123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670671672673674675676677678679680681682683684685686687688689690691692693694695696697698699700701702703704705706707708709710711712713714715716717718719720721722723724725726727728729730731732733734735736737738739740741742743744745746747748749750751752753754755756757758759760761762763764765766767768769770771772773774775776777778779780781782783784785786787788789790791792793794795796797798799800801802803804805806807808809810811812813814815816817818819820821822823824825826827828829830831832833834835836837838839840841842843844845846847848849850851852853854855856857858859860861862863864865866867868869870871872873874875876877878879880881882883884885886887888889890891892893894895896897898899900901902903904905906907908909910911912913914915916917918919920921922923924925926927928929930931932933934935936937938939940941942943944945946947948949950951952953954955956957958959960961962963964965966967968969970971972973974975976977978979980981982983984985986987988989990991992993994995996997998999100010011002100310041005100610071008100910101011101210131014101510161017101810191020102110221023102410251026102710281029103010311032103310341035103610371038103910401041104210431044104510461047104810491050105110521053105410551056105710581059106010611062106310641065106610671068106910701071107210731074107510761077107810791080108110821083108410851086108710881089109010911092109310941095109610971098109911001101110211031104110511061107110811091110111111121113111411151116111711181119112011211122112311241125112611271128112911301131113211331134113511361137113811391140114111421143114411451146114711481149115011511152115311541155115611571158115911601161116211631164116511661167116811691170117111721173117411751176117711781179118011811182118311841185118611871188118911901191119211931194119511961197119811991200120112021203120412051206120712081209121012111212121312141215121612171218121912201221122212231224122512261227122812291230123112321233123412351236123712381239124012411242124312441245124612471248124912501251125212531254125512561257125812591260126112621263126412651266126712681269127012711272127312741275127612771278127912801281128212831284128512861287128812891290129112921293129412951296129712981299130013011302130313041305130613071308130913101311131213131314131513161317131813191320132113221323132413251326132713281329133013311332133313341335133613371338133913401341134213431344134513461347134813491350135113521353135413551356135713581359136013611362136313641365136613671368136913701371137213731374137513761377137813791380138113821383138413851386138713881389139013911392139313941395139613971398139914001401140214031404140514061407140814091410141114121413141414151416141714181419142014211422142314241425142614271428142914301431143214331434143514361437143814391440144114421443144414451446144714481449145014511452145314541455145614571458145914601461146214631464146514661467146814691470147114721473147414751476147714781479148014811482148314841485148614871488148914901491149214931494149514961497149814991500150115021503150415051506150715081509151015111512151315141515151615171518151915201521152215231524152515261527152815291530153115321533153415351536153715381539154015411542154315441545154615471548154915501551155215531554155515561557155815591560156115621563156415651566156715681569157015711572157315741575157615771578157915801581158215831584158515861587158815891590159115921593159415951596159715981599160016011602160316041605160616071608160916101611161216131614161516161617161816191620162116221623162416251626162716281629163016311632163316341635163616371638163916401641164216431644164516461647164816491650165116521653165416551656165716581659166016611662166316641665166616671668166916701671167216731674167516761677167816791680168116821683168416851686168716881689169016911692169316941695169616971698169917001701170217031704170517061707170817091710171117121713171417151716171717181719172017211722172317241725172617271728172917301731173217331734173517361737173817391740174117421743174417451746174717481749175017511752175317541755175617571758175917601761176217631764176517661767176817691770177117721773177417751776177717781779178017811782178317841785178617871788178917901791179217931794179517961797179817991800180118021803180418051806180718081809181018111812181318141815181618171818181918201821182218231824182518261827182818291830 |
- /*****************************************************************************
- * pixel.c: ppc pixel metrics
- *****************************************************************************
- * Copyright (C) 2003-2018 x264 project
- *
- * Authors: Eric Petit <eric.petit@lapsus.org>
- * Guillaume Poirier <gpoirier@mplayerhq.hu>
- *
- * This program is free software; you can redistribute it and/or modify
- * it under the terms of the GNU General Public License as published by
- * the Free Software Foundation; either version 2 of the License, or
- * (at your option) any later version.
- *
- * This program is distributed in the hope that it will be useful,
- * but WITHOUT ANY WARRANTY; without even the implied warranty of
- * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
- * GNU General Public License for more details.
- *
- * You should have received a copy of the GNU General Public License
- * along with this program; if not, write to the Free Software
- * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02111, USA.
- *
- * This program is also available under a commercial proprietary license.
- * For more information, contact us at licensing@x264.com.
- *****************************************************************************/
- #include "common/common.h"
- #include "ppccommon.h"
- #include "pixel.h"
- #if !HIGH_BIT_DEPTH
- /***********************************************************************
- * SAD routines
- **********************************************************************/
- #define PIXEL_SAD_ALTIVEC( name, lx, ly, a, b ) \
- static int name( uint8_t *pix1, intptr_t i_pix1, \
- uint8_t *pix2, intptr_t i_pix2 ) \
- { \
- ALIGNED_16( int sum ); \
- \
- LOAD_ZERO; \
- vec_u8_t pix1v, pix2v; \
- vec_s32_t sumv = zero_s32v; \
- for( int y = 0; y < ly; y++ ) \
- { \
- pix1v = vec_vsx_ld( 0, pix1 ); \
- pix2v = vec_vsx_ld( 0, pix2 ); \
- sumv = (vec_s32_t) vec_sum4s( \
- vec_absd( pix1v, pix2v ), \
- (vec_u32_t) sumv ); \
- pix1 += i_pix1; \
- pix2 += i_pix2; \
- } \
- sumv = vec_sum##a( sumv, zero_s32v ); \
- sumv = vec_splat( sumv, b ); \
- vec_ste( sumv, 0, &sum ); \
- return sum; \
- }
- PIXEL_SAD_ALTIVEC( pixel_sad_16x16_altivec, 16, 16, s, 3 )
- PIXEL_SAD_ALTIVEC( pixel_sad_8x16_altivec, 8, 16, 2s, 1 )
- PIXEL_SAD_ALTIVEC( pixel_sad_16x8_altivec, 16, 8, s, 3 )
- PIXEL_SAD_ALTIVEC( pixel_sad_8x8_altivec, 8, 8, 2s, 1 )
- /***********************************************************************
- * SATD routines
- **********************************************************************/
- /***********************************************************************
- * VEC_HADAMAR
- ***********************************************************************
- * b[0] = a[0] + a[1] + a[2] + a[3]
- * b[1] = a[0] + a[1] - a[2] - a[3]
- * b[2] = a[0] - a[1] - a[2] + a[3]
- * b[3] = a[0] - a[1] + a[2] - a[3]
- **********************************************************************/
- #define VEC_HADAMAR(a0,a1,a2,a3,b0,b1,b2,b3) \
- b2 = vec_add( a0, a1 ); \
- b3 = vec_add( a2, a3 ); \
- a0 = vec_sub( a0, a1 ); \
- a2 = vec_sub( a2, a3 ); \
- b0 = vec_add( b2, b3 ); \
- b1 = vec_sub( b2, b3 ); \
- b2 = vec_sub( a0, a2 ); \
- b3 = vec_add( a0, a2 )
- /***********************************************************************
- * VEC_ABS
- ***********************************************************************
- * a: s16v
- *
- * a = abs(a)
- *
- * Call vec_sub()/vec_max() instead of vec_abs() because vec_abs()
