diff -bBru src-ori/bitstream/bitstream.c src/bitstream/bitstream.c --- src-ori/bitstream/bitstream.c Wed Jan 12 10:57:45 2005 +++ src/bitstream/bitstream.c Wed Jan 12 12:39:29 2005 @@ -24,13 +24,234 @@ * ****************************************************************************/ -#include -#include +#include +#include "../global_all_dll.h" + BitstreamInit(Bitstream * const bs, + void *const bitstream, + uint32_t length) + { + + uint32_t tmp; + size_t bitpos; + ptr_t adjbitstream = (ptr_t)bitstream; + + /* + * Start the stream on a uint32_t boundary, by rounding down to the + * previous uint32_t and skipping the intervening bytes. + */ + bitpos = ((sizeof(uint32_t)-1) & (size_t)bitstream); + adjbitstream = adjbitstream - bitpos; + bs->start = bs->tail = (uint32_t *) adjbitstream; + + tmp = *bs->start; +#ifndef ARCH_IS_BIG_ENDIAN + BSWAP(tmp); +#endif + bs->bufa = tmp; + + tmp = *(bs->start + 1); +#ifndef ARCH_IS_BIG_ENDIAN + BSWAP(tmp); +#endif + bs->bufb = tmp; + + bs->buf = 0; + bs->pos = bs->initpos = bitpos*8; + bs->length = length; +} + BitstreamReset(Bitstream * const bs) + { + + uint32_t tmp; + + bs->tail = bs->start; + + tmp = *bs->start; +#ifndef ARCH_IS_BIG_ENDIAN + BSWAP(tmp); +#endif + bs->bufa = tmp; + + tmp = *(bs->start + 1); +#ifndef ARCH_IS_BIG_ENDIAN + BSWAP(tmp); +#endif + bs->bufb = tmp; + + bs->buf = 0; + bs->pos = bs->initpos; +} + BitstreamShowBits(Bitstream * const bs, + const uint32_t bits) + { + + int nbit = (bits + bs->pos) - 32; + + if (nbit > 0) { + return ((bs->bufa & (0xffffffff >> bs->pos)) << nbit) | (bs-> + bufb >> (32 - + nbit)); + } else { + return (bs->bufa & (0xffffffff >> bs->pos)) >> (32 - bs->pos - bits); + } +} + void BitstreamSkip(Bitstream * const bs, + const uint32_t bits) + { + + bs->pos += bits; + + if (bs->pos >= 32) { + uint32_t tmp; + + bs->bufa = bs->bufb; + tmp = *((uint32_t *) bs->tail + 2); +#ifndef ARCH_IS_BIG_ENDIAN + BSWAP(tmp); +#endif + bs->bufb = tmp; + bs->tail++; + bs->pos -= 32; + } +} + uint32_t BitstreamNumBitsToByteAlign(Bitstream *bs) + { + + uint32_t n = (32 - bs->pos) % 8; + return n == 0 ? 8 : n; +} + uint32_t BitstreamShowBitsFromByteAlign(Bitstream *bs, int bits) + { + + int bspos = bs->pos + BitstreamNumBitsToByteAlign(bs); + int nbit = (bits + bspos) - 32; + + if (bspos >= 32) { + return bs->bufb >> (32 - nbit); + } else if (nbit > 0) { + return ((bs->bufa & (0xffffffff >> bspos)) << nbit) | (bs-> + bufb >> (32 - + nbit)); + } else { + return (bs->bufa & (0xffffffff >> bspos)) >> (32 - bspos - bits); + } + +} + void BitstreamByteAlign(Bitstream * const bs) + { + + uint32_t remainder = bs->pos % 8; + + if (remainder) { + BitstreamSkip(bs, 8 - remainder); + } +} + BitstreamPos(const Bitstream * const bs) + { + + return((uint32_t)(8*((ptr_t)bs->tail - (ptr_t)bs->start) + bs->pos - bs->initpos)); +} + BitstreamLength(Bitstream * const bs) + { + + uint32_t len = (uint32_t)((ptr_t)bs->tail - (ptr_t)bs->start); + + if (bs->pos) { + uint32_t b = bs->buf; + +#ifndef ARCH_IS_BIG_ENDIAN + BSWAP(b); +#endif + *bs->tail = b; + + len += (bs->pos + 7) / 8; + } + + /* initpos is always on a byte boundary */ + if (bs->initpos) + len -= bs->initpos/8; + + return len; +} + BitstreamForward(Bitstream * const bs, + const uint32_t bits) + { + + bs->pos += bits; + + if (bs->pos >= 32) { + uint32_t b = bs->buf; + +#ifndef ARCH_IS_BIG_ENDIAN + BSWAP(b); +#endif + *bs->tail++ = b; + bs->buf = 0; + bs->pos -= 32; + } +} + BitstreamGetBits(Bitstream * const bs, + const uint32_t n) + { + + uint32_t ret = BitstreamShowBits(bs, n); + + BitstreamSkip(bs, n); + return ret; +} + BitstreamGetBit(Bitstream * const bs) + { + + return BitstreamGetBits(bs, 1); +} + BitstreamPutBit(Bitstream * const bs, + const uint32_t bit) + { + + if (bit) + bs->buf |= (0x80000000 >> bs->pos); + + BitstreamForward(bs, 1); +} + BitstreamPutBits(Bitstream * const bs, + const uint32_t value, + const uint32_t size) + { + + uint32_t shift = 32 - bs->pos - size; + + if (shift <= 32) { + bs->buf |= value << shift; + BitstreamForward(bs, size); + } else { + uint32_t remainder; + + shift = size - (32 - bs->pos); + bs->buf |= value >> shift; + BitstreamForward(bs, size - shift); + remainder = shift; + + shift = 32 - shift; + + bs->buf |= value << shift; + BitstreamForward(bs, remainder); + } +} + BitstreamPad(Bitstream * const bs) + { + + int bits = 8 - (bs->pos % 8); + if (bits < 8) + BitstreamPutBits(bs, stuffing_codes[bits - 1], bits); +} + BitstreamPadAlways(Bitstream * const bs) + { + + int bits = 8 - (bs->pos % 8); + BitstreamPutBits(bs, stuffing_codes[bits - 1], bits); +} + -#include "bitstream.h" -#include "zigzag.h" -#include "../quant/quant_matrix.h" -#include "mbcoding.h" /* multiple definitions when merging all .h files into global_all_dll.h */ @@ -48,7 +269,7 @@ static const uint32_t intra_dc_threshold_table[] = { - 32, /* never use */ + 32, 13, 15, 17, @@ -378,6 +599,15 @@ #define VIDOBJ_START_CODE_MASK 0x0000001f #define VIDOBJLAY_START_CODE_MASK 0x0000000f +#if 0 + dec->time_inc_resolution--; +#endif +#if 0 + dec->time_inc_bits = 0; +#endif +#if 0 + DPRINTF("*** WARNING: no vop_start_code found"); +#endif int BitstreamReadHeaders(Bitstream * bs, DECODER * dec, @@ -545,16 +775,10 @@ dec->time_inc_resolution = BitstreamGetBits(bs, 16); /* vop_time_increment_resolution */ DPRINTF(XVID_DEBUG_HEADER,"vop_time_increment_resolution %i\n", dec->time_inc_resolution); -#if 0 - dec->time_inc_resolution--; -#endif if (dec->time_inc_resolution > 0) { dec->time_inc_bits = MAX(log2bin(dec->time_inc_resolution-1), 1); } else { -#if 0 - dec->time_inc_bits = 0; -#endif /* for "old" xvid compatibility, set time_inc_bits = 1 */ dec->time_inc_bits = 1; } @@ -1039,9 +1263,6 @@ } } -#if 0 - DPRINTF("*** WARNING: no vop_start_code found"); -#endif return -1; /* ignore it */ } @@ -1070,6 +1291,10 @@ /* write vol header */ +#if 0 + BitstreamPad(bs); +#endif +#define DIVX5_ID "DivX999b000p" void BitstreamWriteVolHeader(Bitstream * const bs, const MBParam * pParam, @@ -1096,9 +1321,6 @@ } /* visual_object_sequence_start_code */ -#if 0 - BitstreamPad(bs); -#endif /* * no padding here, anymore. You have to make sure that you are @@ -1257,7 +1479,6 @@ BitstreamPadAlways(bs); /* next_start_code(); */ /* fake divx5 id, to ensure compatibility with divx5 decoder */ -#define DIVX5_ID "DivX999b000p" if (pParam->max_bframes > 0 && (pParam->global_flags & XVID_GLOBAL_PACKED)) { BitstreamWriteUserData(bs, DIVX5_ID, strlen(DIVX5_ID)); } @@ -1278,6 +1499,25 @@ /* write vop header */ +#if 0 + BitstreamPad(bs); +#endif +#if 0 + DPRINTF(XVID_DEBUG_HEADER, "coding_type = %i\n", frame->coding_type); +#endif +#if 0 + DPRINTF("[%i:%i] %c", + frame->seconds, frame->ticks, + frame->coding_type == I_VOP ? 'I' : + frame->coding_type == P_VOP ? 'P' : + frame->coding_type == S_VOP ? 'S' : 'B'); +#endif +#if 0 + BitstreamPadAlways(bs); /* next_start_code() */ +#endif +#if 0 + DPRINTF(XVID_DEBUG_HEADER, "quant = %i\n", quant); +#endif void BitstreamWriteVopHeader( Bitstream * const bs, @@ -1288,9 +1528,6 @@ { uint32_t i; -#if 0 - BitstreamPad(bs); -#endif /* * no padding here, anymore. You have to make sure that you are @@ -1300,9 +1537,6 @@ BitstreamPutBits(bs, VOP_START_CODE, 32); BitstreamPutBits(bs, frame->coding_type, 2); -#if 0 - DPRINTF(XVID_DEBUG_HEADER, "coding_type = %i\n", frame->coding_type); -#endif for (i = 0; i < frame->seconds; i++) { BitstreamPutBit(bs, 1); @@ -1313,21 +1547,11 @@ /* time_increment: value=nth_of_sec, nbits = log2(resolution) */ BitstreamPutBits(bs, frame->ticks, MAX(log2bin(pParam->fbase-1), 1)); -#if 0 - DPRINTF("[%i:%i] %c", - frame->seconds, frame->ticks, - frame->coding_type == I_VOP ? 'I' : - frame->coding_type == P_VOP ? 'P' : - frame->coding_type == S_VOP ? 'S' : 'B'); -#endif WRITE_MARKER(); if (!vop_coded) { BitstreamPutBits(bs, 0, 1); -#if 0 - BitstreamPadAlways(bs); /* next_start_code() */ -#endif /* NB: It's up to the function caller to write the next_start_code(). * At the moment encoder.c respects that requisite because a VOP * always ends with a next_start_code either if it's coded or not @@ -1375,9 +1599,6 @@ } -#if 0 - DPRINTF(XVID_DEBUG_HEADER, "quant = %i\n", quant); -#endif BitstreamPutBits(bs, quant, 5); /* quantizer */ diff -bBru src-ori/bitstream/bitstream.h src/bitstream/bitstream.h --- src-ori/bitstream/bitstream.h Wed Jan 12 10:32:43 2005 +++ src/bitstream/bitstream.h Wed Jan 12 12:37:47 2005 @@ -152,158 +152,53 @@ /* initialise bitstream structure */ -static void __inline -BitstreamInit(Bitstream * const bs, +extern BitstreamInit(Bitstream * const bs, void *const bitstream, uint32_t length) -{ - uint32_t tmp; - size_t bitpos; - ptr_t adjbitstream = (ptr_t)bitstream; - - /* - * Start the stream on a uint32_t boundary, by rounding down to the - * previous uint32_t and skipping the intervening bytes. - */ - bitpos = ((sizeof(uint32_t)-1) & (size_t)bitstream); - adjbitstream = adjbitstream - bitpos; - bs->start = bs->tail = (uint32_t *) adjbitstream; - - tmp = *bs->start; -#ifndef ARCH_IS_BIG_ENDIAN - BSWAP(tmp); -#endif - bs->bufa = tmp; - - tmp = *(bs->start + 1); -#ifndef ARCH_IS_BIG_ENDIAN - BSWAP(tmp); -#endif - bs->bufb = tmp; - - bs->buf = 0; - bs->pos = bs->initpos = bitpos*8; - bs->length = length; -} +; /* reset bitstream state */ -static void __inline -BitstreamReset(Bitstream * const bs) -{ - uint32_t tmp; - - bs->tail = bs->start; - - tmp = *bs->start; -#ifndef ARCH_IS_BIG_ENDIAN - BSWAP(tmp); -#endif - bs->bufa = tmp; - - tmp = *(bs->start + 1); -#ifndef ARCH_IS_BIG_ENDIAN - BSWAP(tmp); -#endif - bs->bufb = tmp; - - bs->buf = 0; - bs->pos = bs->initpos; -} +extern BitstreamReset(Bitstream * const bs) +; /* reads n bits from bitstream without changing the stream pos */ -static uint32_t __inline -BitstreamShowBits(Bitstream * const bs, +extern BitstreamShowBits(Bitstream * const bs, const uint32_t bits) -{ - int nbit = (bits + bs->pos) - 32; - - if (nbit > 0) { - return ((bs->bufa & (0xffffffff >> bs->pos)) << nbit) | (bs-> - bufb >> (32 - - nbit)); - } else { - return (bs->bufa & (0xffffffff >> bs->pos)) >> (32 - bs->pos - bits); - } -} +; /* skip n bits forward in bitstream */ -static __inline void -BitstreamSkip(Bitstream * const bs, +extern void BitstreamSkip(Bitstream * const bs, const uint32_t bits) -{ - bs->pos += bits; - - if (bs->pos >= 32) { - uint32_t tmp; - - bs->bufa = bs->bufb; - tmp = *((uint32_t *) bs->tail + 2); -#ifndef ARCH_IS_BIG_ENDIAN - BSWAP(tmp); -#endif - bs->bufb = tmp; - bs->tail++; - bs->pos -= 32; - } -} +; /* number of bits to next byte alignment */ -static __inline uint32_t -BitstreamNumBitsToByteAlign(Bitstream *bs) -{ - uint32_t n = (32 - bs->pos) % 8; - return n == 0 ? 