight, int offset); +typedef void (*h264_biweight_func)(uint8_t *dst, uint8_t *src, int stride, int log2_denom, int weightd, int weights, int offset); #define DEF_OLD_QPEL(name)\ void ff_put_ ## name (uint8_t *dst/*align width (8 or 16)*/, uint8_t *src/*align 1*/, int stride);\ @@ -108,11 +129,13 @@ static void a(uint8_t *block, const uint8_t *pixels, int line_size, int h){\ } /* motion estimation */ +// h is limited to {width/2, width, 2*width} but never larger than 16 and never smaller then 2 +// allthough currently h<4 is not used as functions with width <8 are not used and neither implemented +typedef int (*me_cmp_func)(void /*MpegEncContext*/ *s, uint8_t *blk1/*align width (8 or 16)*/, uint8_t *blk2/*align 1*/, int line_size, int h)/* __attribute__ ((const))*/; -typedef int (*op_pixels_abs_func)(uint8_t *blk1/*align width (8 or 16)*/, uint8_t *blk2/*align 1*/, int line_size)/* __attribute__ ((const))*/; - -typedef int (*me_cmp_func)(void /*MpegEncContext*/ *s, uint8_t *blk1/*align width (8 or 16)*/, uint8_t *blk2/*align 1*/, int line_size)/* __attribute__ ((const))*/; +// for snow slices +typedef struct slice_buffer_s slice_buffer; /** * DSPContext. @@ -122,7 +145,10 @@ typedef struct DSPContext { void (*get_pixels)(DCTELEM *block/*align 16*/, const uint8_t *pixels/*align 8*/, int line_size); void (*diff_pixels)(DCTELEM *block/*align 16*/, const uint8_t *s1/*align 8*/, const uint8_t *s2/*align 8*/, int stride); void (*put_pixels_clamped)(const DCTELEM *block/*align 16*/, uint8_t *pixels/*align 8*/, int line_size); + void (*put_signed_pixels_clamped)(const DCTELEM *block/*align 16*/, uint8_t *pixels/*align 8*/, int line_size); void (*add_pixels_clamped)(const DCTELEM *block/*align 16*/, uint8_t *pixels/*align 8*/, int line_size); + void (*add_pixels8)(uint8_t *pixels, DCTELEM *block, int line_size); + void (*add_pixels4)(uint8_t *pixels, DCTELEM *block, int line_size); /** * translational global motion compensation. */ @@ -131,29 +157,38 @@ typedef struct DSPContext { * global motion compensation. */ 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); + int dxx, int dxy, int dyx, int dyy, int shift, int r, int width, int height); 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); - me_cmp_func sad[2]; /* identical to pix_absAxA except additional void * */ - me_cmp_func sse[2]; - me_cmp_func hadamard8_diff[2]; - me_cmp_func dct_sad[2]; - me_cmp_func quant_psnr[2]; - me_cmp_func bit[2]; - me_cmp_func rd[2]; - int (*hadamard8_abs )(uint8_t *src, int stride, int mean); - - me_cmp_func me_pre_cmp[11]; - me_cmp_func me_cmp[11]; - me_cmp_func me_sub_cmp[11]; - me_cmp_func mb_cmp[11]; - - /* maybe create an array for 16/8/4/2 functions */ +// 16x16 8x8 4x4 2x2 16x8 8x4 4x2 8x16 4x8 2x4 + + me_cmp_func sad[5]; /* identical to pix_absAxA except additional void * */ + me_cmp_func sse[5]; + me_cmp_func hadamard8_diff[5]; + me_cmp_func dct_sad[5]; + me_cmp_func quant_psnr[5]; + me_cmp_func bit[5]; + me_cmp_func rd[5]; + me_cmp_func vsad[5]; + me_cmp_func vsse[5]; + me_cmp_func nsse[5]; + me_cmp_func w53[5]; + me_cmp_func w97[5]; + me_cmp_func dct_max[5]; + me_cmp_func dct264_sad[5]; + + me_cmp_func me_pre_cmp[5]; + me_cmp_func me_cmp[5]; + me_cmp_func me_sub_cmp[5]; + me_cmp_func mb_cmp[5]; + me_cmp_func ildct_cmp[5]; //only width 16 used + me_cmp_func frame_skip_cmp[5]; //only width 8 used + /** * Halfpel motion compensation with rounding (a+b+1)>>1. - * this is an array[4][4] of motion compensation funcions for 4 - * horizontal blocksizes (2,4,8,16) and the 4 halfpel positions