- * actually also calls vec_splat(0), but we already have a null vector.
- **********************************************************************/
- #define VEC_ABS(a) \
- a = vec_max( a, vec_sub( zero_s16v, a ) );
- #define VEC_ABSOLUTE(a) (vec_u16_t)vec_max( a, vec_sub( zero_s16v, a ) )
- /***********************************************************************
- * VEC_ADD_ABS
- ***********************************************************************
- * a: s16v
- * b, c: s32v
- *
- * c[i] = abs(a[2*i]) + abs(a[2*i+1]) + [bi]
- **********************************************************************/
- #define VEC_ADD_ABS(a,b,c) \
- VEC_ABS( a ); \
- c = vec_sum4s( a, b )
- /***********************************************************************
- * SATD 4x4
- **********************************************************************/
- static int pixel_satd_4x4_altivec( uint8_t *pix1, intptr_t i_pix1,
- uint8_t *pix2, intptr_t i_pix2 )
- {
- ALIGNED_16( int i_satd );
- PREP_DIFF;
- vec_s16_t diff0v, diff1v, diff2v, diff3v;
- vec_s16_t temp0v, temp1v, temp2v, temp3v;
- vec_s32_t satdv;
- VEC_DIFF_H( pix1, i_pix1, pix2, i_pix2, 4, diff0v );
- VEC_DIFF_H( pix1, i_pix1, pix2, i_pix2, 4, diff1v );
- VEC_DIFF_H( pix1, i_pix1, pix2, i_pix2, 4, diff2v );
- VEC_DIFF_H( pix1, i_pix1, pix2, i_pix2, 4, diff3v );
- /* Hadamar H */
- VEC_HADAMAR( diff0v, diff1v, diff2v, diff3v,
- temp0v, temp1v, temp2v, temp3v );
- VEC_TRANSPOSE_4( temp0v, temp1v, temp2v, temp3v,
- diff0v, diff1v, diff2v, diff3v );
- /* Hadamar V */
- VEC_HADAMAR( diff0v, diff1v, diff2v, diff3v,
- temp0v, temp1v, temp2v, temp3v );
- VEC_ADD_ABS( temp0v, zero_s32v, satdv );
- VEC_ADD_ABS( temp1v, satdv, satdv );
- VEC_ADD_ABS( temp2v, satdv, satdv );
- VEC_ADD_ABS( temp3v, satdv, satdv );
- satdv = vec_sum2s( satdv, zero_s32v );
- satdv = vec_splat( satdv, 1 );
- vec_ste( satdv, 0, &i_satd );
- return i_satd >> 1;
- }
- /***********************************************************************
- * SATD 4x8
- **********************************************************************/
- static int pixel_satd_4x8_altivec( uint8_t *pix1, intptr_t i_pix1,
- uint8_t *pix2, intptr_t i_pix2 )
- {
- ALIGNED_16( int i_satd );
- PREP_DIFF;
- vec_s16_t diff0v, diff1v, diff2v, diff3v;
- vec_s16_t temp0v, temp1v, temp2v, temp3v;
- vec_s32_t satdv;
- VEC_DIFF_H( pix1, i_pix1, pix2, i_pix2, 4, diff0v );
- VEC_DIFF_H( pix1, i_pix1, pix2, i_pix2, 4, diff1v );
- VEC_DIFF_H( pix1, i_pix1, pix2, i_pix2, 4, diff2v );
- VEC_DIFF_H( pix1, i_pix1, pix2, i_pix2, 4, diff3v );
- VEC_HADAMAR( diff0v, diff1v, diff2v, diff3v,
- temp0v, temp1v, temp2v, temp3v );
- VEC_TRANSPOSE_4( temp0v, temp1v, temp2v, temp3v,
- diff0v, diff1v, diff2v, diff3v );
- VEC_HADAMAR( diff0v, diff1v, diff2v, diff3v,
- temp0v, temp1v, temp2v, temp3v );
- VEC_ADD_ABS( temp0v, zero_s32v, satdv );
- VEC_ADD_ABS( temp1v, satdv, satdv );
- VEC_ADD_ABS( temp2v, satdv, satdv );
- VEC_ADD_ABS( temp3v, satdv, satdv );
- VEC_DIFF_H( pix1, i_pix1, pix2, i_pix2, 4, diff0v );
- VEC_DIFF_H( pix1, i_pix1, pix2, i_pix2, 4, diff1v );
- VEC_DIFF_H( pix1, i_pix1, pix2, i_pix2, 4, diff2v );
- VEC_DIFF_H( pix1, i_pix1, pix2, i_pix2, 4, diff3v );
- VEC_HADAMAR( diff0v, diff1v, diff2v, diff3v,
- temp0v, temp1v, temp2v, temp3v );
- VEC_TRANSPOSE_4( temp0v, temp1v, temp2v, temp3v,
- diff0v, diff1v, diff2v, diff3v );
- VEC_HADAMAR( diff0v, diff1v, diff2v, diff3v,
- temp0v, temp1v, temp2v, temp3v );
- VEC_ADD_ABS( temp0v, satdv, satdv );
- VEC_ADD_ABS( temp1v, satdv, satdv );
- VEC_ADD_ABS( temp2v, satdv, satdv );
- VEC_ADD_ABS( temp3v, satdv, satdv );
- satdv = vec_sum2s( satdv, zero_s32v );
- satdv = vec_splat( satdv, 1 );
- vec_ste( satdv, 0, &i_satd );
- return i_satd >> 1;
- }
- /***********************************************************************
- * SATD 8x4
- **********************************************************************/
- static int pixel_satd_8x4_altivec( uint8_t *pix1, intptr_t i_pix1,
- uint8_t *pix2, intptr_t i_pix2 )
- {
- ALIGNED_16( int i_satd );
- PREP_DIFF;
- vec_s16_t diff0v, diff1v, diff2v, diff3v,
- diff4v, diff5v, diff6v, diff7v;
- vec_s16_t temp0v, temp1v, temp2v, temp3v,
- temp4v, temp5v, temp6v, temp7v;
- vec_s32_t satdv;
- VEC_DIFF_H( pix1, i_pix1, pix2, i_pix2, 8, diff0v );
- VEC_DIFF_H( pix1, i_pix1, pix2, i_pix2, 8, diff1v );
- VEC_DIFF_H( pix1, i_pix1, pix2, i_pix2, 8, diff2v );
- VEC_DIFF_H( pix1, i_pix1, pix2, i_pix2, 8, diff3v );
- VEC_HADAMAR( diff0v, diff1v, diff2v, diff3v,
- temp0v, temp1v, temp2v, temp3v );
- /* This causes warnings because temp4v...temp7v haven't be set,
- but we don't care */
- VEC_TRANSPOSE_8( temp0v, temp1v, temp2v, temp3v,
- temp4v, temp5v, temp6v, temp7v,
- diff0v, diff1v, diff2v, diff3v,
- diff4v, diff5v, diff6v, diff7v );
- VEC_HADAMAR( diff0v, diff1v, diff2v, diff3v,
- temp0v, temp1v, temp2v, temp3v );
- VEC_HADAMAR( diff4v, diff5v, diff6v, diff7v,
- temp4v, temp5v, temp6v, temp7v );
- VEC_ADD_ABS( temp0v, zero_s32v, satdv );
- VEC_ADD_ABS( temp1v, satdv, satdv );
- VEC_ADD_ABS( temp2v, satdv, satdv );
- VEC_ADD_ABS( temp3v, satdv, satdv );
- VEC_ADD_ABS( temp4v, satdv, satdv );
- VEC_ADD_ABS( temp5v, satdv, satdv );
- VEC_ADD_ABS( temp6v, satdv, satdv );
- VEC_ADD_ABS( temp7v, satdv, satdv );
- satdv = vec_sum2s( satdv, zero_s32v );
- satdv = vec_splat( satdv, 1 );
- vec_ste( satdv, 0, &i_satd );
- return i_satd >> 1;
- }
- /***********************************************************************
- * SATD 8x8
- **********************************************************************/
- static int pixel_satd_8x8_altivec( uint8_t *pix1, intptr_t i_pix1,
- uint8_t *pix2, intptr_t i_pix2 )
- {
- ALIGNED_16( int i_satd );
- PREP_DIFF;
- vec_s16_t diff0v, diff1v, diff2v, diff3v,
- diff4v, diff5v, diff6v, diff7v;
- vec_s16_t temp0v, temp1v, temp2v, temp3v,
- temp4v, temp5v, temp6v, temp7v;
- vec_s32_t satdv;
- VEC_DIFF_H( pix1, i_pix1, pix2, i_pix2, 8, diff0v );
- VEC_DIFF_H( pix1, i_pix1, pix2, i_pix2, 8, diff1v );
- VEC_DIFF_H( pix1, i_pix1, pix2, i_pix2, 8, diff2v );
- VEC_DIFF_H( pix1, i_pix1, pix2, i_pix2, 8, diff3v );
- VEC_DIFF_H( pix1, i_pix1, pix2, i_pix2, 8, diff4v );
- VEC_DIFF_H( pix1, i_pix1, pix2, i_pix2, 8, diff5v );
- VEC_DIFF_H( pix1, i_pix1, pix2, i_pix2, 8, diff6v );
- VEC_DIFF_H( pix1, i_pix1, pix2, i_pix2, 8, diff7v );
- VEC_HADAMAR( diff0v, diff1v, diff2v, diff3v,
- temp0v, temp1v, temp2v, temp3v );
- VEC_HADAMAR( diff4v, diff5v, diff6v, diff7v,
- temp4v, temp5v, temp6v, temp7v );
- VEC_TRANSPOSE_8( temp0v, temp1v, temp2v, temp3v,
- temp4v, temp5v, temp6v, temp7v,
- diff0v, diff1v, diff2v, diff3v,
- diff4v, diff5v, diff6v, diff7v );
- VEC_HADAMAR( diff0v, diff1v, diff2v, diff3v,
- temp0v, temp1v, temp2v, temp3v );
- VEC_HADAMAR( diff4v, diff5v, diff6v, diff7v,
- temp4v, temp5v, temp6v, temp7v );
- VEC_ADD_ABS( temp0v, zero_s32v, satdv );
- VEC_ADD_ABS( temp1v, satdv, satdv );
- VEC_ADD_ABS( temp2v, satdv, satdv );
- VEC_ADD_ABS( temp3v, satdv, satdv );
- VEC_ADD_ABS( temp4v, satdv, satdv );
- VEC_ADD_ABS( temp5v, satdv, satdv );
- VEC_ADD_ABS( temp6v, satdv, satdv );
- VEC_ADD_ABS( temp7v, satdv, satdv );
- satdv = vec_sums( satdv, zero_s32v );
- satdv = vec_splat( satdv, 3 );
- vec_ste( satdv, 0, &i_satd );
- return i_satd >> 1;
- }
- /***********************************************************************
- * SATD 8x16
- **********************************************************************/
- static int pixel_satd_8x16_altivec( uint8_t *pix1, intptr_t i_pix1,
- uint8_t *pix2, intptr_t i_pix2 )
- {
- ALIGNED_16( int i_satd );
- PREP_DIFF;
- vec_s16_t diff0v, diff1v, diff2v, diff3v,
- diff4v, diff5v, diff6v, diff7v;
- vec_s16_t temp0v, temp1v, temp2v, temp3v,
- temp4v, temp5v, temp6v, temp7v;
- vec_s32_t satdv;
- VEC_DIFF_H( pix1, i_pix1, pix2, i_pix2, 8, diff0v );
- VEC_DIFF_H( pix1, i_pix1, pix2, i_pix2, 8, diff1v );