8 : n; -} +extern uint32_t BitstreamNumBitsToByteAlign(Bitstream *bs) +; /* show nbits from next byte alignment */ -static __inline uint32_t -BitstreamShowBitsFromByteAlign(Bitstream *bs, int bits) -{ - int bspos = bs->pos + BitstreamNumBitsToByteAlign(bs); - int nbit = (bits + bspos) - 32; - - if (bspos >= 32) { - return bs->bufb >> (32 - nbit); - } else if (nbit > 0) { - return ((bs->bufa & (0xffffffff >> bspos)) << nbit) | (bs-> - bufb >> (32 - - nbit)); - } else { - return (bs->bufa & (0xffffffff >> bspos)) >> (32 - bspos - bits); - } - -} +extern uint32_t BitstreamShowBitsFromByteAlign(Bitstream *bs, int bits) +; /* move forward to the next byte boundary */ -static __inline void -BitstreamByteAlign(Bitstream * const bs) -{ - uint32_t remainder = bs->pos % 8; - - if (remainder) { - BitstreamSkip(bs, 8 - remainder); - } -} +extern void BitstreamByteAlign(Bitstream * const bs) +; /* bitstream length (unit bits) */ -static uint32_t __inline -BitstreamPos(const Bitstream * const bs) -{ - return((uint32_t)(8*((ptr_t)bs->tail - (ptr_t)bs->start) + bs->pos - bs->initpos)); -} +extern BitstreamPos(const Bitstream * const bs) +; /* @@ -311,111 +206,42 @@ * NOTE: assumes no futher bitstream functions will be called. */ -static uint32_t __inline -BitstreamLength(Bitstream * const bs) -{ - uint32_t len = (uint32_t)((ptr_t)bs->tail - (ptr_t)bs->start); - - if (bs->pos) { - uint32_t b = bs->buf; - -#ifndef ARCH_IS_BIG_ENDIAN - BSWAP(b); -#endif - *bs->tail = b; - - len += (bs->pos + 7) / 8; - } - - /* initpos is always on a byte boundary */ - if (bs->initpos) - len -= bs->initpos/8; - - return len; -} +extern BitstreamLength(Bitstream * const bs) +; /* move bitstream position forward by n bits and write out buffer if needed */ -static void __inline -BitstreamForward(Bitstream * const bs, +extern BitstreamForward(Bitstream * const bs, const uint32_t bits) -{ - bs->pos += bits; - - if (bs->pos >= 32) { - uint32_t b = bs->buf; - -#ifndef ARCH_IS_BIG_ENDIAN - BSWAP(b); -#endif - *bs->tail++ = b; - bs->buf = 0; - bs->pos -= 32; - } -} +; /* read n bits from bitstream */ -static uint32_t __inline -BitstreamGetBits(Bitstream * const bs, +extern BitstreamGetBits(Bitstream * const bs, const uint32_t n) -{ - uint32_t ret = BitstreamShowBits(bs, n); - - BitstreamSkip(bs, n); - return ret; -} +; /* read single bit from bitstream */ -static uint32_t __inline -BitstreamGetBit(Bitstream * const bs) -{ - return BitstreamGetBits(bs, 1); -} +extern BitstreamGetBit(Bitstream * const bs) +; /* write single bit to bitstream */ -static void __inline -BitstreamPutBit(Bitstream * const bs, +extern BitstreamPutBit(Bitstream * const bs, const uint32_t bit) -{ - if (bit) - bs->buf |= (0x80000000 >> bs->pos); - - BitstreamForward(bs, 1); -} +; /* write n bits to bitstream */ -static void __inline -BitstreamPutBits(Bitstream * const bs, +extern BitstreamPutBits(Bitstream * const bs, const uint32_t value, const uint32_t size) -{ - uint32_t shift = 32 - bs->pos - size; - - if (shift <= 32) { - bs->buf |= value << shift; - BitstreamForward(bs, size); - } else { - uint32_t remainder; - - shift = size - (32 - bs->pos); - bs->buf |= value >> shift; - BitstreamForward(bs, size - shift); - remainder = shift; - - shift = 32 - shift; - - bs->buf |= value << shift; - BitstreamForward(bs, remainder); - } -} +; static const int stuffing_codes[8] = { @@ -432,13 +258,8 @@ /* pad bitstream to the next byte boundary */ -static void __inline -BitstreamPad(Bitstream * const bs) -{ - int bits = 8 - (bs->pos % 8); - if (bits < 8) - BitstreamPutBits(bs, stuffing_codes[bits - 1], bits); -} +extern BitstreamPad(Bitstream * const bs) +; /* @@ -446,11 +267,7 @@ * alway pad: even if currently at the byte boundary */ -static void __inline -BitstreamPadAlways(Bitstream * const bs) -{ - int bits = 8 - (bs->pos % 8); - BitstreamPutBits(bs, stuffing_codes[bits - 1], bits); -} +extern BitstreamPadAlways(Bitstream * const bs) +; #endif /* _BITSTREAM_H_ */ diff -bBru src-ori/bitstream/cbp.c src/bitstream/cbp.c --- src-ori/bitstream/cbp.c Wed Jan 12 10:32:43 2005 +++ src/bitstream/cbp.c Wed Jan 12 12:38:38 2005 @@ -24,10 +24,9 @@ * ****************************************************************************/ -#include "../portab.h" -#include "cbp.h" +#include +#include "../global_all_dll.h" -cbpFuncPtr calc_cbp; /* * Returns a field of bits that indicates non zero ac blocks diff -bBru src-ori/bitstream/cbp.h src/bitstream/cbp.h --- src-ori/bitstream/cbp.h Wed Jan 12 10:32:43 2005 +++ src/bitstream/cbp.h Wed Jan 12 12:37:42 2005 @@ -33,7 +33,7 @@ typedef cbpFunc *cbpFuncPtr; -extern cbpFuncPtr calc_cbp; + extern cbpFunc calc_cbp_c; extern cbpFunc calc_cbp_plain; diff -bBru src-ori/bitstream/mbcoding.c src/bitstream/mbcoding.c --- src-ori/bitstream/mbcoding.c Wed Jan 12 10:32:43 2005 +++ src/bitstream/mbcoding.c Wed Jan 12 12:38:58 2005 @@ -23,38 +23,32 @@ * ****************************************************************************/ -#include -#include -#include - -#include "../portab.h" -#include "../global.h" -#include "bitstream.h" -#include "zigzag.h" -#include "vlc_codes.h" -#include "mbcoding.h" +#include +#include "../global_all_dll.h" + void MBSkip(Bitstream * bs) + { + + BitstreamPutBit(bs, 1); /* not coded */ +} + + -#include "../utils/mbfunctions.h" #define LEVELOFFSET 32 -/* Initialized once during xvid_global call - * RO access is thread safe */ -static REVERSE_EVENT DCT3D[2][4096]; -static VLC coeff_VLC[2][2][64][64]; /* not really MB related, but VLCs are only available here */ -void bs_put_spritetrajectory(Bitstream * bs, const int val) -{ - const int code = sprite_trajectory_code[val+16384].code; - const int len = sprite_trajectory_code[val+16384].len; - const int code2 = sprite_trajectory_len[len].code; - const int len2 = sprite_trajectory_len[len].len; - #if 0 printf("GMC=%d Code/Len = %d / %d ",val, code,len); printf("Code2 / Len2 = %d / %d \n",code2,len2); #endif +void bs_put_spritetrajectory(Bitstream * bs, const int val) +{ + const int code = (((struct global_all_dll *)Dll::Tls())->sprite_trajectory_code)[val+16384].code; + const int len = (((struct global_all_dll *)Dll::Tls())->sprite_trajectory_code)[val+16384].len; + const int code2 = (((struct global_all_dll *)Dll::Tls())->sprite_trajectory_len)[len].code; + const int len2 = (((struct global_all_dll *)Dll::Tls())->sprite_trajectory_len)[len].len; + BitstreamPutBits(bs, code2, len2); if (len) BitstreamPutBits(bs, code, len); @@ -65,9 +59,9 @@ int i; for (i = 0; i < 12; i++) { - if (BitstreamShowBits(bs, sprite_trajectory_len[i].len) == sprite_trajectory_len[i].code) + if (BitstreamShowBits(bs, (((struct global_all_dll *)Dll::Tls())->sprite_trajectory_len)[i].len) == (((struct global_all_dll *)Dll::Tls())->sprite_trajectory_len)[i].code) { - BitstreamSkip(bs, sprite_trajectory_len[i].len); + BitstreamSkip(bs, (((struct global_all_dll *)Dll::Tls())->sprite_trajectory_len)[i].len); return i; } } @@ -82,14 +76,14 @@ for (intra = 0; intra < 2; intra++) for (i = 0; i < 4096; i++) - DCT3D[intra][i].event.level = 0; + (((struct global_all_dll *)Dll::Tls())->DCT3D)[intra][i].event.level = 0; for (intra = 0; intra < 2; intra++) { for (last = 0; last < 2; last++) { for (run = 0; run < 63 + last; run++) { for (level = 0; level < (uint32_t)(32 << intra); level++) { offset = !intra * LEVELOFFSET; - coeff_VLC[intra][last][level + offset][run].len = 128; + (((struct global_all_dll *)Dll::Tls())->coeff_VLC)[intra][last][level + offset][run].len = 128; } } } @@ -100,19 +94,19 @@ offset = !intra * LEVELOFFSET; for (j = 0; j < (uint32_t)(1 << (12 - coeff_tab[intra][i].vlc.len)); j++) { - DCT3D[intra][(coeff_tab[intra][i].vlc.code << (12 - coeff_tab[intra][i].vlc.len)) | j].len = coeff_tab[intra][i].vlc.len; - DCT3D[intra][(coeff_tab[intra][i].vlc.code << (12 - coeff_tab[intra][i].vlc.len)) | j].event = coeff_tab[intra][i].event; + (((struct global_all_dll *)Dll::Tls())->DCT3D)[intra][(coeff_tab[intra][i].vlc.code << (12 - coeff_tab[intra][i].vlc.len)) | j].len = coeff_tab[intra][i].vlc.len; + (((struct global_all_dll *)Dll::Tls())->DCT3D)[intra][(coeff_tab[intra][i].vlc.code << (12 - coeff_tab[intra][i].vlc.len)) | j].event = coeff_tab[intra][i].event; } - coeff_VLC[intra][coeff_tab[intra][i].event.last][coeff_tab[intra][i].event.level + offset][coeff_tab[intra][i].event.run].code + (((struct global_all_dll *)Dll::Tls())->coeff_VLC)[intra][coeff_tab[intra][i].event.last][coeff_tab[intra][i].event.level + offset][coeff_tab[intra][i].event.run].code = coeff_tab[intra][i].vlc.code << 1; - coeff_VLC[intra][coeff_tab[intra][i].event.last][coeff_tab[intra][i].event.level + offset][coeff_tab[intra][i].event.run].len + (((struct global_all_dll *)Dll::Tls())->coeff_VLC)[intra][coeff_tab[intra][i].event.last][coeff_tab[intra][i].event.level + offset][coeff_tab[intra][i].event.run].len = coeff_tab[intra][i].vlc.len + 1; if (!intra) { - coeff_VLC[intra][coeff_tab[intra][i].event.last][offset - coeff_tab[intra][i].event.level][coeff_tab[intra][i].event.run].code + (((struct global_all_dll *)Dll::Tls())->coeff_VLC)[intra][coeff_tab[intra][i].event.last][offset - coeff_tab[intra][i].event.level][coeff_tab[intra][i].event.run].code = (coeff_tab[intra][i].vlc.code << 1) | 1; - coeff_VLC[intra][coeff_tab[intra][i].event.last][offset - coeff_tab[intra][i].event.level][coeff_tab[intra][i].event.run].len + (((struct global_all_dll *)Dll::Tls())->coeff_VLC)[intra][coeff_tab[intra][i].event.last][offset - coeff_tab[intra][i].event.level][coeff_tab[intra][i].event.run].len = coeff_tab[intra][i].vlc.len + 1; } } @@ -141,36 +135,36 @@ level_esc = level; } else { if (!intra) { - coeff_VLC[intra][last][level + offset][run].code + (((struct global_all_dll *)Dll::Tls())->coeff_VLC)[intra][last][level + offset][run].code = (ESCAPE3 << 21) | (last << 20) | (run << 14) | (1 << 13) | ((level & 0xfff) << 1) | 1; - coeff_VLC[intra][last][level + offset][run].len = 30; - coeff_VLC[intra][last][offset - level][run].code + (((struct global_all_dll *)Dll::Tls())->coeff_VLC)[intra][last][level + offset][run].len = 30; + (((struct global_all_dll *)Dll::Tls())->coeff_VLC)[intra][last][offset - level][run].code = (ESCAPE3 << 21) | (last << 20) | (run << 14) | (1 << 13) | ((-(int32_t)level & 0xfff) << 1) | 1; - coeff_VLC[intra][last][offset - level][run].len = 30; + (((struct global_all_dll *)Dll::Tls())->coeff_VLC)[intra][last][offset - level][run].len = 30; } continue; } } - coeff_VLC[intra][last][level + offset][run].code - = (escape << coeff_VLC[intra][last][level_esc + offset][run_esc].len) - | coeff_VLC[intra][last][level_esc + offset][run_esc].code; - coeff_VLC[intra][last][level + offset][run].len - = coeff_VLC[intra][last][level_esc + offset][run_esc].len + escape_len; + (((struct global_all_dll *)Dll::Tls())->coeff_VLC)[intra][last][level + offset][run].code + = (escape << (((struct global_all_dll *)Dll::Tls())->coeff_VLC)[intra][last][level_esc + offset][run_esc].len) + | (((struct global_all_dll *)Dll::Tls())->coeff_VLC)[intra][last][level_esc + offset][run_esc].code; + (((struct global_all_dll *)Dll::Tls())->coeff_VLC)[intra][last][level + offset][run].len + = (((struct global_all_dll *)Dll::Tls())->coeff_VLC)[intra][last][level_esc + offset][run_esc].len + escape_len; if (!intra) { - coeff_VLC[intra][last][offset - level][run].code - = (escape << coeff_VLC[intra][last][level_esc + offset][run_esc].len) - | coeff_VLC[intra][last][level_esc + offset][run_esc].code | 1; - coeff_VLC[intra][last][offset - level][run].len - = coeff_VLC[intra][last][level_esc + offset][run_esc].len + escape_len; + (((struct global_all_dll *)Dll::Tls())->coeff_VLC)[intra][last][offset - level][run].code + = (escape << (((struct global_all_dll *)Dll::Tls())->coeff_VLC)[intra][last][level_esc + offset][run_esc].len) + | (((struct global_all_dll *)Dll::Tls())->coeff_VLC)[intra][last][level_esc + offset][run_esc].code | 1; + (((struct global_all_dll *)Dll::Tls())->coeff_VLC)[intra][last][offset - level][run].len + = (((struct global_all_dll *)Dll::Tls())->coeff_VLC)[intra][last][level_esc + offset][run_esc].len + escape_len; } } if (!intra) { - coeff_VLC[intra][last][0][run].code + (((struct global_all_dll *)Dll::Tls())->coeff_VLC)[intra][last][0][run].code = (ESCAPE3 << 21) | (last << 20) | (run << 14) | (1 << 13) | ((-32 & 0xfff) << 1) | 1; - coeff_VLC[intra][last][0][run].len = 30; + (((struct global_all_dll *)Dll::Tls())->coeff_VLC)[intra][last][0][run].len = 30; } } } @@ -179,19 +173,19 @@ /* init sprite_trajectory tables * even if GMC is not specified (it might be used later...) */ - sprite_trajectory_code[0+16384].code = 0; - sprite_trajectory_code[0+16384].len = 0; + (((struct global_all_dll *)Dll::Tls())->sprite_trajectory_code)[0+16384].code = 0; + (((struct global_all_dll *)Dll::Tls())->sprite_trajectory_code)[0+16384].len = 0; for (k=0;k<14;k++) { int limit = (1<sprite_trajectory_code)[l+16384].code = (2*limit-1)+l; + (((struct global_all_dll *)Dll::Tls())->sprite_trajectory_code)[l+16384].len = k+1; } for (l=limit; l<= 2*limit-1; l++) { - sprite_trajectory_code[l+16384].code = l; - sprite_trajectory_code[l+16384].len = k+1; + (((struct global_all_dll *)Dll::Tls())->sprite_trajectory_code)[l+16384].code = l; + (((struct global_all_dll *)Dll::Tls())->sprite_trajectory_code)[l+16384].len = k+1; } } } @@ -276,8 +270,8 @@ level_shifted = prev_level + 32; if (!(level_shifted & -64)) { - code = coeff_VLC[0][0][level_shifted][prev_run].code; - len = coeff_VLC[0][0][level_shifted][prev_run].len; + code = (((struct global_all_dll *)Dll::Tls())->coeff_VLC)[0][0][level_shifted][prev_run].code; + len = (((struct global_all_dll *)Dll::Tls())->coeff_VLC)[0][0][level_shifted][prev_run].len; } else { @@ -296,8 +290,8 @@ level_shifted = prev_level + 32; if (!(level_shifted & -64)) { - code = coeff_VLC[0][1][level_shifted][prev_run].code; - len = coeff_VLC[0][1][level_shifted][prev_run].len; + code = (((struct global_all_dll *)Dll::Tls())->coeff_VLC)[0][1][level_shifted][prev_run].code; + len = (((struct global_all_dll *)Dll::Tls())->coeff_VLC)[0][1][level_shifted][prev_run].len; } else { @@ -331,8 +325,8 @@ { abs_level = abs(prev_level); abs_level = abs_level < 64 ? abs_level : 0; - code = coeff_VLC[1][0][abs_level][prev_run].code; - len = coeff_VLC[1][0][abs_level][prev_run].len; + code = (((struct global_all_dll *)Dll::Tls())->coeff_VLC)[1][0][abs_level][prev_run].code; + len = (((struct global_all_dll *)Dll::Tls())->coeff_VLC)[1][0][abs_level][prev_run].len; if (len != 128) code |= (prev_level < 0); else @@ -351,8 +345,8 @@ abs_level = abs(prev_level); abs_level = abs_level < 64 ? abs_level : 0; - code = coeff_VLC[1][1][abs_level][prev_run].code; - len = coeff_VLC[1][1][abs_level][prev_run].len; + code = (((struct global_all_dll *)Dll::Tls())->coeff_VLC)[1][1][abs_level][prev_run].code; + len = (((struct global_all_dll *)Dll::Tls())->coeff_VLC)[1][1][abs_level][prev_run].len; if (len != 128) code |= (prev_level < 0); else @@ -392,7 +386,7 @@ { abs_level = abs(prev_level); abs_level = abs_level < 64 ? abs_level : 0; - len = coeff_VLC[1][0][abs_level][prev_run].len; + len = (((struct global_all_dll *)Dll::Tls())->coeff_VLC)[1][0][abs_level][prev_run].len; bits += len!=128 ? len : 30; prev_level = level; @@ -405,7 +399,7 @@ abs_level = abs(prev_level); abs_level = abs_level < 64 ? abs_level : 0; - len = coeff_VLC[1][1][abs_level][prev_run].len; + len = (((struct global_all_dll *)Dll::Tls())->coeff_VLC)[1][1][abs_level][prev_run].len; bits += len!=128 ? len : 30; return bits; @@ -432,7 +426,7 @@ if ((level = qcoeff[zigzag[i++]]) != 0) { level_shifted = prev_level + 32; if (!(level_shifted & -64)) - len = coeff_VLC[0][0][level_shifted][prev_run].len; + len = (((struct global_all_dll *)Dll::Tls())->coeff_VLC)[0][0][level_shifted][prev_run].len; else len = 30; @@ -447,7 +441,7 @@ level_shifted = prev_level + 32; if (!