+ * this is an array[4][4] of motion compensation funcions for 4 + * horizontal blocksizes (8,16) and the 4 halfpel positions
* *pixels_tab[ 0->16xH 1->8xH ][ xhalfpel + 2*yhalfpel ] * @param block destination where the result is stored * @param pixels source @@ -164,8 +199,8 @@ typedef struct DSPContext { /** * Halfpel motion compensation with rounding (a+b+1)>>1. - * This is an array[4][4] of motion compensation functions for 4 - * horizontal blocksizes (2,4,8,16) and the 4 halfpel positions
+ * This is an array[4][4] of motion compensation functions for 4 + * horizontal blocksizes (8,16) and the 4 halfpel positions
* *pixels_tab[ 0->16xH 1->8xH ][ xhalfpel + 2*yhalfpel ] * @param block destination into which the result is averaged (a+b+1)>>1 * @param pixels source @@ -176,7 +211,7 @@ typedef struct DSPContext { /** * Halfpel motion compensation with no rounding (a+b)>>1. - * this is an array[2][4] of motion compensation funcions for 2 + * this is an array[2][4] of motion compensation funcions for 2 * horizontal blocksizes (8,16) and the 4 halfpel positions
* *pixels_tab[ 0->16xH 1->8xH ][ xhalfpel + 2*yhalfpel ] * @param block destination where the result is stored @@ -184,11 +219,11 @@ typedef struct DSPContext { * @param line_size number of bytes in a horizontal line of block * @param h height */ - op_pixels_func put_no_rnd_pixels_tab[2][4]; + op_pixels_func put_no_rnd_pixels_tab[4][4]; /** * Halfpel motion compensation with no rounding (a+b)>>1. - * this is an array[2][4] of motion compensation funcions for 2 + * this is an array[2][4] of motion compensation funcions for 2 * horizontal blocksizes (8,16) and the 4 halfpel positions
* *pixels_tab[ 0->16xH 1->8xH ][ xhalfpel + 2*yhalfpel ] * @param block destination into which the result is averaged (a+b)>>1 @@ -196,8 +231,10 @@ typedef struct DSPContext { * @param line_size number of bytes in a horizontal line of block * @param h height */ - op_pixels_func avg_no_rnd_pixels_tab[2][4]; - + op_pixels_func avg_no_rnd_pixels_tab[4][4]; + + void (*put_no_rnd_pixels_l2[2])(uint8_t *block/*align width (8 or 16)*/, const uint8_t *a/*align 1*/, const uint8_t *b/*align 1*/, int line_size, int h); + /** * Thirdpel motion compensation with rounding (a+b+1)>>1. * this is an array[12] of motion compensation funcions for the 9 thirdpel positions