- VEC_DIFF_H( pix1, i_pix1, pix2, i_pix2, 8, diff2v );
- VEC_DIFF_H( pix1, i_pix1, pix2, i_pix2, 8, diff3v );
- VEC_DIFF_H( pix1, i_pix1, pix2, i_pix2, 8, diff4v );
- VEC_DIFF_H( pix1, i_pix1, pix2, i_pix2, 8, diff5v );
- VEC_DIFF_H( pix1, i_pix1, pix2, i_pix2, 8, diff6v );
- VEC_DIFF_H( pix1, i_pix1, pix2, i_pix2, 8, diff7v );
- VEC_HADAMAR( diff0v, diff1v, diff2v, diff3v,
- temp0v, temp1v, temp2v, temp3v );
- VEC_HADAMAR( diff4v, diff5v, diff6v, diff7v,
- temp4v, temp5v, temp6v, temp7v );
- VEC_TRANSPOSE_8( temp0v, temp1v, temp2v, temp3v,
- temp4v, temp5v, temp6v, temp7v,
- diff0v, diff1v, diff2v, diff3v,
- diff4v, diff5v, diff6v, diff7v );
- VEC_HADAMAR( diff0v, diff1v, diff2v, diff3v,
- temp0v, temp1v, temp2v, temp3v );
- VEC_HADAMAR( diff4v, diff5v, diff6v, diff7v,
- temp4v, temp5v, temp6v, temp7v );
- VEC_ADD_ABS( temp0v, zero_s32v, satdv );
- VEC_ADD_ABS( temp1v, satdv, satdv );
- VEC_ADD_ABS( temp2v, satdv, satdv );
- VEC_ADD_ABS( temp3v, satdv, satdv );
- VEC_ADD_ABS( temp4v, satdv, satdv );
- VEC_ADD_ABS( temp5v, satdv, satdv );
- VEC_ADD_ABS( temp6v, satdv, satdv );
- VEC_ADD_ABS( temp7v, satdv, satdv );
- VEC_DIFF_H( pix1, i_pix1, pix2, i_pix2, 8, diff0v );
- VEC_DIFF_H( pix1, i_pix1, pix2, i_pix2, 8, diff1v );
- VEC_DIFF_H( pix1, i_pix1, pix2, i_pix2, 8, diff2v );
- VEC_DIFF_H( pix1, i_pix1, pix2, i_pix2, 8, diff3v );
- VEC_DIFF_H( pix1, i_pix1, pix2, i_pix2, 8, diff4v );
- VEC_DIFF_H( pix1, i_pix1, pix2, i_pix2, 8, diff5v );
- VEC_DIFF_H( pix1, i_pix1, pix2, i_pix2, 8, diff6v );
- VEC_DIFF_H( pix1, i_pix1, pix2, i_pix2, 8, diff7v );
- VEC_HADAMAR( diff0v, diff1v, diff2v, diff3v,
- temp0v, temp1v, temp2v, temp3v );
- VEC_HADAMAR( diff4v, diff5v, diff6v, diff7v,
- temp4v, temp5v, temp6v, temp7v );
- VEC_TRANSPOSE_8( temp0v, temp1v, temp2v, temp3v,
- temp4v, temp5v, temp6v, temp7v,
- diff0v, diff1v, diff2v, diff3v,
- diff4v, diff5v, diff6v, diff7v );
- VEC_HADAMAR( diff0v, diff1v, diff2v, diff3v,
- temp0v, temp1v, temp2v, temp3v );
- VEC_HADAMAR( diff4v, diff5v, diff6v, diff7v,
- temp4v, temp5v, temp6v, temp7v );
- VEC_ADD_ABS( temp0v, satdv, satdv );
- VEC_ADD_ABS( temp1v, satdv, satdv );
- VEC_ADD_ABS( temp2v, satdv, satdv );
- VEC_ADD_ABS( temp3v, satdv, satdv );
- VEC_ADD_ABS( temp4v, satdv, satdv );
- VEC_ADD_ABS( temp5v, satdv, satdv );
- VEC_ADD_ABS( temp6v, satdv, satdv );
- VEC_ADD_ABS( temp7v, satdv, satdv );
- satdv = vec_sums( satdv, zero_s32v );
- satdv = vec_splat( satdv, 3 );
- vec_ste( satdv, 0, &i_satd );
- return i_satd >> 1;
- }
- /***********************************************************************
- * SATD 16x8
- **********************************************************************/
- static int pixel_satd_16x8_altivec( uint8_t *pix1, intptr_t i_pix1,
- uint8_t *pix2, intptr_t i_pix2 )
- {
- ALIGNED_16( int i_satd );
- LOAD_ZERO;
- vec_s32_t satdv;
- vec_s16_t pix1v, pix2v;
- vec_s16_t diffh0v, diffh1v, diffh2v, diffh3v,
- diffh4v, diffh5v, diffh6v, diffh7v;
- vec_s16_t diffl0v, diffl1v, diffl2v, diffl3v,
- diffl4v, diffl5v, diffl6v, diffl7v;
- vec_s16_t temp0v, temp1v, temp2v, temp3v,
- temp4v, temp5v, temp6v, temp7v;
- VEC_DIFF_HL( pix1, i_pix1, pix2, i_pix2, diffh0v, diffl0v );
- VEC_DIFF_HL( pix1, i_pix1, pix2, i_pix2, diffh1v, diffl1v );
- VEC_DIFF_HL( pix1, i_pix1, pix2, i_pix2, diffh2v, diffl2v );
- VEC_DIFF_HL( pix1, i_pix1, pix2, i_pix2, diffh3v, diffl3v );
- VEC_DIFF_HL( pix1, i_pix1, pix2, i_pix2, diffh4v, diffl4v );
- VEC_DIFF_HL( pix1, i_pix1, pix2, i_pix2, diffh5v, diffl5v );
- VEC_DIFF_HL( pix1, i_pix1, pix2, i_pix2, diffh6v, diffl6v );
- VEC_DIFF_HL( pix1, i_pix1, pix2, i_pix2, diffh7v, diffl7v );
- VEC_HADAMAR( diffh0v, diffh1v, diffh2v, diffh3v,
- temp0v, temp1v, temp2v, temp3v );
- VEC_HADAMAR( diffh4v, diffh5v, diffh6v, diffh7v,
- temp4v, temp5v, temp6v, temp7v );
- VEC_TRANSPOSE_8( temp0v, temp1v, temp2v, temp3v,
- temp4v, temp5v, temp6v, temp7v,
- diffh0v, diffh1v, diffh2v, diffh3v,
- diffh4v, diffh5v, diffh6v, diffh7v );
- VEC_HADAMAR( diffh0v, diffh1v, diffh2v, diffh3v,
- temp0v, temp1v, temp2v, temp3v );
- VEC_HADAMAR( diffh4v, diffh5v, diffh6v, diffh7v,
- temp4v, temp5v, temp6v, temp7v );
- VEC_ADD_ABS( temp0v, zero_s32v, satdv );
- VEC_ADD_ABS( temp1v, satdv, satdv );
- VEC_ADD_ABS( temp2v, satdv, satdv );
- VEC_ADD_ABS( temp3v, satdv, satdv );
- VEC_ADD_ABS( temp4v, satdv, satdv );
- VEC_ADD_ABS( temp5v, satdv, satdv );
- VEC_ADD_ABS( temp6v, satdv, satdv );
- VEC_ADD_ABS( temp7v, satdv, satdv );
- VEC_HADAMAR( diffl0v, diffl1v, diffl2v, diffl3v,
- temp0v, temp1v, temp2v, temp3v );
- VEC_HADAMAR( diffl4v, diffl5v, diffl6v, diffl7v,
- temp4v, temp5v, temp6v, temp7v );
- VEC_TRANSPOSE_8( temp0v, temp1v, temp2v, temp3v,
- temp4v, temp5v, temp6v, temp7v,
- diffl0v, diffl1v, diffl2v, diffl3v,
- diffl4v, diffl5v, diffl6v, diffl7v );
- VEC_HADAMAR( diffl0v, diffl1v, diffl2v, diffl3v,
- temp0v, temp1v, temp2v, temp3v );
- VEC_HADAMAR( diffl4v, diffl5v, diffl6v, diffl7v,
- temp4v, temp5v, temp6v, temp7v );
- VEC_ADD_ABS( temp0v, satdv, satdv );
- VEC_ADD_ABS( temp1v, satdv, satdv );
- VEC_ADD_ABS( temp2v, satdv, satdv );
- VEC_ADD_ABS( temp3v, satdv, satdv );
- VEC_ADD_ABS( temp4v, satdv, satdv );
- VEC_ADD_ABS( temp5v, satdv, satdv );
- VEC_ADD_ABS( temp6v, satdv, satdv );
- VEC_ADD_ABS( temp7v, satdv, satdv );
- satdv = vec_sums( satdv, zero_s32v );
- satdv = vec_splat( satdv, 3 );
- vec_ste( satdv, 0, &i_satd );
- return i_satd >> 1;
- }
- /***********************************************************************
- * SATD 16x16
- **********************************************************************/
- static int pixel_satd_16x16_altivec( uint8_t *pix1, intptr_t i_pix1,
- uint8_t *pix2, intptr_t i_pix2 )
- {
- ALIGNED_16( int i_satd );
- LOAD_ZERO;
- vec_s32_t satdv;
- vec_s16_t pix1v, pix2v;
- vec_s16_t diffh0v, diffh1v, diffh2v, diffh3v,
- diffh4v, diffh5v, diffh6v, diffh7v;
- vec_s16_t diffl0v, diffl1v, diffl2v, diffl3v,
- diffl4v, diffl5v, diffl6v, diffl7v;
- vec_s16_t temp0v, temp1v, temp2v, temp3v,
- temp4v, temp5v, temp6v, temp7v;
- VEC_DIFF_HL( pix1, i_pix1, pix2, i_pix2, diffh0v, diffl0v );
- VEC_DIFF_HL( pix1, i_pix1, pix2, i_pix2, diffh1v, diffl1v );
- VEC_DIFF_HL( pix1, i_pix1, pix2, i_pix2, diffh2v, diffl2v );
- VEC_DIFF_HL( pix1, i_pix1, pix2, i_pix2, diffh3v, diffl3v );
- VEC_DIFF_HL( pix1, i_pix1, pix2, i_pix2, diffh4v, diffl4v );
- VEC_DIFF_HL( pix1, i_pix1, pix2, i_pix2, diffh5v, diffl5v );
- VEC_DIFF_HL( pix1, i_pix1, pix2, i_pix2, diffh6v, diffl6v );
- VEC_DIFF_HL( pix1, i_pix1, pix2, i_pix2, diffh7v, diffl7v );
- VEC_HADAMAR( diffh0v, diffh1v, diffh2v, diffh3v,
- temp0v, temp1v, temp2v, temp3v );
- VEC_HADAMAR( diffh4v, diffh5v, diffh6v, diffh7v,
- temp4v, temp5v, temp6v, temp7v );
- VEC_TRANSPOSE_8( temp0v, temp1v, temp2v, temp3v,
- temp4v, temp5v, temp6v, temp7v,
- diffh0v, diffh1v, diffh2v, diffh3v,
- diffh4v, diffh5v, diffh6v, diffh7v );
- VEC_HADAMAR( diffh0v, diffh1v, diffh2v, diffh3v,
- temp0v, temp1v, temp2v, temp3v );
- VEC_HADAMAR( diffh4v, diffh5v, diffh6v, diffh7v,
- temp4v, temp5v, temp6v, temp7v );
- VEC_ADD_ABS( temp0v, zero_s32v, satdv );
- VEC_ADD_ABS( temp1v, satdv, satdv );
- VEC_ADD_ABS( temp2v, satdv, satdv );
- VEC_ADD_ABS( temp3v, satdv, satdv );
- VEC_ADD_ABS( temp4v, satdv, satdv );
- VEC_ADD_ABS( temp5v, satdv, satdv );
- VEC_ADD_ABS( temp6v, satdv, satdv );
- VEC_ADD_ABS( temp7v, satdv, satdv );
- VEC_HADAMAR( diffl0v, diffl1v, diffl2v, diffl3v,
- temp0v, temp1v, temp2v, temp3v );
- VEC_HADAMAR( diffl4v, diffl5v, diffl6v, diffl7v,
- temp4v, temp5v, temp6v, temp7v );
- VEC_TRANSPOSE_8( temp0v, temp1v, temp2v, temp3v,
- temp4v, temp5v, temp6v, temp7v,
- diffl0v, diffl1v, diffl2v, diffl3v,
- diffl4v, diffl5v, diffl6v, diffl7v );
- VEC_HADAMAR( diffl0v, diffl1v, diffl2v, diffl3v,
- temp0v, temp1v, temp2v, temp3v );
- VEC_HADAMAR( diffl4v, diffl5v, diffl6v, diffl7v,
