yading@10: /*
yading@10:  * simple math operations
yading@10:  * Copyright (c) 2001, 2002 Fabrice Bellard
yading@10:  * Copyright (c) 2006 Michael Niedermayer <michaelni@gmx.at> et al
yading@10:  *
yading@10:  * This file is part of FFmpeg.
yading@10:  *
yading@10:  * FFmpeg is free software; you can redistribute it and/or
yading@10:  * modify it under the terms of the GNU Lesser General Public
yading@10:  * License as published by the Free Software Foundation; either
yading@10:  * version 2.1 of the License, or (at your option) any later version.
yading@10:  *
yading@10:  * FFmpeg is distributed in the hope that it will be useful,
yading@10:  * but WITHOUT ANY WARRANTY; without even the implied warranty of
yading@10:  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
yading@10:  * Lesser General Public License for more details.
yading@10:  *
yading@10:  * You should have received a copy of the GNU Lesser General Public
yading@10:  * License along with FFmpeg; if not, write to the Free Software
yading@10:  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
yading@10:  */
yading@10: #ifndef AVCODEC_MATHOPS_H
yading@10: #define AVCODEC_MATHOPS_H
yading@10: 
yading@10: #include <stdint.h>
yading@10: 
yading@10: #include "libavutil/common.h"
yading@10: #include "config.h"
yading@10: 
yading@10: extern const uint32_t ff_inverse[257];
yading@10: extern const uint8_t  ff_reverse[256];
yading@10: extern const uint8_t ff_sqrt_tab[256];
yading@10: 
yading@10: #if   ARCH_ARM
yading@10: #   include "arm/mathops.h"
yading@10: #elif ARCH_AVR32
yading@10: #   include "avr32/mathops.h"
yading@10: #elif ARCH_BFIN
yading@10: #   include "bfin/mathops.h"
yading@10: #elif ARCH_MIPS
yading@10: #   include "mips/mathops.h"
yading@10: #elif ARCH_PPC
yading@10: #   include "ppc/mathops.h"
yading@10: #elif ARCH_X86
yading@10: #   include "x86/mathops.h"
yading@10: #endif
yading@10: 
yading@10: /* generic implementation */
yading@10: 
yading@10: #ifndef MUL64
yading@10: #   define MUL64(a,b) ((int64_t)(a) * (int64_t)(b))
yading@10: #endif
yading@10: 
yading@10: #ifndef MULL
yading@10: #   define MULL(a,b,s) (MUL64(a, b) >> (s))
yading@10: #endif
yading@10: 
yading@10: #ifndef MULH
yading@10: static av_always_inline int MULH(int a, int b){
yading@10:     return MUL64(a, b) >> 32;
yading@10: }
yading@10: #endif
yading@10: 
yading@10: #ifndef UMULH
yading@10: static av_always_inline unsigned UMULH(unsigned a, unsigned b){
yading@10:     return ((uint64_t)(a) * (uint64_t)(b))>>32;
yading@10: }
yading@10: #endif
yading@10: 
yading@10: #ifndef MAC64
yading@10: #   define MAC64(d, a, b) ((d) += MUL64(a, b))
yading@10: #endif
yading@10: 
yading@10: #ifndef MLS64
yading@10: #   define MLS64(d, a, b) ((d) -= MUL64(a, b))
yading@10: #endif
yading@10: 
yading@10: /* signed 16x16 -> 32 multiply add accumulate */
yading@10: #ifndef MAC16
yading@10: #   define MAC16(rt, ra, rb) rt += (ra) * (rb)
yading@10: #endif
yading@10: 
yading@10: /* signed 16x16 -> 32 multiply */
yading@10: #ifndef MUL16
yading@10: #   define MUL16(ra, rb) ((ra) * (rb))
yading@10: #endif
yading@10: 
yading@10: #ifndef MLS16
yading@10: #   define MLS16(rt, ra, rb) ((rt) -= (ra) * (rb))
yading@10: #endif
yading@10: 
yading@10: /* median of 3 */
yading@10: #ifndef mid_pred
yading@10: #define mid_pred mid_pred
yading@10: static inline av_const int mid_pred(int a, int b, int c)
yading@10: {
yading@10: #if 0
yading@10:     int t= (a-b)&((a-b)>>31);
yading@10:     a-=t;
yading@10:     b+=t;
yading@10:     b-= (b-c)&((b-c)>>31);
yading@10:     b+= (a-b)&((a-b)>>31);
yading@10: 
yading@10:     return b;
yading@10: #else
yading@10:     if(a>b){
yading@10:         if(c>b){
yading@10:             if(c>a) b=a;
yading@10:             else    b=c;