@@ -215,25 +252,21 @@ typedef struct DSPContext { qpel_mc_func put_no_rnd_qpel_pixels_tab[2][16]; qpel_mc_func avg_no_rnd_qpel_pixels_tab[2][16]; qpel_mc_func put_mspel_pixels_tab[8]; - + /** * h264 Chram MC */ h264_chroma_mc_func put_h264_chroma_pixels_tab[3]; h264_chroma_mc_func avg_h264_chroma_pixels_tab[3]; - qpel_mc_func put_h264_qpel_pixels_tab[3][16]; - qpel_mc_func avg_h264_qpel_pixels_tab[3][16]; - - op_pixels_abs_func pix_abs16x16; - op_pixels_abs_func pix_abs16x16_x2; - op_pixels_abs_func pix_abs16x16_y2; - op_pixels_abs_func pix_abs16x16_xy2; - op_pixels_abs_func pix_abs8x8; - op_pixels_abs_func pix_abs8x8_x2; - op_pixels_abs_func pix_abs8x8_y2; - op_pixels_abs_func pix_abs8x8_xy2; - + qpel_mc_func put_h264_qpel_pixels_tab[4][16]; + qpel_mc_func avg_h264_qpel_pixels_tab[4][16]; + + h264_weight_func weight_h264_pixels_tab[10]; + h264_biweight_func biweight_h264_pixels_tab[10]; + + me_cmp_func pix_abs[2][4]; + /* huffyuv specific */ void (*add_bytes)(uint8_t *dst/*align 16*/, uint8_t *src/*align 16*/, int w); void (*diff_bytes)(uint8_t *dst/*align 16*/, uint8_t *src1/*align 16*/, uint8_t *src2/*align 1*/,int w); @@ -243,27 +276,39 @@ typedef struct DSPContext { */ void (*sub_hfyu_median_prediction)(uint8_t *dst, uint8_t *src1, uint8_t *src2, int w, int *left, int *left_top); void (*bswap_buf)(uint32_t *dst, 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 (*h263_v_loop_filter)(uint8_t *src, int stride, int qscale); + void (*h263_h_loop_filter)(uint8_t *src, int stride, int qscale); + + void (*h261_loop_filter)(uint8_t *src, int stride); + /* (I)DCT */ void (*fdct)(DCTELEM *block/* align 16*/); void (*fdct248)(DCTELEM *block/* align 16*/); - + /* IDCT really*/ void (*idct)(DCTELEM *block/* align 16*/); - + /** * block -> idct -> clip to unsigned 8 bit -> dest. * (-1392, 0, 0, ...) -> idct -> (-174, -174, ...) -> put -> (0, 0, ...) * @param line_size size in bytes of a horizotal line of dest */ void (*idct_put)(uint8_t *dest/*align 8*/, int line_size, DCTELEM *block/*align 16*/); - + /** * block -> idct -> add dest -> clip to unsigned 8 bit -> dest. * @param line_size size in bytes of a horizotal line of dest */ void (*idct_add)(uint8_t *dest/*align 8*/, int line_size, DCTELEM *block/*align 16*/); - + /** * idct input permutation. * several optimized IDCTs need a permutated input (relative to the normal order of the reference @@ -282,7 +327,22 @@ typedef struct DSPContext { #define FF_LIBMPEG2_IDCT_PERM 2 #define FF_SIMPLE_IDCT_PERM 3 #define FF_TRANSPOSE_IDCT_PERM 4 - +#define FF_PARTTRANS_IDCT_PERM 5 + + int (*try_8x8basis)(int16_t rem[64], int16_t weight[64], int16_t basis[64], int scale); + void (*add_8x8basis)(int16_t rem[64], int16_t basis[64], int scale); +#define BASIS_SHIFT 16 +#define RECON_SHIFT 6 + + 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); + + /* snow wavelet */ + void (*vertical_compose97i)(DWTELEM *b0, DWTELEM *b1, DWTELEM *b2, DWTELEM *b3, DWTELEM *b4, DWTELEM *b5, int width); + void (*horizontal_compose97i)(DWTELEM *b, int width); + void (*inner_add_yblock)(uint8_t *obmc, const int obmc_stride, uint8_t * * block, int b_w, int b_h, int src_x, int src_y, int src_stride, slice_buffer * sb, int add, uint8_t * dst8); } DSPContext; void dsputil_static_init(void); @@ -294,7 +354,9 @@ void dsputil_init(DSPContext* p, AVCodecContext *avctx); */ void ff_block_permute(DCTELEM *block, uint8_t *permutation, const uint8_t *scantable, int last); -#define