- temp4v, temp5v, temp6v, temp7v );
- VEC_ADD_ABS( temp0v, satdv, satdv );
- VEC_ADD_ABS( temp1v, satdv, satdv );
- VEC_ADD_ABS( temp2v, satdv, satdv );
- VEC_ADD_ABS( temp3v, satdv, satdv );
- VEC_ADD_ABS( temp4v, satdv, satdv );
- VEC_ADD_ABS( temp5v, satdv, satdv );
- VEC_ADD_ABS( temp6v, satdv, satdv );
- VEC_ADD_ABS( temp7v, satdv, satdv );
- VEC_DIFF_HL( pix1, i_pix1, pix2, i_pix2, diffh0v, diffl0v );
- VEC_DIFF_HL( pix1, i_pix1, pix2, i_pix2, diffh1v, diffl1v );
- VEC_DIFF_HL( pix1, i_pix1, pix2, i_pix2, diffh2v, diffl2v );
- VEC_DIFF_HL( pix1, i_pix1, pix2, i_pix2, diffh3v, diffl3v );
- VEC_DIFF_HL( pix1, i_pix1, pix2, i_pix2, diffh4v, diffl4v );
- VEC_DIFF_HL( pix1, i_pix1, pix2, i_pix2, diffh5v, diffl5v );
- VEC_DIFF_HL( pix1, i_pix1, pix2, i_pix2, diffh6v, diffl6v );
- VEC_DIFF_HL( pix1, i_pix1, pix2, i_pix2, diffh7v, diffl7v );
- VEC_HADAMAR( diffh0v, diffh1v, diffh2v, diffh3v,
- temp0v, temp1v, temp2v, temp3v );
- VEC_HADAMAR( diffh4v, diffh5v, diffh6v, diffh7v,
- temp4v, temp5v, temp6v, temp7v );
- VEC_TRANSPOSE_8( temp0v, temp1v, temp2v, temp3v,
- temp4v, temp5v, temp6v, temp7v,
- diffh0v, diffh1v, diffh2v, diffh3v,
- diffh4v, diffh5v, diffh6v, diffh7v );
- VEC_HADAMAR( diffh0v, diffh1v, diffh2v, diffh3v,
- temp0v, temp1v, temp2v, temp3v );
- VEC_HADAMAR( diffh4v, diffh5v, diffh6v, diffh7v,
- temp4v, temp5v, temp6v, temp7v );
- VEC_ADD_ABS( temp0v, satdv, satdv );
- VEC_ADD_ABS( temp1v, satdv, satdv );
- VEC_ADD_ABS( temp2v, satdv, satdv );
- VEC_ADD_ABS( temp3v, satdv, satdv );
- VEC_ADD_ABS( temp4v, satdv, satdv );
- VEC_ADD_ABS( temp5v, satdv, satdv );
- VEC_ADD_ABS( temp6v, satdv, satdv );
- VEC_ADD_ABS( temp7v, satdv, satdv );
- VEC_HADAMAR( diffl0v, diffl1v, diffl2v, diffl3v,
- temp0v, temp1v, temp2v, temp3v );
- VEC_HADAMAR( diffl4v, diffl5v, diffl6v, diffl7v,
- temp4v, temp5v, temp6v, temp7v );
- VEC_TRANSPOSE_8( temp0v, temp1v, temp2v, temp3v,
- temp4v, temp5v, temp6v, temp7v,
- diffl0v, diffl1v, diffl2v, diffl3v,
- diffl4v, diffl5v, diffl6v, diffl7v );
- VEC_HADAMAR( diffl0v, diffl1v, diffl2v, diffl3v,
- temp0v, temp1v, temp2v, temp3v );
- VEC_HADAMAR( diffl4v, diffl5v, diffl6v, diffl7v,
- temp4v, temp5v, temp6v, temp7v );
- VEC_ADD_ABS( temp0v, satdv, satdv );
- VEC_ADD_ABS( temp1v, satdv, satdv );
- VEC_ADD_ABS( temp2v, satdv, satdv );
- VEC_ADD_ABS( temp3v, satdv, satdv );
- VEC_ADD_ABS( temp4v, satdv, satdv );
- VEC_ADD_ABS( temp5v, satdv, satdv );
- VEC_ADD_ABS( temp6v, satdv, satdv );
- VEC_ADD_ABS( temp7v, satdv, satdv );
- satdv = vec_sums( satdv, zero_s32v );
- satdv = vec_splat( satdv, 3 );
- vec_ste( satdv, 0, &i_satd );
- return i_satd >> 1;
- }
- /***********************************************************************
- * Interleaved SAD routines
- **********************************************************************/
- static void pixel_sad_x4_16x16_altivec( uint8_t *fenc,
- uint8_t *pix0, uint8_t *pix1,
- uint8_t *pix2, uint8_t *pix3,
- intptr_t i_stride, int scores[4] )
- {
- ALIGNED_16( int sum0 );
- ALIGNED_16( int sum1 );
- ALIGNED_16( int sum2 );
- ALIGNED_16( int sum3 );
- LOAD_ZERO;
- vec_u8_t fencv, pix0v, pix1v, pix2v, pix3v;
- vec_s32_t sum0v, sum1v, sum2v, sum3v;
- sum0v = vec_splat_s32(0);
- sum1v = vec_splat_s32(0);
- sum2v = vec_splat_s32(0);
- sum3v = vec_splat_s32(0);
- for( int y = 0; y < 8; y++ )
- {
- pix0v = vec_vsx_ld( 0, pix0 );
- pix0 += i_stride;
- pix1v = vec_vsx_ld( 0, pix1 );
- pix1 += i_stride;
- fencv = vec_ld(0, fenc);
- fenc += FENC_STRIDE;
- pix2v = vec_vsx_ld( 0, pix2 );
- pix2 += i_stride;
- pix3v = vec_vsx_ld( 0, pix3 );
- pix3 += i_stride;
- sum0v = (vec_s32_t) vec_sum4s( vec_absd( fencv, pix0v ), (vec_u32_t) sum0v );
- sum1v = (vec_s32_t) vec_sum4s( vec_absd( fencv, pix1v ), (vec_u32_t) sum1v );
- sum2v = (vec_s32_t) vec_sum4s( vec_absd( fencv, pix2v ), (vec_u32_t) sum2v );
- sum3v = (vec_s32_t) vec_sum4s( vec_absd( fencv, pix3v ), (vec_u32_t) sum3v );
- pix0v = vec_vsx_ld( 0, pix0 );
- pix0 += i_stride;
- pix1v = vec_vsx_ld( 0, pix1 );
- pix1 += i_stride;
- fencv = vec_ld(0, fenc);
- fenc += FENC_STRIDE;
- pix2v = vec_vsx_ld( 0, pix2 );
- pix2 += i_stride;
- pix3v = vec_vsx_ld( 0, pix3 );
- pix3 += i_stride;
- sum0v = (vec_s32_t) vec_sum4s( vec_absd( fencv, pix0v ), (vec_u32_t) sum0v );
- sum1v = (vec_s32_t) vec_sum4s( vec_absd( fencv, pix1v ), (vec_u32_t) sum1v );
- sum2v = (vec_s32_t) vec_sum4s( vec_absd( fencv, pix2v ), (vec_u32_t) sum2v );
- sum3v = (vec_s32_t) vec_sum4s( vec_absd( fencv, pix3v ), (vec_u32_t) sum3v );
- }
- sum0v = vec_sums( sum0v, zero_s32v );
- sum1v = vec_sums( sum1v, zero_s32v );
- sum2v = vec_sums( sum2v, zero_s32v );
- sum3v = vec_sums( sum3v, zero_s32v );
- sum0v = vec_splat( sum0v, 3 );
- sum1v = vec_splat( sum1v, 3 );
- sum2v = vec_splat( sum2v, 3 );
- sum3v = vec_splat( sum3v, 3 );
- vec_ste( sum0v, 0, &sum0);
- vec_ste( sum1v, 0, &sum1);
- vec_ste( sum2v, 0, &sum2);
- vec_ste( sum3v, 0, &sum3);
- scores[0] = sum0;
- scores[1] = sum1;
- scores[2] = sum2;
- scores[3] = sum3;
- }
- static void pixel_sad_x3_16x16_altivec( uint8_t *fenc, uint8_t *pix0,
- uint8_t *pix1, uint8_t *pix2,
- intptr_t i_stride, int scores[3] )
- {
- ALIGNED_16( int sum0 );
- ALIGNED_16( int sum1 );
- ALIGNED_16( int sum2 );
- LOAD_ZERO;
- vec_u8_t fencv, pix0v, pix1v, pix2v;
- vec_s32_t sum0v, sum1v, sum2v;
- sum0v = vec_splat_s32(0);
- sum1v = vec_splat_s32(0);
- sum2v = vec_splat_s32(0);
- for( int y = 0; y < 8; y++ )
- {
- pix0v = vec_vsx_ld( 0, pix0 );
- pix0 += i_stride;
- pix1v = vec_vsx_ld( 0, pix1 );
- pix1 += i_stride;
- fencv = vec_ld(0, fenc);
- fenc += FENC_STRIDE;
- pix2v = vec_vsx_ld( 0, pix2 );
- pix2 += i_stride;
- sum0v = (vec_s32_t) vec_sum4s( vec_absd( fencv, pix0v ), (vec_u32_t) sum0v );
- sum1v = (vec_s32_t) vec_sum4s( vec_absd( fencv, pix1v ), (vec_u32_t) sum1v );
- sum2v = (vec_s32_t) vec_sum4s( vec_absd( fencv, pix2v ), (vec_u32_t) sum2v );
- pix0v = vec_vsx_ld( 0, pix0 );
- pix0 += i_stride;
- pix1v = vec_vsx_ld( 0, pix1 );
- pix1 += i_stride;
- fencv = vec_ld(0, fenc);
- fenc += FENC_STRIDE;
- pix2v = vec_vsx_ld( 0, pix2 );
- pix2 += i_stride;
- sum0v = (vec_s32_t) vec_sum4s( vec_absd( fencv, pix0v ), (vec_u32_t) sum0v );
- sum1v = (vec_s32_t) vec_sum4s( vec_absd( fencv, pix1v ), (vec_u32_t) sum1v );
- sum2v = (vec_s32_t) vec_sum4s( vec_absd( fencv, pix2v ), (vec_u32_t) sum2v );
- }
- sum0v = vec_sums( sum0v, zero_s32v );
- sum1v = vec_sums( sum1v, zero_s32v );
- sum2v = vec_sums( sum2v, zero_s32v );
- sum0v = vec_splat( sum0v, 3 );
- sum1v = vec_splat( sum1v, 3 );
- sum2v = vec_splat( sum2v, 3 );
- vec_ste( sum0v, 0, &sum0);
- vec_ste( sum1v, 0, &sum1);
- vec_ste( sum2v, 0, &sum2);
- scores[0] = sum0;
- scores[1] = sum1;
- scores[2] = sum2;
- }
- static void pixel_sad_x4_16x8_altivec( uint8_t *fenc, uint8_t *pix0, uint8_t *pix1, uint8_t *pix2,
- uint8_t *pix3, intptr_t i_stride, int scores[4] )
- {
- ALIGNED_16( int sum0 );
- ALIGNED_16( int sum1 );
- ALIGNED_16( int sum2 );
- ALIGNED_16( int sum3 );
- LOAD_ZERO;
- vec_u8_t fencv, pix0v, pix1v, pix2v, pix3v;
- vec_s32_t sum0v, sum1v, sum2v, sum3v;
- sum0v = vec_splat_s32(0);
- sum1v = vec_splat_s32(0);
- sum2v = vec_splat_s32(0);
- sum3v = vec_splat_s32(0);
- for( int y = 0; y < 4; y++ )
- {
- pix0v = vec_vsx_ld( 0, pix0 );
- pix0 += i_stride;
- pix1v = vec_vsx_ld( 0, pix1 );
- pix1 += i_stride;
- fencv = vec_ld( 0, fenc );
- fenc += FENC_STRIDE;
- pix2v = vec_vsx_ld( 0, pix2 );
- pix2 += i_stride;
- pix3v = vec_vsx_ld( 0, pix3 );
- pix3 += i_stride;
- sum0v = (vec_s32_t) vec_sum4s( vec_absd( fencv, pix0v ), (vec_u32_t) sum0v );
- sum1v = (vec_s32_t) vec_sum4s( vec_absd( fencv, pix1v ), (vec_u32_t) sum1v );
- sum2v = (vec_s32_t) vec_sum4s( vec_absd( fencv, pix2v ), (vec_u32_t) sum2v );
- sum3v = (vec_s32_t) vec_sum4s( vec_absd( fencv, pix3v ), (vec_u32_t) sum3v );
- pix0v = vec_vsx_ld( 0, pix0 );
- pix0 += i_stride;