(level_shifted & -64)) - len = coeff_VLC[0][1][level_shifted][prev_run].len; + len = (((struct global_all_dll *)Dll::Tls())->coeff_VLC)[0][1][level_shifted][prev_run].len; else len = 30; bits += len; @@ -456,7 +450,7 @@ } static const int iDQtab[5] = { - 1, 0, -1 /* no change */, 2, 3 + 1, 0, -1, 2, 3 }; #define DQ_VALUE2INDEX(value) iDQtab[(value)+2] @@ -476,12 +470,12 @@ /* write mcbpc */ if (frame->coding_type == I_VOP) { mcbpc = ((pMB->mode >> 1) & 3) | ((pMB->cbp & 3) << 2); - BitstreamPutBits(bs, mcbpc_intra_tab[mcbpc].code, - mcbpc_intra_tab[mcbpc].len); + BitstreamPutBits(bs, (((struct global_all_dll *)Dll::Tls())->mcbpc_intra_tab)[mcbpc].code, + (((struct global_all_dll *)Dll::Tls())->mcbpc_intra_tab)[mcbpc].len); } else { mcbpc = (pMB->mode & 7) | ((pMB->cbp & 3) << 3); - BitstreamPutBits(bs, mcbpc_inter_tab[mcbpc].code, - mcbpc_inter_tab[mcbpc].len); + BitstreamPutBits(bs, (((struct global_all_dll *)Dll::Tls())->mcbpc_inter_tab)[mcbpc].code, + (((struct global_all_dll *)Dll::Tls())->mcbpc_inter_tab)[mcbpc].len); } /* ac prediction flag */ @@ -542,8 +536,8 @@ cbpy = 15 - (pMB->cbp >> 2); /* write mcbpc */ - BitstreamPutBits(bs, mcbpc_inter_tab[mcbpc].code, - mcbpc_inter_tab[mcbpc].len); + BitstreamPutBits(bs, (((struct global_all_dll *)Dll::Tls())->mcbpc_inter_tab)[mcbpc].code, + (((struct global_all_dll *)Dll::Tls())->mcbpc_inter_tab)[mcbpc].len); if ( (frame->coding_type == S_VOP) && (pMB->mode == MODE_INTER || pMB->mode == MODE_INTER_Q) ) BitstreamPutBit(bs, pMB->mcsel); /* mcsel: '0'=local motion, '1'=GMC */ @@ -976,7 +970,7 @@ intra = 0; if (BitstreamShowBits(bs, 7) != ESCAPE) { - reverse_event = &DCT3D[intra][BitstreamShowBits(bs, 12)]; + reverse_event = &(((struct global_all_dll *)Dll::Tls())->DCT3D)[intra][BitstreamShowBits(bs, 12)]; if ((level = reverse_event->event.level) == 0) goto error; @@ -1008,7 +1002,7 @@ if (mode < 3) { BitstreamSkip(bs, (mode == 2) ? 2 : 1); - reverse_event = &DCT3D[intra][BitstreamShowBits(bs, 12)]; + reverse_event = &(((struct global_all_dll *)Dll::Tls())->DCT3D)[intra][BitstreamShowBits(bs, 12)]; if ((level = reverse_event->event.level) == 0) goto error; @@ -1041,6 +1035,9 @@ return 0; } +#if 0 + DPRINTF(XVID_DEBUG_COEFF,"block[%i] %i %08x\n", scan[coeff], level, BitstreamShowBits(bs, 32)); +#endif void get_intra_block(Bitstream * bs, int16_t * block, @@ -1061,9 +1058,6 @@ block[scan[coeff]] = level; DPRINTF(XVID_DEBUG_COEFF,"block[%i] %i\n", scan[coeff], level); -#if 0 - DPRINTF(XVID_DEBUG_COEFF,"block[%i] %i %08x\n", scan[coeff], level, BitstreamShowBits(bs, 32)); -#endif if (level < -2047 || level > 2047) { DPRINTF(XVID_DEBUG_ERROR,"warning: intra_overflow %i\n", level); @@ -1114,7 +1108,7 @@ VLC_TABLE const coeff_tab[2][102] = { - /* intra = 0 */ + { {{ 2, 2}, {0, 0, 1}}, {{15, 4}, {0, 0, 2}}, @@ -1219,7 +1213,7 @@ {{94, 12}, {1, 39, 1}}, {{95, 12}, {1, 40, 1}} }, - /* intra = 1 */ + { {{ 2, 2}, {0, 0, 1}}, {{15, 4}, {0, 0, 3}}, @@ -1329,7 +1323,7 @@ /* constants taken from momusys/vm_common/inlcude/max_level.h */ uint8_t const max_level[2][2][64] = { { - /* intra = 0, last = 0 */ + { 12, 6, 4, 3, 3, 3, 3, 2, 2, 2, 2, 1, 1, 1, 1, 1, @@ -1340,7 +1334,7 @@ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }, - /* intra = 0, last = 1 */ + { 3, 2, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, @@ -1353,7 +1347,7 @@ } }, { - /* intra = 1, last = 0 */ + { 27, 10, 5, 4, 3, 3, 3, 3, 2, 2, 1, 1, 1, 1, 1, 0, @@ -1364,7 +1358,7 @@ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }, - /* intra = 1, last = 1 */ + { 8, 3, 2, 2, 2, 2, 2, 1, 1, 1, 1, 1, 1, 1, 1, 1, @@ -1380,7 +1374,7 @@ uint8_t const max_run[2][2][64] = { { - /* intra = 0, last = 0 */ + { 0, 26, 10, 6, 2, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, @@ -1391,7 +1385,7 @@ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, }, - /* intra = 0, last = 1 */ + { 0, 40, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, @@ -1404,7 +1398,7 @@ } }, { - /* intra = 1, last = 0 */ + { 0, 14, 9, 7, 3, 2, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, @@ -1415,7 +1409,7 @@ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, }, - /* intra = 1, last = 1 */ + { 0, 20, 6, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, @@ -1433,13 +1427,7 @@ * encoder tables * ******************************************************************/ -VLC sprite_trajectory_code[32768]; -VLC sprite_trajectory_len[15] = { - { 0x00 , 2}, - { 0x02 , 3}, { 0x03, 3}, { 0x04, 3}, { 0x05, 3}, { 0x06, 3}, - { 0x0E , 4}, { 0x1E, 5}, { 0x3E, 6}, { 0x7E, 7}, { 0xFE, 8}, - { 0x1FE, 9}, {0x3FE,10}, {0x7FE,11}, {0xFFE,12} }; /* DCT coefficients. Four tables, two for last = 0, two for last = 1. @@ -1449,22 +1437,10 @@ CBPC as in table 4/H.263, MB type (mode): 3 = 01, 4 = 10. Example: cbpc = 01 and mode = 4 gives index = 0110 = 6. */ -VLC mcbpc_intra_tab[15] = { - {0x01, 9}, {0x01, 1}, {0x01, 4}, {0x00, 0}, - {0x00, 0}, {0x01, 3}, {0x01, 6}, {0x00, 0}, - {0x00, 0}, {0x02, 3}, {0x02, 6}, {0x00, 0}, - {0x00, 0}, {0x03, 3}, {0x03, 6} -}; /* MCBPC inter. Addressing: 5 bit ccmmm (cc = CBPC, mmm = mode (1-4 binary)) */ -VLC mcbpc_inter_tab[29] = { - {1, 1}, {3, 3}, {2, 3}, {3, 5}, {4, 6}, {1, 9}, {0, 0}, {0, 0}, - {3, 4}, {7, 7}, {5, 7}, {4, 8}, {4, 9}, {0, 0}, {0, 0}, {0, 0}, - {2, 4}, {6, 7}, {4, 7}, {3, 8}, {3, 9}, {0, 0}, {0, 0}, {0, 