BYTE_VEC32(c) ((c)*0x01010101UL) +void ff_set_cmp(DSPContext* c, me_cmp_func *cmp, int type); + +#define BYTE_VEC32(c) ((c)*0x01010101UL) static inline uint32_t rnd_avg32(uint32_t a, uint32_t b) { @@ -306,6 +368,30 @@ static inline uint32_t no_rnd_avg32(uint32_t a, uint32_t b) return (a & b) + (((a ^ b) & ~BYTE_VEC32(0x01)) >> 1); } +static inline int get_penalty_factor(int lambda, int lambda2, int type){ + switch(type&0xFF){ + default: + case FF_CMP_SAD: + return lambda>>FF_LAMBDA_SHIFT; + case FF_CMP_DCT: + return (3*lambda)>>(FF_LAMBDA_SHIFT+1); + case FF_CMP_W53: + return (4*lambda)>>(FF_LAMBDA_SHIFT); + case FF_CMP_W97: + return (2*lambda)>>(FF_LAMBDA_SHIFT); + case FF_CMP_SATD: + case FF_CMP_DCT264: + return (2*lambda)>>FF_LAMBDA_SHIFT; + case FF_CMP_RD: + case FF_CMP_PSNR: + case FF_CMP_SSE: + case FF_CMP_NSSE: + return lambda2>>FF_LAMBDA_SHIFT; + case FF_CMP_BIT: + return 1; + } +} + /** * Empty mmx state. * this must be called between any dsp function and float/double code. @@ -317,6 +403,12 @@ static inline uint32_t no_rnd_avg32(uint32_t a, uint32_t b) one or more MultiMedia extension */ int mm_support(void); +#ifdef __GNUC__ + #define DECLARE_ALIGNED_16(t,v) t v __attribute__ ((aligned (16))) +#else + #define DECLARE_ALIGNED_16(t,v) __declspec(align(16)) t v +#endif + #if defined(HAVE_MMX) #undef emms_c @@ -326,11 +418,13 @@ int mm_support(void); #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 */ 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) { @@ -344,7 +438,13 @@ static inline void emms(void) emms();\ } -#define __align8 __attribute__ ((aligned (8))) +#ifdef __GNUC__ + #define DECLARE_ALIGNED_8(t,v) t v __attribute__ ((aligned (8))) +#else + #define DECLARE_ALIGNED_8(t,v) __declspec(align(8)) t v +#endif + +#define STRIDE_ALIGN 8 void dsputil_init_mmx(DSPContext* c, AVCodecContext *avctx); void dsputil_init_pix_mmx(DSPContext* c, AVCodecContext *avctx); @@ -352,21 +452,35 @@ void dsputil_init_pix_mmx(DSPContext* c, AVCodecContext *avctx); #elif defined(ARCH_ARMV4L) /* This is to use 4 bytes read to the IDCT pointers for some 'zero' - line ptimizations */ -#define __align8 __attribute__ ((aligned (4))) + line optimizations */ +#define DECLARE_ALIGNED_8(t,v) t v __attribute__ ((aligned (4))) +#define STRIDE_ALIGN 4 + +#define MM_IWMMXT 0x0100 /* XScale IWMMXT */ + +extern int mm_flags; void dsputil_init_armv4l(DSPContext* c, AVCodecContext *avctx); #elif defined(HAVE_MLIB) /* SPARC/VIS IDCT needs 8-byte aligned DCT blocks */ -#define __align8 __attribute__ ((aligned (8))) +#define DECLARE_ALIGNED_8(t,v) t v __attribute__ ((aligned (8))) +#define STRIDE_ALIGN 8 void dsputil_init_mlib(DSPContext* c, AVCodecContext *avctx); +#elif defined(ARCH_SPARC) + +/* SPARC/VIS IDCT needs 8-byte aligned DCT blocks */ +#define DECLARE_ALIGNED_8(t,v) t v __attribute__ ((aligned (8))) +#define STRIDE_ALIGN 8 +void dsputil_init_vis(DSPContext* c, AVCodecContext *avctx); + #elif defined(ARCH_ALPHA) -#define __align8 __attribute__ ((aligned (8))) +#define DECLARE_ALIGNED_8(t,v) t v __attribute__ ((aligned (8))) +#define STRIDE_ALIGN 8 void