- pix1v = vec_vsx_ld( 0, pix1 );
- pix1 += i_stride;
- fencv = vec_ld(0, fenc);
- fenc += FENC_STRIDE;
- pix2v = vec_vsx_ld( 0, pix2 );
- pix2 += i_stride;
- pix3v = vec_vsx_ld( 0, pix3 );
- pix3 += i_stride;
- sum0v = (vec_s32_t) vec_sum4s( vec_absd( fencv, pix0v ), (vec_u32_t) sum0v );
- sum1v = (vec_s32_t) vec_sum4s( vec_absd( fencv, pix1v ), (vec_u32_t) sum1v );
- sum2v = (vec_s32_t) vec_sum4s( vec_absd( fencv, pix2v ), (vec_u32_t) sum2v );
- sum3v = (vec_s32_t) vec_sum4s( vec_absd( fencv, pix3v ), (vec_u32_t) sum3v );
- }
- sum0v = vec_sums( sum0v, zero_s32v );
- sum1v = vec_sums( sum1v, zero_s32v );
- sum2v = vec_sums( sum2v, zero_s32v );
- sum3v = vec_sums( sum3v, zero_s32v );
- sum0v = vec_splat( sum0v, 3 );
- sum1v = vec_splat( sum1v, 3 );
- sum2v = vec_splat( sum2v, 3 );
- sum3v = vec_splat( sum3v, 3 );
- vec_ste( sum0v, 0, &sum0);
- vec_ste( sum1v, 0, &sum1);
- vec_ste( sum2v, 0, &sum2);
- vec_ste( sum3v, 0, &sum3);
- scores[0] = sum0;
- scores[1] = sum1;
- scores[2] = sum2;
- scores[3] = sum3;
- }
- static void pixel_sad_x3_16x8_altivec( uint8_t *fenc, uint8_t *pix0,
- uint8_t *pix1, uint8_t *pix2,
- intptr_t i_stride, int scores[3] )
- {
- ALIGNED_16( int sum0 );
- ALIGNED_16( int sum1 );
- ALIGNED_16( int sum2 );
- LOAD_ZERO;
- vec_u8_t fencv, pix0v, pix1v, pix2v;
- vec_s32_t sum0v, sum1v, sum2v;
- sum0v = vec_splat_s32(0);
- sum1v = vec_splat_s32(0);
- sum2v = vec_splat_s32(0);
- for( int y = 0; y < 4; y++ )
- {
- pix0v = vec_vsx_ld(0, pix0);
- pix0 += i_stride;
- pix1v = vec_vsx_ld(0, pix1);
- pix1 += i_stride;
- fencv = vec_ld(0, fenc);
- fenc += FENC_STRIDE;
- pix2v = vec_vsx_ld(0, pix2);
- pix2 += i_stride;
- sum0v = (vec_s32_t) vec_sum4s( vec_absd( fencv, pix0v ), (vec_u32_t) sum0v );
- sum1v = (vec_s32_t) vec_sum4s( vec_absd( fencv, pix1v ), (vec_u32_t) sum1v );
- sum2v = (vec_s32_t) vec_sum4s( vec_absd( fencv, pix2v ), (vec_u32_t) sum2v );
- pix0v = vec_vsx_ld(0, pix0);
- pix0 += i_stride;
- pix1v = vec_vsx_ld(0, pix1);
- pix1 += i_stride;
- fencv = vec_ld(0, fenc);
- fenc += FENC_STRIDE;
- pix2v = vec_vsx_ld(0, pix2);
- pix2 += i_stride;
- sum0v = (vec_s32_t) vec_sum4s( vec_absd( fencv, pix0v ), (vec_u32_t) sum0v );
- sum1v = (vec_s32_t) vec_sum4s( vec_absd( fencv, pix1v ), (vec_u32_t) sum1v );
- sum2v = (vec_s32_t) vec_sum4s( vec_absd( fencv, pix2v ), (vec_u32_t) sum2v );
- }
- sum0v = vec_sums( sum0v, zero_s32v );
- sum1v = vec_sums( sum1v, zero_s32v );
- sum2v = vec_sums( sum2v, zero_s32v );
- sum0v = vec_splat( sum0v, 3 );
- sum1v = vec_splat( sum1v, 3 );
- sum2v = vec_splat( sum2v, 3 );
- vec_ste( sum0v, 0, &sum0);
- vec_ste( sum1v, 0, &sum1);
- vec_ste( sum2v, 0, &sum2);
- scores[0] = sum0;
- scores[1] = sum1;
- scores[2] = sum2;
- }
- static void pixel_sad_x4_8x16_altivec( uint8_t *fenc,
- uint8_t *pix0, uint8_t *pix1,
- uint8_t *pix2, uint8_t *pix3,
- intptr_t i_stride, int scores[4] )
- {
- ALIGNED_16( int sum0 );
- ALIGNED_16( int sum1 );
- ALIGNED_16( int sum2 );
- ALIGNED_16( int sum3 );
- LOAD_ZERO;
- vec_u8_t fencv, pix0v, pix1v, pix2v, pix3v;
- vec_s32_t sum0v, sum1v, sum2v, sum3v;
- sum0v = vec_splat_s32(0);
- sum1v = vec_splat_s32(0);
- sum2v = vec_splat_s32(0);
- sum3v = vec_splat_s32(0);
- for( int y = 0; y < 8; y++ )
- {
- pix0v = vec_vsx_ld(0, pix0);
- pix0 += i_stride;
- pix1v = vec_vsx_ld(0, pix1);
- pix1 += i_stride;
- fencv = vec_vsx_ld(0, fenc);
- fenc += FENC_STRIDE;
- pix2v = vec_vsx_ld(0, pix2);
- pix2 += i_stride;
- pix3v = vec_vsx_ld(0, pix3);
- pix3 += i_stride;
- sum0v = (vec_s32_t) vec_sum4s( vec_absd( fencv, pix0v ), (vec_u32_t) sum0v );
- sum1v = (vec_s32_t) vec_sum4s( vec_absd( fencv, pix1v ), (vec_u32_t) sum1v );
- sum2v = (vec_s32_t) vec_sum4s( vec_absd( fencv, pix2v ), (vec_u32_t) sum2v );
- sum3v = (vec_s32_t) vec_sum4s( vec_absd( fencv, pix3v ), (vec_u32_t) sum3v );
- pix0v = vec_vsx_ld(0, pix0);
- pix0 += i_stride;
- pix1v = vec_vsx_ld(0, pix1);
- pix1 += i_stride;
- fencv = vec_vsx_ld(0, fenc);
- fenc += FENC_STRIDE;
- pix2v = vec_vsx_ld(0, pix2);
- pix2 += i_stride;
- pix3v = vec_vsx_ld(0, pix3);
- pix3 += i_stride;
- sum0v = (vec_s32_t) vec_sum4s( vec_absd( fencv, pix0v ), (vec_u32_t) sum0v );
- sum1v = (vec_s32_t) vec_sum4s( vec_absd( fencv, pix1v ), (vec_u32_t) sum1v );
- sum2v = (vec_s32_t) vec_sum4s( vec_absd( fencv, pix2v ), (vec_u32_t) sum2v );
- sum3v = (vec_s32_t) vec_sum4s( vec_absd( fencv, pix3v ), (vec_u32_t) sum3v );
- }
- sum0v = vec_sum2s( sum0v, zero_s32v );
- sum1v = vec_sum2s( sum1v, zero_s32v );
- sum2v = vec_sum2s( sum2v, zero_s32v );
- sum3v = vec_sum2s( sum3v, zero_s32v );
- sum0v = vec_splat( sum0v, 1 );
- sum1v = vec_splat( sum1v, 1 );
- sum2v = vec_splat( sum2v, 1 );
- sum3v = vec_splat( sum3v, 1 );
- vec_ste( sum0v, 0, &sum0);
- vec_ste( sum1v, 0, &sum1);
- vec_ste( sum2v, 0, &sum2);
- vec_ste( sum3v, 0, &sum3);
- scores[0] = sum0;
- scores[1] = sum1;
- scores[2] = sum2;
- scores[3] = sum3;
- }
- static void pixel_sad_x3_8x16_altivec( uint8_t *fenc, uint8_t *pix0,
- uint8_t *pix1, uint8_t *pix2,
- intptr_t i_stride, int scores[3] )
- {
- ALIGNED_16( int sum0 );
- ALIGNED_16( int sum1 );
- ALIGNED_16( int sum2 );
- LOAD_ZERO;
- vec_u8_t fencv, pix0v, pix1v, pix2v;
- vec_s32_t sum0v, sum1v, sum2v;
- sum0v = vec_splat_s32(0);
- sum1v = vec_splat_s32(0);
- sum2v = vec_splat_s32(0);
- for( int y = 0; y < 8; y++ )
- {
- pix0v = vec_vsx_ld(0, pix0);
- pix0 += i_stride;
- pix1v = vec_vsx_ld(0, pix1);
- pix1 += i_stride;
- fencv = vec_vsx_ld(0, fenc);
- fenc += FENC_STRIDE;
- pix2v = vec_vsx_ld(0, pix2);
- pix2 += i_stride;
- sum0v = (vec_s32_t) vec_sum4s( vec_absd( fencv, pix0v ), (vec_u32_t) sum0v );
- sum1v = (vec_s32_t) vec_sum4s( vec_absd( fencv, pix1v ), (vec_u32_t) sum1v );
- sum2v = (vec_s32_t) vec_sum4s( vec_absd( fencv, pix2v ), (vec_u32_t) sum2v );
- pix0v = vec_vsx_ld(0, pix0);
- pix0 += i_stride;
- pix1v = vec_vsx_ld(0, pix1);
- pix1 += i_stride;
- fencv = vec_vsx_ld(0, fenc);
- fenc += FENC_STRIDE;
- pix2v = vec_vsx_ld(0, pix2);
- pix2 += i_stride;
- sum0v = (vec_s32_t) vec_sum4s( vec_absd( fencv, pix0v ), (vec_u32_t) sum0v );
- sum1v = (vec_s32_t) vec_sum4s( vec_absd( fencv, pix1v ), (vec_u32_t) sum1v );
- sum2v = (vec_s32_t) vec_sum4s( vec_absd( fencv, pix2v ), (vec_u32_t) sum2v );
- }
- sum0v = vec_sum2s( sum0v, zero_s32v );
- sum1v = vec_sum2s( sum1v, zero_s32v );
- sum2v = vec_sum2s( sum2v, zero_s32v );
- sum0v = vec_splat( sum0v, 1 );
- sum1v = vec_splat( sum1v, 1 );
- sum2v = vec_splat( sum2v, 1 );
- vec_ste( sum0v, 0, &sum0);
- vec_ste( sum1v, 0, &sum1);
- vec_ste( sum2v, 0, &sum2);
- scores[0] = sum0;
- scores[1] = sum1;
- scores[2] = sum2;
- }
- static void pixel_sad_x4_8x8_altivec( uint8_t *fenc,
- uint8_t *pix0, uint8_t *pix1,
- uint8_t *pix2, uint8_t *pix3,
- intptr_t i_stride, int scores[4] )
- {
- ALIGNED_16( int sum0 );
- ALIGNED_16( int sum1 );
- ALIGNED_16( int sum2 );
- ALIGNED_16( int sum3 );
- LOAD_ZERO;
- vec_u8_t fencv, pix0v, pix1v, pix2v, pix3v;
- vec_s32_t sum0v, sum1v, sum2v, sum3v;
- sum0v = vec_splat_s32(0);
- sum1v = vec_splat_s32(0);
- sum2v = vec_splat_s32(0);
- sum3v = vec_splat_s32(0);
- for( int y = 0; y < 4; y++ )
- {
- pix0v = vec_vsx_ld(0, pix0);
- pix0 += i_stride;
- pix1v = vec_vsx_ld(0, pix1);
- pix1 += i_stride;
- fencv = vec_vsx_ld(0, fenc);
- fenc += FENC_STRIDE;
- pix2v = vec_vsx_ld(0, pix2);
- pix2 += i_stride;
- pix3v = vec_vsx_ld(0, pix3);
- pix3 += i_stride;
- sum0v = (vec_s32_t) vec_sum4s( vec_absd( fencv, pix0v ), (vec_u32_t) sum0v );
- sum1v = (vec_s32_t) vec_sum4s( vec_absd( fencv, pix1v ), (vec_u32_t) sum1v );
- sum2v = (vec_s32_t) vec_sum4s( vec_absd( fencv, pix2v ), (vec_u32_t) sum2v );
- sum3v = (vec_s32_t) vec_sum4s( vec_absd( fencv, pix3v ), (vec_u32_t) sum3v );
- pix0v = vec_vsx_ld(0, pix0);
- pix0 += i_stride;
- pix1v = vec_vsx_ld(0, pix1);
- pix1 += i_stride;
- fencv = vec_vsx_ld(0, fenc);
- fenc += FENC_STRIDE;
- pix2v = vec_vsx_ld(0, pix2);