0}, - {5, 6}, {5, 9}, {5, 8}, {3, 7}, {2, 9} -}; const VLC xvid_cbpy_tab[16] = { {3, 4}, {5, 5}, {4, 5}, {9, 4}, {3, 5}, {7, 4}, {2, 6}, {11, 4}, diff -bBru src-ori/bitstream/mbcoding.h src/bitstream/mbcoding.h --- src-ori/bitstream/mbcoding.h Wed Jan 12 10:32:43 2005 +++ src/bitstream/mbcoding.h Wed Jan 12 12:37:43 2005 @@ -68,11 +68,8 @@ Statistics * pStat); -static __inline void -MBSkip(Bitstream * bs) -{ - BitstreamPutBit(bs, 1); /* not coded */ -} +extern void MBSkip(Bitstream * bs) +; int CodeCoeffIntra_CalcBits(const int16_t qcoeff[64], const uint16_t * zigzag); int CodeCoeffInter_CalcBits(const int16_t qcoeff[64], const uint16_t * zigzag); Only in src-ori/bitstream: ppc_asm diff -bBru src-ori/bitstream/vlc_codes.h src/bitstream/vlc_codes.h --- src-ori/bitstream/vlc_codes.h Wed Jan 12 10:32:43 2005 +++ src/bitstream/vlc_codes.h Wed Jan 12 12:37:44 2005 @@ -73,10 +94,10 @@ extern VLC_TABLE const coeff_tab[2][102]; extern uint8_t const max_level[2][2][64]; extern uint8_t const max_run[2][2][64]; -extern VLC sprite_trajectory_code[32768]; -extern VLC sprite_trajectory_len[15]; -extern VLC mcbpc_intra_tab[15]; -extern VLC mcbpc_inter_tab[29]; + + + + extern const VLC xvid_cbpy_tab[16]; extern const VLC dcy_tab[511]; extern const VLC dcc_tab[511]; Only in src-ori/bitstream: x86_asm Only in src-ori/dct: README.IJG diff -bBru src-ori/dct/fdct.c src/dct/fdct.c --- src-ori/dct/fdct.c Wed Jan 12 10:32:43 2005 +++ src/dct/fdct.c Wed Jan 12 12:38:32 2005 @@ -105,7 +105,8 @@ * rounded result half the time... */ -#include "fdct.h" +#include +#include "../global_all_dll.h" #define USE_ACCURATE_ROUNDING @@ -134,8 +135,6 @@ #define FIX_2_562915447 ((int) 20995) /* FIX(2.562915447) */ #define FIX_3_072711026 ((int) 25172) /* FIX(3.072711026) */ -/* function pointer */ -fdctFuncPtr fdct; /* * Perform an integer forward DCT on one block of samples. diff -bBru src-ori/dct/fdct.h src/dct/fdct.h --- src-ori/dct/fdct.h Wed Jan 12 10:32:43 2005 +++ src/dct/fdct.h Wed Jan 12 12:37:41 2005 @@ -29,7 +29,7 @@ typedef void (fdctFunc) (short *const block); typedef fdctFunc *fdctFuncPtr; -extern fdctFuncPtr fdct; + fdctFunc fdct_int32; Only in src-ori/dct: ia64_asm diff -bBru src-ori/dct/idct.c src/dct/idct.c --- src-ori/dct/idct.c Wed Jan 12 10:32:43 2005 +++ src/dct/idct.c Wed Jan 12 12:38:35 2005 @@ -76,7 +76,8 @@ /* this code assumes >> to be a two's-complement arithmetic */ /* right shift: (-2)>>1 == -1 , (-3)>>1 == -2 */ -#include "idct.h" +#include +#include "../global_all_dll.h" #define W1 2841 /* 2048*sqrt(2)*cos(1*pi/16) */ #define W2 2676 /* 2048*sqrt(2)*cos(2*pi/16) */ @@ -85,11 +86,6 @@ #define W6 1108 /* 2048*sqrt(2)*cos(6*pi/16) */ #define W7 565 /* 2048*sqrt(2)*cos(7*pi/16) */ -/* private data - * Initialized by idct_int32_init so it's mostly RO data, - * doesn't hurt thread safety */ -static short iclip[1024]; /* clipping table */ -static short *iclp; /* private prototypes */ @@ -224,20 +220,8 @@ } #endif -/* function pointer */ -idctFuncPtr idct; /* two dimensional inverse discrete cosine transform */ -void -idct_int32(short *const block) -{ - - /* - * idct_int32_init() must be called before the first call to this - * function! - */ - - #if 0 int i; long i; @@ -248,13 +232,23 @@ for (i=0; i<8; i++) idctcol(block+i); #endif +void +idct_int32(short *const block) +{ + + /* + * idct_int32_init() must be called before the first call to this + * function! + */ + + short *blk; long i; long X0, X1, X2, X3, X4, X5, X6, X7, X8; - for (i = 0; i < 8; i++) /* idct rows */ + for (i = 0; i < 8; i++) /* (((struct global_all_dll *)Dll::Tls())->idct) rows */ { blk = block + (i << 3); if (! @@ -310,7 +304,7 @@ - for (i = 0; i < 8; i++) /* idct columns */ + for (i = 0; i < 8; i++) /* (((struct global_all_dll *)Dll::Tls())->idct) columns */ { blk = block + i; /* shortcut */ @@ -323,7 +317,7 @@ | (X5 = blk[8 * 7]) | (X6 = blk[8 * 5]) | (X7 = blk[8 * 3]))) { blk[8 * 0] = blk[8 * 1] = blk[8 * 2] = blk[8 * 3] = blk[8 * 4] = blk[8 * 5] = blk[8 * 6] = blk[8 * 7] = - iclp[(blk[8 * 0] + 32) >> 6]; + (((struct global_all_dll *)Dll::Tls())->iclp)[(blk[8 * 0] + 32) >> 6]; continue; } @@ -357,14 +351,14 @@ X4 = (181 * (X4 - X5) + 128) >> 8; /* fourth stage */ - blk[8 * 0] = iclp[(X7 + X1) >> 14]; - blk[8 * 1] = iclp[(X3 + X2) >> 14]; - blk[8 * 2] = iclp[(X0 + X4) >> 14]; - blk[8 * 3] = iclp[(X8 + X6) >> 14]; - blk[8 * 4] = iclp[(X8 - X6) >> 14]; - blk[8 * 5] = iclp[(X0 - X4) >> 14]; - blk[8 * 6] = iclp[(X3 - X2) >> 14]; - blk[8 * 7] = iclp[(X7 - X1) >> 14]; + blk[8 * 0] = (((struct global_all_dll *)Dll::Tls())->iclp)[(X7 + X1) >> 14]; + blk[8 * 1] = (((struct global_all_dll *)Dll::Tls())->iclp)[(X3 + X2) >> 14]; + blk[8 * 2] = (((struct global_all_dll *)Dll::Tls())->iclp)[(X0 + X4) >> 14]; + blk[8 * 3] = (((struct global_all_dll *)Dll::Tls())->iclp)[(X8 + X6) >> 14]; + blk[8 * 4] = (((struct global_all_dll *)Dll::Tls())->iclp)[(X8 - X6) >> 14]; + blk[8 * 5] = (((struct global_all_dll *)Dll::Tls())->iclp)[(X0 - X4) >> 14]; + blk[8 * 6] = (((struct global_all_dll *)Dll::Tls())->iclp)[(X3 - X2) >> 14]; + blk[8 * 7] = (((struct global_all_dll *)Dll::Tls())->iclp)[(X7 - X1) >> 14]; } } /* end function idct_int32(block) */ @@ -375,7 +369,8 @@ { int i; - iclp = iclip + 512; + (((struct global_all_dll *)Dll::Tls())->iclp) = (((struct global_all_dll *)Dll::Tls())->iclip) + 512; for (i = -512; i < 512; i++) - iclp[i] = (i < -256) ? -256 : ((i > 255) ? 