dsputil_init_alpha(DSPContext* c, AVCodecContext *avctx); @@ -377,28 +491,34 @@ void dsputil_init_alpha(DSPContext* c, AVCodecContext *avctx); extern int mm_flags; #if defined(HAVE_ALTIVEC) && !defined(CONFIG_DARWIN) +#define pixel altivec_pixel #include +#undef pixel #endif -#define __align8 __attribute__ ((aligned (16))) +#define DECLARE_ALIGNED_8(t,v) t v __attribute__ ((aligned (16))) +#define STRIDE_ALIGN 16 void dsputil_init_ppc(DSPContext* c, AVCodecContext *avctx); #elif defined(HAVE_MMI) -#define __align8 __attribute__ ((aligned (16))) +#define DECLARE_ALIGNED_8(t,v) t v __attribute__ ((aligned (16))) +#define STRIDE_ALIGN 16 void dsputil_init_mmi(DSPContext* c, AVCodecContext *avctx); #elif defined(ARCH_SH4) -#define __align8 __attribute__ ((aligned (8))) +#define DECLARE_ALIGNED_8(t,v) t v __attribute__ ((aligned (8))) +#define STRIDE_ALIGN 8 void dsputil_init_sh4(DSPContext* c, AVCodecContext *avctx); #else -#define __align8 +#define DECLARE_ALIGNED_8(t,v) t v __attribute__ ((aligned (8))) +#define STRIDE_ALIGN 8 #endif @@ -412,6 +532,7 @@ struct unaligned_16 { uint16_t l; } __attribute__((packed)); #define LD32(a) (((const struct unaligned_32 *) (a))->l) #define LD64(a) (((const struct unaligned_64 *) (a))->l) +#define ST16(a, b) (((struct unaligned_16 *) (a))->l) = (b) #define ST32(a, b) (((struct unaligned_32 *) (a))->l) = (b) #else /* __GNUC__ */ @@ -448,17 +569,19 @@ typedef struct FFTContext { void (*fft_calc)(struct FFTContext *s, FFTComplex *z); } FFTContext; -int fft_init(FFTContext *s, int nbits, int inverse); -void fft_permute(FFTContext *s, FFTComplex *z); -void fft_calc_c(FFTContext *s, FFTComplex *z); -void fft_calc_sse(FFTContext *s, FFTComplex *z); -void fft_calc_altivec(FFTContext *s, FFTComplex *z); +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 fft_calc(FFTContext *s, FFTComplex *z) +static inline void ff_fft_calc(FFTContext *s, FFTComplex *z) { s->fft_calc(s, z); } -void fft_end(FFTContext *s); +void ff_fft_end(FFTContext *s); /* MDCT computation */ @@ -478,33 +601,24 @@ void ff_mdct_calc(MDCTContext *s, FFTSample *out, const FFTSample *input, FFTSample *tmp); void ff_mdct_end(MDCTContext *s); -#define WARPER88_1616(name8, name16)\ -static int name16(void /*MpegEncContext*/ *s, uint8_t *dst, uint8_t *src, int stride){\ - return name8(s, dst , src , stride)\ - +name8(s, dst+8 , src+8 , stride)\ - +name8(s, dst +8*stride, src +8*stride, stride)\ - +name8(s, dst+8+8*stride, src+8+8*stride, stride);\ +#define WARPER8_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);\ } -#ifndef HAVE_LRINTF -/* XXX: add ISOC specific test to avoid specific BSD testing. */ -/* better than nothing implementation. */ -/* btw, rintf() is existing on fbsd too -- alex */ -static inline long int lrintf(float x) -{ -#ifdef CONFIG_WIN32 - /* XXX: incorrect, but make it compile */ - return (int)(x); -#else - return (int)(rint(x)); -#endif +#define WARPER8_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;\ } -#endif - -#if defined(CONFIG_OS2) || defined(CONFIG_SUNOS) -static inline float floorf(float f) { - return floor(f); -} -#endif #endif