- pix2 += i_stride;
- pix3v = vec_vsx_ld(0, pix3);
- pix3 += i_stride;
- sum0v = (vec_s32_t) vec_sum4s( vec_absd( fencv, pix0v ), (vec_u32_t) sum0v );
- sum1v = (vec_s32_t) vec_sum4s( vec_absd( fencv, pix1v ), (vec_u32_t) sum1v );
- sum2v = (vec_s32_t) vec_sum4s( vec_absd( fencv, pix2v ), (vec_u32_t) sum2v );
- sum3v = (vec_s32_t) vec_sum4s( vec_absd( fencv, pix3v ), (vec_u32_t) sum3v );
- }
- sum0v = vec_sum2s( sum0v, zero_s32v );
- sum1v = vec_sum2s( sum1v, zero_s32v );
- sum2v = vec_sum2s( sum2v, zero_s32v );
- sum3v = vec_sum2s( sum3v, zero_s32v );
- sum0v = vec_splat( sum0v, 1 );
- sum1v = vec_splat( sum1v, 1 );
- sum2v = vec_splat( sum2v, 1 );
- sum3v = vec_splat( sum3v, 1 );
- vec_ste( sum0v, 0, &sum0);
- vec_ste( sum1v, 0, &sum1);
- vec_ste( sum2v, 0, &sum2);
- vec_ste( sum3v, 0, &sum3);
- scores[0] = sum0;
- scores[1] = sum1;
- scores[2] = sum2;
- scores[3] = sum3;
- }
- static void pixel_sad_x3_8x8_altivec( uint8_t *fenc, uint8_t *pix0,
- uint8_t *pix1, uint8_t *pix2,
- intptr_t i_stride, int scores[3] )
- {
- ALIGNED_16( int sum0 );
- ALIGNED_16( int sum1 );
- ALIGNED_16( int sum2 );
- LOAD_ZERO;
- vec_u8_t fencv, pix0v, pix1v, pix2v;
- vec_s32_t sum0v, sum1v, sum2v;
- sum0v = vec_splat_s32(0);
- sum1v = vec_splat_s32(0);
- sum2v = vec_splat_s32(0);
- for( int y = 0; y < 4; y++ )
- {
- pix0v = vec_vsx_ld(0, pix0);
- pix0 += i_stride;
- pix1v = vec_vsx_ld(0, pix1);
- pix1 += i_stride;
- fencv = vec_vsx_ld(0, fenc);
- fenc += FENC_STRIDE;
- pix2v = vec_vsx_ld(0, pix2);
- pix2 += i_stride;
- sum0v = (vec_s32_t) vec_sum4s( vec_absd( fencv, pix0v ), (vec_u32_t) sum0v );
- sum1v = (vec_s32_t) vec_sum4s( vec_absd( fencv, pix1v ), (vec_u32_t) sum1v );
- sum2v = (vec_s32_t) vec_sum4s( vec_absd( fencv, pix2v ), (vec_u32_t) sum2v );
- pix0v = vec_vsx_ld(0, pix0);
- pix0 += i_stride;
- pix1v = vec_vsx_ld(0, pix1);
- pix1 += i_stride;
- fencv = vec_vsx_ld(0, fenc);
- fenc += FENC_STRIDE;
- pix2v = vec_vsx_ld(0, pix2);
- pix2 += i_stride;
- sum0v = (vec_s32_t) vec_sum4s( vec_absd( fencv, pix0v ), (vec_u32_t) sum0v );
- sum1v = (vec_s32_t) vec_sum4s( vec_absd( fencv, pix1v ), (vec_u32_t) sum1v );
- sum2v = (vec_s32_t) vec_sum4s( vec_absd( fencv, pix2v ), (vec_u32_t) sum2v );
- }
- sum0v = vec_sum2s( sum0v, zero_s32v );
- sum1v = vec_sum2s( sum1v, zero_s32v );
- sum2v = vec_sum2s( sum2v, zero_s32v );
- sum0v = vec_splat( sum0v, 1 );
- sum1v = vec_splat( sum1v, 1 );
- sum2v = vec_splat( sum2v, 1 );
- vec_ste( sum0v, 0, &sum0);
- vec_ste( sum1v, 0, &sum1);
- vec_ste( sum2v, 0, &sum2);
- scores[0] = sum0;
- scores[1] = sum1;
- scores[2] = sum2;
- }
- /***********************************************************************
- * SSD routines
- **********************************************************************/
- static int pixel_ssd_16x16_altivec( uint8_t *pix1, intptr_t i_stride_pix1,
- uint8_t *pix2, intptr_t i_stride_pix2 )
- {
- ALIGNED_16( int sum );
- LOAD_ZERO;
- vec_u8_t pix1vA, pix2vA, pix1vB, pix2vB;
- vec_u32_t sumv;
- vec_u8_t diffA, diffB;
- sumv = vec_splat_u32(0);
- pix2vA = vec_vsx_ld(0, pix2);
- pix1vA = vec_ld(0, pix1);
- for( int y = 0; y < 7; y++ )
- {
- pix1 += i_stride_pix1;
- pix2 += i_stride_pix2;
- pix2vB = vec_vsx_ld(0, pix2);
- pix1vB = vec_ld(0, pix1);
- diffA = vec_absd(pix1vA, pix2vA);
- sumv = vec_msum(diffA, diffA, sumv);
- pix1 += i_stride_pix1;
- pix2 += i_stride_pix2;
- pix2vA = vec_vsx_ld(0, pix2);
- pix1vA = vec_ld(0, pix1);
- diffB = vec_absd(pix1vB, pix2vB);
- sumv = vec_msum(diffB, diffB, sumv);
- }
- pix1 += i_stride_pix1;
- pix2 += i_stride_pix2;
- pix2vB = vec_vsx_ld(0, pix2);
- pix1vB = vec_ld(0, pix1);
- diffA = vec_absd(pix1vA, pix2vA);
- sumv = vec_msum(diffA, diffA, sumv);
- diffB = vec_absd(pix1vB, pix2vB);
- sumv = vec_msum(diffB, diffB, sumv);
- sumv = (vec_u32_t) vec_sums((vec_s32_t) sumv, zero_s32v);
- sumv = vec_splat(sumv, 3);
- vec_ste((vec_s32_t) sumv, 0, &sum);
- return sum;
- }
- static int pixel_ssd_8x8_altivec( uint8_t *pix1, intptr_t i_stride_pix1,
- uint8_t *pix2, intptr_t i_stride_pix2 )
- {
- ALIGNED_16( int sum );
- LOAD_ZERO;
- vec_u8_t pix1v, pix2v;
- vec_u32_t sumv;
- vec_u8_t diffv;
- const vec_u32_t sel = (vec_u32_t)CV(-1,-1,0,0);
- sumv = vec_splat_u32(0);
- for( int y = 0; y < 8; y++ )
- {
- pix1v = vec_vsx_ld(0, pix1);
- pix2v = vec_vsx_ld(0, pix2);
- diffv = vec_absd( pix1v, pix2v );
- sumv = vec_msum(diffv, diffv, sumv);
- pix1 += i_stride_pix1;
- pix2 += i_stride_pix2;
- }
- sumv = vec_sel( zero_u32v, sumv, sel );
- sumv = (vec_u32_t) vec_sums((vec_s32_t) sumv, zero_s32v);
- sumv = vec_splat(sumv, 3);
- vec_ste((vec_s32_t) sumv, 0, &sum);
- return sum;
- }
- /****************************************************************************
- * variance
- ****************************************************************************/
- static uint64_t pixel_var_16x16_altivec( uint8_t *pix, intptr_t i_stride )
- {
- ALIGNED_16(uint32_t sum_tab[4]);
- ALIGNED_16(uint32_t sqr_tab[4]);
- LOAD_ZERO;
- vec_u32_t sqr_v = zero_u32v;
- vec_u32_t sum_v = zero_u32v;
- for( int y = 0; y < 16; y++ )
- {
- vec_u8_t pix0_v = vec_ld(0, pix);
- sum_v = vec_sum4s(pix0_v, sum_v);
- sqr_v = vec_msum(pix0_v, pix0_v, sqr_v);
- pix += i_stride;
- }
- sum_v = (vec_u32_t)vec_sums( (vec_s32_t)sum_v, zero_s32v );
- sqr_v = (vec_u32_t)vec_sums( (vec_s32_t)sqr_v, zero_s32v );
- vec_ste(sum_v, 12, sum_tab);
- vec_ste(sqr_v, 12, sqr_tab);
- uint32_t sum = sum_tab[3];
- uint32_t sqr = sqr_tab[3];
- return sum + ((uint64_t)sqr<<32);
- }
- static uint64_t pixel_var_8x8_altivec( uint8_t *pix, intptr_t i_stride )
- {
- ALIGNED_16(uint32_t sum_tab[4]);
- ALIGNED_16(uint32_t sqr_tab[4]);
- LOAD_ZERO;
- vec_u32_t sqr_v = zero_u32v;
- vec_u32_t sum_v = zero_u32v;
- static const vec_u8_t perm_tab[] =
- {
- CV(0x00, 0x01, 0x02, 0x03, 0x04, 0x05, 0x06, 0x07, /* pix=mod16, i_stride=mod16 */
- 0x10, 0x11, 0x12, 0x13, 0x14, 0x15, 0x16, 0x17),
- CV(0x08, 0x09, 0x0A, 0x0B, 0x0C, 0x0D, 0x0E, 0x0F, /* pix=mod8, i_stride=mod16 */
- 0x18, 0x19, 0x1A, 0x1B, 0x1C, 0x1D, 0x1E, 0x1F),
- };
- vec_u8_t perm = perm_tab[ ((uintptr_t)pix & 8) >> 3 ];
- for( int y = 0; y < 4; y++ )
- {
- vec_u8_t pix0_v = vec_ld(0, pix);
- vec_u8_t pix1_v = vec_ld(i_stride, pix);
- vec_u8_t pix_v = vec_perm(pix0_v, pix1_v, perm);
- sum_v = vec_sum4s(pix_v, sum_v);
- sqr_v = vec_msum(pix_v, pix_v, sqr_v);
- pix += i_stride<<1;
- }
- sum_v = (vec_u32_t)vec_sums( (vec_s32_t)sum_v, zero_s32v );
- sqr_v = (vec_u32_t)vec_sums( (vec_s32_t)sqr_v, zero_s32v );
- vec_ste(sum_v, 12, sum_tab);
- vec_ste(sqr_v, 12, sqr_tab);
- uint32_t sum = sum_tab[3];
- uint32_t sqr = sqr_tab[3];
- return sum + ((uint64_t)sqr<<32);
- }
- /**********************************************************************
- * SA8D routines: sum of 8x8 Hadamard transformed differences
- **********************************************************************/
- /* SA8D_1D unrolled by 8 in Altivec */
- #define SA8D_1D_ALTIVEC( sa8d0v, sa8d1v, sa8d2v, sa8d3v, \
- sa8d4v, sa8d5v, sa8d6v, sa8d7v ) \
- { \
- /* int a0 = SRC(0) + SRC(4) */ \
- vec_s16_t a0v = vec_add(sa8d0v, sa8d4v); \
- /* int a4 = SRC(0) - SRC(4) */ \
- vec_s16_t a4v = vec_sub(sa8d0v, sa8d4v); \
- /* int a1 = SRC(1) + SRC(5) */ \
- vec_s16_t a1v = vec_add(sa8d1v, sa8d5v); \
- /* int a5 = SRC(1) - SRC(5) */ \
- vec_s16_t a5v = vec_sub(sa8d1v, sa8d5v); \
- /* int a2 = SRC(2) + SRC(6) */ \
- vec_s16_t a2v = vec_add(sa8d2v, sa8d6v); \
- /* int a6 = SRC(2) - SRC(6) */ \
- vec_s16_t a6v = vec_sub(sa8d2v, sa8d6v); \
- /* int a3 = SRC(3) + SRC(7) */ \
- vec_s16_t a3v = vec_add(sa8d3v, sa8d7v); \
- /* int a7 = SRC(3) - SRC(7) */ \
- vec_s16_t a7v = vec_sub(sa8d3v, sa8d7v); \
- \
- /* int b0 = a0 + a2 */ \
- vec_s16_t b0v = vec_add(a0v, a2v); \
- /* int b2 = a0 - a2; */ \
- vec_s16_t b2v = vec_sub(a0v, a2v); \
- /* int b1 = a1 + a3; */ \
- vec_s16_t b1v = vec_add(a1v, a3v); \