yading@10:         }
yading@10:     }else{
yading@10:         if(b>c){
yading@10:             if(c>a) b=c;
yading@10:             else    b=a;
yading@10:         }
yading@10:     }
yading@10:     return b;
yading@10: #endif
yading@10: }
yading@10: #endif
yading@10: 
yading@10: #ifndef sign_extend
yading@10: static inline av_const int sign_extend(int val, unsigned bits)
yading@10: {
yading@10:     unsigned shift = 8 * sizeof(int) - bits;
yading@10:     union { unsigned u; int s; } v = { (unsigned) val << shift };
yading@10:     return v.s >> shift;
yading@10: }
yading@10: #endif
yading@10: 
yading@10: #ifndef zero_extend
yading@10: static inline av_const unsigned zero_extend(unsigned val, unsigned bits)
yading@10: {
yading@10:     return (val << ((8 * sizeof(int)) - bits)) >> ((8 * sizeof(int)) - bits);
yading@10: }
yading@10: #endif
yading@10: 
yading@10: #ifndef COPY3_IF_LT
yading@10: #define COPY3_IF_LT(x, y, a, b, c, d)\
yading@10: if ((y) < (x)) {\
yading@10:     (x) = (y);\
yading@10:     (a) = (b);\
yading@10:     (c) = (d);\
yading@10: }
yading@10: #endif
yading@10: 
yading@10: #ifndef MASK_ABS
yading@10: #define MASK_ABS(mask, level) do {              \
yading@10:         mask  = level >> 31;                    \
yading@10:         level = (level ^ mask) - mask;          \
yading@10:     } while (0)
yading@10: #endif
yading@10: 
yading@10: #ifndef NEG_SSR32
yading@10: #   define NEG_SSR32(a,s) ((( int32_t)(a))>>(32-(s)))
yading@10: #endif
yading@10: 
yading@10: #ifndef NEG_USR32
yading@10: #   define NEG_USR32(a,s) (((uint32_t)(a))>>(32-(s)))
yading@10: #endif
yading@10: 
yading@10: #if HAVE_BIGENDIAN
yading@10: # ifndef PACK_2U8
yading@10: #   define PACK_2U8(a,b)     (((a) <<  8) | (b))
yading@10: # endif
yading@10: # ifndef PACK_4U8
yading@10: #   define PACK_4U8(a,b,c,d) (((a) << 24) | ((b) << 16) | ((c) << 8) | (d))
yading@10: # endif
yading@10: # ifndef PACK_2U16
yading@10: #   define PACK_2U16(a,b)    (((a) << 16) | (b))
yading@10: # endif
yading@10: #else
yading@10: # ifndef PACK_2U8
yading@10: #   define PACK_2U8(a,b)     (((b) <<  8) | (a))
yading@10: # endif
yading@10: # ifndef PACK_4U2
yading@10: #   define PACK_4U8(a,b,c,d) (((d) << 24) | ((c) << 16) | ((b) << 8) | (a))
yading@10: # endif
yading@10: # ifndef PACK_2U16
yading@10: #   define PACK_2U16(a,b)    (((b) << 16) | (a))
yading@10: # endif
yading@10: #endif
yading@10: 
yading@10: #ifndef PACK_2S8
yading@10: #   define PACK_2S8(a,b)     PACK_2U8((a)&255, (b)&255)
yading@10: #endif
yading@10: #ifndef PACK_4S8
yading@10: #   define PACK_4S8(a,b,c,d) PACK_4U8((a)&255, (b)&255, (c)&255, (d)&255)
yading@10: #endif
yading@10: #ifndef PACK_2S16
yading@10: #   define PACK_2S16(a,b)    PACK_2U16((a)&0xffff, (b)&0xffff)
yading@10: #endif
yading@10: 
yading@10: #ifndef FASTDIV
yading@10: #   define FASTDIV(a,b) ((uint32_t)((((uint64_t)a) * ff_inverse[b]) >> 32))
yading@10: #endif /* FASTDIV */
yading@10: 
yading@10: static inline av_const unsigned int ff_sqrt(unsigned int a)
yading@10: {
yading@10:     unsigned int b;
yading@10: 
yading@10:     if (a < 255) return (ff_sqrt_tab[a + 1] - 1) >> 4;
yading@10:     else if (a < (1 << 12)) b = ff_sqrt_tab[a >> 4] >> 2;
yading@10: #if !CONFIG_SMALL
yading@10:     else if (a < (1 << 14)) b = ff_sqrt_tab[a >> 6] >> 1;
yading@10:     else if (a < (1 << 16)) b = ff_sqrt_tab[a >> 8]   ;
yading@10: #endif
yading@10:     else {
yading@10:         int s = av_log2_16bit(a >> 16) >> 1;
yading@10:         unsigned int c = a >> (s + 2);
yading@10:         b = ff_sqrt_tab[c >> (s + 8)];
yading@10:         b = FASTDIV(c,b) + (b << s);
yading@10:     }
yading@10: 
yading@10:     return b - (a < b * b);
yading@10: }
yading@10: 
yading@10: #endif /* AVCODEC_MATHOPS_H */