- /* int b3 = a1 - a3; */ \
- vec_s16_t b3v = vec_sub(a1v, a3v); \
- /* int b4 = a4 + a6; */ \
- vec_s16_t b4v = vec_add(a4v, a6v); \
- /* int b6 = a4 - a6; */ \
- vec_s16_t b6v = vec_sub(a4v, a6v); \
- /* int b5 = a5 + a7; */ \
- vec_s16_t b5v = vec_add(a5v, a7v); \
- /* int b7 = a5 - a7; */ \
- vec_s16_t b7v = vec_sub(a5v, a7v); \
- \
- /* DST(0, b0 + b1) */ \
- sa8d0v = vec_add(b0v, b1v); \
- /* DST(1, b0 - b1) */ \
- sa8d1v = vec_sub(b0v, b1v); \
- /* DST(2, b2 + b3) */ \
- sa8d2v = vec_add(b2v, b3v); \
- /* DST(3, b2 - b3) */ \
- sa8d3v = vec_sub(b2v, b3v); \
- /* DST(4, b4 + b5) */ \
- sa8d4v = vec_add(b4v, b5v); \
- /* DST(5, b4 - b5) */ \
- sa8d5v = vec_sub(b4v, b5v); \
- /* DST(6, b6 + b7) */ \
- sa8d6v = vec_add(b6v, b7v); \
- /* DST(7, b6 - b7) */ \
- sa8d7v = vec_sub(b6v, b7v); \
- }
- static int pixel_sa8d_8x8_core_altivec( uint8_t *pix1, intptr_t i_pix1,
- uint8_t *pix2, intptr_t i_pix2 )
- {
- int32_t i_satd=0;
- PREP_DIFF;
- vec_s16_t diff0v, diff1v, diff2v, diff3v, diff4v, diff5v, diff6v, diff7v;
- VEC_DIFF_H( pix1, i_pix1, pix2, i_pix2, 8, diff0v );
- VEC_DIFF_H( pix1, i_pix1, pix2, i_pix2, 8, diff1v );
- VEC_DIFF_H( pix1, i_pix1, pix2, i_pix2, 8, diff2v );
- VEC_DIFF_H( pix1, i_pix1, pix2, i_pix2, 8, diff3v );
- VEC_DIFF_H( pix1, i_pix1, pix2, i_pix2, 8, diff4v );
- VEC_DIFF_H( pix1, i_pix1, pix2, i_pix2, 8, diff5v );
- VEC_DIFF_H( pix1, i_pix1, pix2, i_pix2, 8, diff6v );
- VEC_DIFF_H( pix1, i_pix1, pix2, i_pix2, 8, diff7v );
- vec_s16_t sa8d0v, sa8d1v, sa8d2v, sa8d3v, sa8d4v, sa8d5v, sa8d6v, sa8d7v;
- SA8D_1D_ALTIVEC(diff0v, diff1v, diff2v, diff3v,
- diff4v, diff5v, diff6v, diff7v);
- VEC_TRANSPOSE_8(diff0v, diff1v, diff2v, diff3v,
- diff4v, diff5v, diff6v, diff7v,
- sa8d0v, sa8d1v, sa8d2v, sa8d3v,
- sa8d4v, sa8d5v, sa8d6v, sa8d7v );
- SA8D_1D_ALTIVEC(sa8d0v, sa8d1v, sa8d2v, sa8d3v,
- sa8d4v, sa8d5v, sa8d6v, sa8d7v );
- /* accumulation of the absolute value of all elements of the resulting bloc */
- vec_s16_t abs0v = VEC_ABS(sa8d0v);
- vec_s16_t abs1v = VEC_ABS(sa8d1v);
- vec_s16_t sum01v = vec_add(abs0v, abs1v);
- vec_s16_t abs2v = VEC_ABS(sa8d2v);
- vec_s16_t abs3v = VEC_ABS(sa8d3v);
- vec_s16_t sum23v = vec_add(abs2v, abs3v);
- vec_s16_t abs4v = VEC_ABS(sa8d4v);
- vec_s16_t abs5v = VEC_ABS(sa8d5v);
- vec_s16_t sum45v = vec_add(abs4v, abs5v);
- vec_s16_t abs6v = VEC_ABS(sa8d6v);
- vec_s16_t abs7v = VEC_ABS(sa8d7v);
- vec_s16_t sum67v = vec_add(abs6v, abs7v);
- vec_s16_t sum0123v = vec_add(sum01v, sum23v);
- vec_s16_t sum4567v = vec_add(sum45v, sum67v);
- vec_s32_t sumblocv;
- sumblocv = vec_sum4s(sum0123v, (vec_s32_t)zerov );
- sumblocv = vec_sum4s(sum4567v, sumblocv );
- sumblocv = vec_sums(sumblocv, (vec_s32_t)zerov );
- sumblocv = vec_splat(sumblocv, 3);
- vec_ste(sumblocv, 0, &i_satd);
- return i_satd;
- }
- static int pixel_sa8d_8x8_altivec( uint8_t *pix1, intptr_t i_pix1,
- uint8_t *pix2, intptr_t i_pix2 )
- {
- int32_t i_satd;
- i_satd = (pixel_sa8d_8x8_core_altivec( pix1, i_pix1, pix2, i_pix2 )+2)>>2;
- return i_satd;
- }
- static int pixel_sa8d_16x16_altivec( uint8_t *pix1, intptr_t i_pix1,
- uint8_t *pix2, intptr_t i_pix2 )
- {
- int32_t i_satd;
- i_satd = (pixel_sa8d_8x8_core_altivec( &pix1[0], i_pix1, &pix2[0], i_pix2 )
- + pixel_sa8d_8x8_core_altivec( &pix1[8], i_pix1, &pix2[8], i_pix2 )
- + pixel_sa8d_8x8_core_altivec( &pix1[8*i_pix1], i_pix1, &pix2[8*i_pix2], i_pix2 )
- + pixel_sa8d_8x8_core_altivec( &pix1[8*i_pix1+8], i_pix1, &pix2[8*i_pix2+8], i_pix2 ) +2)>>2;
- return i_satd;
- }
- #define HADAMARD4_ALTIVEC(d0,d1,d2,d3,s0,s1,s2,s3) {\
- vec_s16_t t0 = vec_add(s0, s1); \
- vec_s16_t t1 = vec_sub(s0, s1); \
- vec_s16_t t2 = vec_add(s2, s3); \
- vec_s16_t t3 = vec_sub(s2, s3); \
- d0 = vec_add(t0, t2); \
- d2 = vec_sub(t0, t2); \
- d1 = vec_add(t1, t3); \
- d3 = vec_sub(t1, t3); \
- }
- #ifdef WORDS_BIGENDIAN
- #define vec_perm_extend_s16(val, perm) (vec_s16_t)vec_perm(val, zero_u8v, perm)
- #else
- #define vec_perm_extend_s16(val, perm) (vec_s16_t)vec_perm(zero_u8v, val, perm)
- #endif
- #define VEC_LOAD_HIGH( p, num ) \
- vec_u8_t pix8_##num = vec_ld( stride*num, p ); \
- vec_s16_t pix16_s##num = vec_perm_extend_s16( pix8_##num, perm ); \
- vec_s16_t pix16_d##num;
- static uint64_t pixel_hadamard_ac_altivec( uint8_t *pix, intptr_t stride, const vec_u8_t perm )
- {
- ALIGNED_16( int32_t sum4_tab[4] );
- ALIGNED_16( int32_t sum8_tab[4] );
- LOAD_ZERO;
- VEC_LOAD_HIGH( pix, 0 );
- VEC_LOAD_HIGH( pix, 1 );
- VEC_LOAD_HIGH( pix, 2 );
- VEC_LOAD_HIGH( pix, 3 );
- HADAMARD4_ALTIVEC(pix16_d0,pix16_d1,pix16_d2,pix16_d3,
- pix16_s0,pix16_s1,pix16_s2,pix16_s3);
- VEC_LOAD_HIGH( pix, 4 );
- VEC_LOAD_HIGH( pix, 5 );
- VEC_LOAD_HIGH( pix, 6 );
- VEC_LOAD_HIGH( pix, 7 );
- HADAMARD4_ALTIVEC(pix16_d4,pix16_d5,pix16_d6,pix16_d7,
- pix16_s4,pix16_s5,pix16_s6,pix16_s7);
- VEC_TRANSPOSE_8(pix16_d0, pix16_d1, pix16_d2, pix16_d3,
- pix16_d4, pix16_d5, pix16_d6, pix16_d7,
- pix16_s0, pix16_s1, pix16_s2, pix16_s3,
- pix16_s4, pix16_s5, pix16_s6, pix16_s7);
- HADAMARD4_ALTIVEC(pix16_d0,pix16_d1,pix16_d2,pix16_d3,
- pix16_s0,pix16_s1,pix16_s2,pix16_s3);
- HADAMARD4_ALTIVEC(pix16_d4,pix16_d5,pix16_d6,pix16_d7,
- pix16_s4,pix16_s5,pix16_s6,pix16_s7);
- vec_u16_t addabs01 = vec_add( VEC_ABSOLUTE(pix16_d0), VEC_ABSOLUTE(pix16_d1) );
- vec_u16_t addabs23 = vec_add( VEC_ABSOLUTE(pix16_d2), VEC_ABSOLUTE(pix16_d3) );
- vec_u16_t addabs45 = vec_add( VEC_ABSOLUTE(pix16_d4), VEC_ABSOLUTE(pix16_d5) );
- vec_u16_t addabs67 = vec_add( VEC_ABSOLUTE(pix16_d6), VEC_ABSOLUTE(pix16_d7) );
- vec_u16_t sum4_v = vec_add(vec_add(addabs01, addabs23), vec_add(addabs45, addabs67));
- vec_ste(vec_sums(vec_sum4s((vec_s16_t)sum4_v, zero_s32v), zero_s32v), 12, sum4_tab);
- vec_s16_t tmpi0 = vec_add(pix16_d0, pix16_d4);
- vec_s16_t tmpi4 = vec_sub(pix16_d0, pix16_d4);
- vec_s16_t tmpi1 = vec_add(pix16_d1, pix16_d5);
- vec_s16_t tmpi5 = vec_sub(pix16_d1, pix16_d5);
- vec_s16_t tmpi2 = vec_add(pix16_d2, pix16_d6);
- vec_s16_t tmpi6 = vec_sub(pix16_d2, pix16_d6);
- vec_s16_t tmpi3 = vec_add(pix16_d3, pix16_d7);
- vec_s16_t tmpi7 = vec_sub(pix16_d3, pix16_d7);
- int sum4 = sum4_tab[3];
- VEC_TRANSPOSE_8(tmpi0, tmpi1, tmpi2, tmpi3,
- tmpi4, tmpi5, tmpi6, tmpi7,
- pix16_d0, pix16_d1, pix16_d2, pix16_d3,
- pix16_d4, pix16_d5, pix16_d6, pix16_d7);
- vec_u16_t addsum04 = vec_add( VEC_ABSOLUTE( vec_add(pix16_d0, pix16_d4) ),
- VEC_ABSOLUTE( vec_sub(pix16_d0, pix16_d4) ) );
- vec_u16_t addsum15 = vec_add( VEC_ABSOLUTE( vec_add(pix16_d1, pix16_d5) ),
- VEC_ABSOLUTE( vec_sub(pix16_d1, pix16_d5) ) );
- vec_u16_t addsum26 = vec_add( VEC_ABSOLUTE( vec_add(pix16_d2, pix16_d6) ),
- VEC_ABSOLUTE( vec_sub(pix16_d2, pix16_d6) ) );
- vec_u16_t addsum37 = vec_add( VEC_ABSOLUTE( vec_add(pix16_d3, pix16_d7) ),
- VEC_ABSOLUTE( vec_sub(pix16_d3, pix16_d7) ) );
- vec_u16_t sum8_v = vec_add( vec_add(addsum04, addsum15), vec_add(addsum26, addsum37) );
- vec_ste(vec_sums(vec_sum4s((vec_s16_t)sum8_v, zero_s32v), zero_s32v), 12, sum8_tab);
- int sum8 = sum8_tab[3];
- ALIGNED_16( int16_t tmp0_4_tab[8] );
- vec_ste(vec_add(pix16_d0, pix16_d4), 0, tmp0_4_tab);
- sum4 -= tmp0_4_tab[0];
- sum8 -= tmp0_4_tab[0];
- return ((uint64_t)sum8<<32) + sum4;
- }
- static const vec_u8_t hadamard_permtab[] =
- {
- CV(0x10,0x00,0x11,0x01, 0x12,0x02,0x13,0x03, /* pix = mod16 */
- 0x14,0x04,0x15,0x05, 0x16,0x06,0x17,0x07 ),
- CV(0x18,0x08,0x19,0x09, 0x1A,0x0A,0x1B,0x0B, /* pix = mod8 */
- 0x1C,0x0C,0x1D,0x0D, 0x1E,0x0E,0x1F,0x0F )
- };
- static uint64_t pixel_hadamard_ac_16x16_altivec( uint8_t *pix, intptr_t stride )
- {
- int idx = ((uintptr_t)pix & 8) >> 3;
- vec_u8_t permh = hadamard_permtab[idx];
- vec_u8_t perml = hadamard_permtab[!idx];
- uint64_t sum = pixel_hadamard_ac_altivec( pix, stride, permh );
- sum += pixel_hadamard_ac_altivec( pix+8, stride, perml );
- sum += pixel_hadamard_ac_altivec( pix+8*stride, stride, permh );
- sum += pixel_hadamard_ac_altivec( pix+8*stride+8, stride, perml );
- return ((sum>>34)<<32) + ((uint32_t)sum>>1);
- }
- static uint64_t pixel_hadamard_ac_16x8_altivec( uint8_t *pix, intptr_t stride )
- {
- int idx = ((uintptr_t)pix & 8) >> 3;
- vec_u8_t permh = hadamard_permtab[idx];
- vec_u8_t perml = hadamard_permtab[!idx];
- uint64_t sum = pixel_hadamard_ac_altivec( pix, stride, permh );
- sum += pixel_hadamard_ac_altivec( pix+8, stride, perml );
- return ((sum>>34)<<32) + ((uint32_t)sum>>1);
- }
- static uint64_t pixel_hadamard_ac_8x16_altivec( uint8_t *pix, intptr_t stride )
- {
- vec_u8_t perm = hadamard_permtab[ (((uintptr_t)pix & 8) >> 3) ];
- uint64_t sum = pixel_hadamard_ac_altivec( pix, stride, perm );
- sum += pixel_hadamard_ac_altivec( pix+8*stride, stride, perm );
- return ((sum>>34)<<32) + ((uint32_t)sum>>1);
- }
- static uint64_t pixel_hadamard_ac_8x8_altivec( uint8_t *pix, intptr_t stride )
- {
- vec_u8_t perm = hadamard_permtab[ (((uintptr_t)pix & 8) >> 3) ];
- uint64_t sum = pixel_hadamard_ac_altivec( pix, stride, perm );
- return ((sum>>34)<<32) + ((uint32_t)sum>>1);
- }
- /****************************************************************************
- * structural similarity metric
- ****************************************************************************/
- static void ssim_4x4x2_core_altivec( const uint8_t *pix1, intptr_t stride1,
- const uint8_t *pix2, intptr_t stride2,
- int sums[2][4] )
- {
- ALIGNED_16( int temp[4] );
- vec_u8_t pix1v, pix2v;
- vec_u32_t s1v, s2v, ssv, s12v;
- LOAD_ZERO;
- s1v = s2v = ssv = s12v = zero_u32v;
- for( int y = 0; y < 4; y++ )
- {
- pix1v = vec_vsx_ld( y*stride1, pix1 );
- pix2v = vec_vsx_ld( y*stride2, pix2 );
- s1v = vec_sum4s( pix1v, s1v );
- s2v = vec_sum4s( pix2v, s2v );
- ssv = vec_msum( pix1v, pix1v, ssv );
- ssv = vec_msum( pix2v, pix2v, ssv );
- s12v = vec_msum( pix1v, pix2v, s12v );
- }
- vec_st( (vec_s32_t)s1v, 0, temp );
- sums[0][0] = temp[0];
- sums[1][0] = temp[1];
- vec_st( (vec_s32_t)s2v, 0, temp );
- sums[0][1] = temp[0];
- sums[1][1] = temp[1];
- vec_st( (vec_s32_t)ssv, 0, temp );
- sums[0][2] = temp[0];
- sums[1][2] = temp[1];
- vec_st( (vec_s32_t)s12v, 0, temp );
- sums[0][3] = temp[0];
- sums[1][3] = temp[1];
- }
- #define SATD_X( size ) \
- static void pixel_satd_x3_##size##_altivec( uint8_t *fenc, uint8_t *pix0, uint8_t *pix1, uint8_t *pix2,\
- intptr_t i_stride, int scores[3] )\
- {\
- scores[0] = pixel_satd_##size##_altivec( fenc, FENC_STRIDE, pix0, i_stride );\
- scores[1] = pixel_satd_##size##_altivec( fenc, FENC_STRIDE, pix1, i_stride );\
- scores[2] = pixel_satd_##size##_altivec( fenc, FENC_STRIDE, pix2, i_stride );\
- }\
- static void pixel_satd_x4_##size##_altivec( uint8_t *fenc, uint8_t *pix0, uint8_t *pix1, uint8_t *pix2,\
- uint8_t *pix3, intptr_t i_stride, int scores[4] )\
- {\
- scores[0] = pixel_satd_##size##_altivec( fenc, FENC_STRIDE, pix0, i_stride );\
- scores[1] = pixel_satd_##size##_altivec( fenc, FENC_STRIDE, pix1, i_stride );\
- scores[2] = pixel_satd_##size##_altivec( fenc, FENC_STRIDE, pix2, i_stride );\
- scores[3] = pixel_satd_##size##_altivec( fenc, FENC_STRIDE, pix3, i_stride );\
- }
- SATD_X( 16x16 )\
- SATD_X( 16x8 )\
- SATD_X( 8x16 )\
- SATD_X( 8x8 )\
- SATD_X( 8x4 )\
- SATD_X( 4x8 )\
- SATD_X( 4x4 )
- #define INTRA_MBCMP_8x8( mbcmp )\
- static void intra_##mbcmp##_x3_8x8_altivec( uint8_t *fenc, uint8_t edge[36], int res[3] )\
- {\
- ALIGNED_8( uint8_t pix[8*FDEC_STRIDE] );\
- x264_predict_8x8_v_c( pix, edge );\
- res[0] = pixel_##mbcmp##_8x8_altivec( pix, FDEC_STRIDE, fenc, FENC_STRIDE );\
- x264_predict_8x8_h_c( pix, edge );\
- res[1] = pixel_##mbcmp##_8x8_altivec( pix, FDEC_STRIDE, fenc, FENC_STRIDE );\
- x264_predict_8x8_dc_c( pix, edge );\
- res[2] = pixel_##mbcmp##_8x8_altivec( pix, FDEC_STRIDE, fenc, FENC_STRIDE );\
- }
- INTRA_MBCMP_8x8(sad)
- INTRA_MBCMP_8x8(sa8d)
- #define INTRA_MBCMP( mbcmp, size, pred1, pred2, pred3, chroma )\
- static void intra_##mbcmp##_x3_##size##x##size##chroma##_altivec( uint8_t *fenc, uint8_t *fdec, int res[3] )\
- {\
- x264_predict_##size##x##size##chroma##_##pred1##_c( fdec );\
- res[0] = pixel_##mbcmp##_##size##x##size##_altivec( fdec, FDEC_STRIDE, fenc, FENC_STRIDE );\
- x264_predict_##size##x##size##chroma##_##pred2##_c( fdec );\
- res[1] = pixel_##mbcmp##_##size##x##size##_altivec( fdec, FDEC_STRIDE, fenc, FENC_STRIDE );\
- x264_predict_##size##x##size##chroma##_##pred3##_c( fdec );\
- res[2] = pixel_##mbcmp##_##size##x##size##_altivec( fdec, FDEC_STRIDE, fenc, FENC_STRIDE );\
- }
- INTRA_MBCMP(satd, 4, v, h, dc, )
- INTRA_MBCMP(sad, 8, dc, h, v, c )
- INTRA_MBCMP(satd, 8, dc, h, v, c )
- INTRA_MBCMP(sad, 16, v, h, dc, )
- INTRA_MBCMP(satd, 16, v, h, dc, )
- #endif // !HIGH_BIT_DEPTH
- /****************************************************************************
- * x264_pixel_init:
- ****************************************************************************/
- void x264_pixel_init_altivec( x264_pixel_function_t *pixf )
- {
- #if !HIGH_BIT_DEPTH
- pixf->sad[PIXEL_16x16] = pixel_sad_16x16_altivec;
- pixf->sad[PIXEL_8x16] = pixel_sad_8x16_altivec;
- pixf->sad[PIXEL_16x8] = pixel_sad_16x8_altivec;
- pixf->sad[PIXEL_8x8] = pixel_sad_8x8_altivec;
- pixf->sad_x3[PIXEL_16x16] = pixel_sad_x3_16x16_altivec;
- pixf->sad_x3[PIXEL_8x16] = pixel_sad_x3_8x16_altivec;
- pixf->sad_x3[PIXEL_16x8] = pixel_sad_x3_16x8_altivec;
- pixf->sad_x3[PIXEL_8x8] = pixel_sad_x3_8x8_altivec;
- pixf->sad_x4[PIXEL_16x16] = pixel_sad_x4_16x16_altivec;
- pixf->sad_x4[PIXEL_8x16] = pixel_sad_x4_8x16_altivec;
- pixf->sad_x4[PIXEL_16x8] = pixel_sad_x4_16x8_altivec;
- pixf->sad_x4[PIXEL_8x8] = pixel_sad_x4_8x8_altivec;
- pixf->satd[PIXEL_16x16] = pixel_satd_16x16_altivec;
- pixf->satd[PIXEL_8x16] = pixel_satd_8x16_altivec;
- pixf->satd[PIXEL_16x8] = pixel_satd_16x8_altivec;
- pixf->satd[PIXEL_8x8] = pixel_satd_8x8_altivec;
- pixf->satd[PIXEL_8x4] = pixel_satd_8x4_altivec;
- pixf->satd[PIXEL_4x8] = pixel_satd_4x8_altivec;
- pixf->satd[PIXEL_4x4] = pixel_satd_4x4_altivec;
- pixf->satd_x3[PIXEL_16x16] = pixel_satd_x3_16x16_altivec;
- pixf->satd_x3[PIXEL_8x16] = pixel_satd_x3_8x16_altivec;
- pixf->satd_x3[PIXEL_16x8] = pixel_satd_x3_16x8_altivec;
- pixf->satd_x3[PIXEL_8x8] = pixel_satd_x3_8x8_altivec;
- pixf->satd_x3[PIXEL_8x4] = pixel_satd_x3_8x4_altivec;
- pixf->satd_x3[PIXEL_4x8] = pixel_satd_x3_4x8_altivec;
- pixf->satd_x3[PIXEL_4x4] = pixel_satd_x3_4x4_altivec;
- pixf->satd_x4[PIXEL_16x16] = pixel_satd_x4_16x16_altivec;
- pixf->satd_x4[PIXEL_8x16] = pixel_satd_x4_8x16_altivec;
- pixf->satd_x4[PIXEL_16x8] = pixel_satd_x4_16x8_altivec;
- pixf->satd_x4[PIXEL_8x8] = pixel_satd_x4_8x8_altivec;
- pixf->satd_x4[PIXEL_8x4] = pixel_satd_x4_8x4_altivec;
- pixf->satd_x4[PIXEL_4x8] = pixel_satd_x4_4x8_altivec;
- pixf->satd_x4[PIXEL_4x4] = pixel_satd_x4_4x4_altivec;
- pixf->intra_sad_x3_8x8 = intra_sad_x3_8x8_altivec;
- pixf->intra_sad_x3_8x8c = intra_sad_x3_8x8c_altivec;
- pixf->intra_sad_x3_16x16 = intra_sad_x3_16x16_altivec;
- pixf->intra_satd_x3_4x4 = intra_satd_x3_4x4_altivec;
- pixf->intra_satd_x3_8x8c = intra_satd_x3_8x8c_altivec;
- pixf->intra_satd_x3_16x16 = intra_satd_x3_16x16_altivec;
- pixf->ssd[PIXEL_16x16] = pixel_ssd_16x16_altivec;
- pixf->ssd[PIXEL_8x8] = pixel_ssd_8x8_altivec;
- pixf->sa8d[PIXEL_16x16] = pixel_sa8d_16x16_altivec;
- pixf->sa8d[PIXEL_8x8] = pixel_sa8d_8x8_altivec;
- pixf->intra_sa8d_x3_8x8 = intra_sa8d_x3_8x8_altivec;
- pixf->var[PIXEL_16x16] = pixel_var_16x16_altivec;
- pixf->var[PIXEL_8x8] = pixel_var_8x8_altivec;
- pixf->hadamard_ac[PIXEL_16x16] = pixel_hadamard_ac_16x16_altivec;
- pixf->hadamard_ac[PIXEL_16x8] = pixel_hadamard_ac_16x8_altivec;
- pixf->hadamard_ac[PIXEL_8x16] = pixel_hadamard_ac_8x16_altivec;
- pixf->hadamard_ac[PIXEL_8x8] = pixel_hadamard_ac_8x8_altivec;
- pixf->ssim_4x4x2_core = ssim_4x4x2_core_altivec;
- #endif // !HIGH_BIT_DEPTH
- }
|