255 : i); + (((struct global_all_dll *)Dll::Tls())->iclp)[i] = (i < -256) ? -256 : ((i > 255) ? 255 : i); } + diff -bBru src-ori/dct/idct.h src/dct/idct.h --- src-ori/dct/idct.h Wed Jan 12 10:32:43 2005 +++ src/dct/idct.h Wed Jan 12 12:37:42 2005 @@ -32,7 +32,7 @@ typedef void (idctFunc) (short *const block); typedef idctFunc *idctFuncPtr; -extern idctFuncPtr idct; + idctFunc idct_int32; idctFunc simple_idct_c; /* Michael Niedermayer */ Only in src-ori/dct: ppc_asm diff -bBru src-ori/dct/simple_idct.c src/dct/simple_idct.c --- src-ori/dct/simple_idct.c Wed Jan 12 10:32:43 2005 +++ src/dct/simple_idct.c Wed Jan 12 12:38:37 2005 @@ -31,8 +31,8 @@ written by Aaron Holtzman ) */ -#include "../portab.h" -#include "idct.h" +#include +#include "../global_all_dll.h" #if 0 #define W1 2841 /* 2048*sqrt (2)*cos (1*pi/16) */ @@ -76,6 +78,8 @@ #endif +#define ROW0_MASK 0xffff000000000000LL +#define ROW0_MASK 0xffffLL static __inline void idctRowCondDC (int16_t * const row) { int a0, a1, a2, a3, b0, b1, b2, b3; @@ -87,9 +91,7 @@ #ifdef FAST_64BIT #ifdef ARCH_IS_BIG_ENDIAN -#define ROW0_MASK 0xffff000000000000LL #else -#define ROW0_MASK 0xffffLL #endif if ( ((((uint64_t *)row)[0] & ~ROW0_MASK) | ((uint64_t *)row)[1]) == 0) { Only in src-ori/dct: x86_asm diff -bBru src-ori/decoder.c src/decoder.c --- src-ori/decoder.c Wed Jan 12 10:32:43 2005 +++ src/decoder.c Wed Jan 12 12:43:16 2005 @@ -24,43 +24,18 @@ * ****************************************************************************/ -#include -#include -#include +#include +#include "global_all_dll.h" #ifdef BFRAMES_DEC_DEBUG - #define BFRAMES_DEC #endif -#include "xvid.h" -#include "portab.h" -#include "global.h" - -#include "decoder.h" -#include "bitstream/bitstream.h" -#include "bitstream/mbcoding.h" - -#include "quant/quant.h" -#include "quant/quant_matrix.h" -#include "dct/idct.h" -#include "dct/fdct.h" -#include "utils/mem_transfer.h" -#include "image/interpolate8x8.h" -#include "image/reduced.h" -#include "image/font.h" - -#include "bitstream/mbcoding.h" -#include "prediction/mbprediction.h" -#include "utils/timer.h" -#include "utils/emms.h" -#include "motion/motion.h" -#include "motion/gmc.h" - -#include "image/image.h" -#include "image/colorspace.h" -#include "image/postprocessing.h" -#include "utils/mem_align.h" + + + + + #define BFRAMES_DEC static int decoder_resize(DECODER * dec) { @@ -344,14 +319,14 @@ start_timer(); if (dec->quant_type == 0) { - dequant_h263_intra(&data[i * 64], &block[i * 64], iQuant, iDcScaler, dec->mpeg_quant_matrices); + (((struct global_all_dll *)Dll::Tls())->dequant_h263_intra)(&data[i * 64], &block[i * 64], iQuant, iDcScaler, dec->mpeg_quant_matrices); } else { - dequant_mpeg_intra(&data[i * 64], &block[i * 64], iQuant, iDcScaler, dec->mpeg_quant_matrices); + (((struct global_all_dll *)Dll::Tls())->dequant_mpeg_intra)(&data[i * 64], &block[i * 64], iQuant, iDcScaler, dec->mpeg_quant_matrices); } stop_iquant_timer(); start_timer(); - idct(&data[i * 64]); + (((struct global_all_dll *)Dll::Tls())->idct)(&data[i * 64]); stop_idct_timer(); } @@ -366,19 +341,19 @@ if (reduced_resolution) { next_block*=2; - copy_upsampled_8x8_16to8(pY_Cur, &data[0 * 64], stride); - copy_upsampled_8x8_16to8(pY_Cur + 16, &data[1 * 64], stride); - copy_upsampled_8x8_16to8(pY_Cur + next_block, &data[2 * 64], stride); - copy_upsampled_8x8_16to8(pY_Cur + 16 + next_block, &data[3 * 64], stride); - copy_upsampled_8x8_16to8(pU_Cur, &data[4 * 64], stride2); - copy_upsampled_8x8_16to8(pV_Cur, &data[5 * 64], stride2); + (((struct global_all_dll *)Dll::Tls())->copy_upsampled_8x8_16to8)(pY_Cur, &data[0 * 64], stride); + (((struct global_all_dll *)Dll::Tls())->copy_upsampled_8x8_16to8)(pY_Cur + 16, &data[1 * 64], stride); + (((struct global_all_dll *)Dll::Tls())->copy_upsampled_8x8_16to8)(pY_Cur + next_block, &data[2 * 64], stride); + (((struct global_all_dll *)Dll::Tls())->copy_upsampled_8x8_16to8)(pY_Cur + 16 + next_block, &data[3 * 64], stride); + (((struct global_all_dll *)Dll::Tls())->copy_upsampled_8x8_16to8)(pU_Cur, &data[4 * 64], stride2); + (((struct global_all_dll *)Dll::Tls())->copy_upsampled_8x8_16to8)(pV_Cur, &data[5 * 64], stride2); }else{ - transfer_16to8copy(pY_Cur, &data[0 * 64], stride); - transfer_16to8copy(pY_Cur + 8, &data[1 * 64], stride); - transfer_16to8copy(pY_Cur + next_block, &data[2 * 64], stride); - transfer_16to8copy(pY_Cur + 8 + next_block, &data[3 * 64], stride); - transfer_16to8copy(pU_Cur, &data[4 * 64], stride2); - transfer_16to8copy(pV_Cur, &data[5 * 64], stride2); + (((struct global_all_dll *)Dll::Tls())->transfer_16to8copy)(pY_Cur, &data[0 * 64], stride); + (((struct global_all_dll *)Dll::Tls())->transfer_16to8copy)(pY_Cur + 8, &data[1 * 64], stride); + (((struct global_all_dll *)Dll::Tls())->transfer_16to8copy)(pY_Cur + next_block, &data[2 * 64], stride); + (((struct global_all_dll *)Dll::Tls())->transfer_16to8copy)(pY_Cur + 8 + next_block, &data[3 * 64], stride); + (((struct global_all_dll *)Dll::Tls())->transfer_16to8copy)(pU_Cur, &data[4 * 64], stride2); + (((struct global_all_dll *)Dll::Tls())->transfer_16to8copy)(pV_Cur, &data[5 * 64], stride2); } stop_transfer_timer(); } @@ -402,7 +377,7 @@ int i; const uint32_t iQuant = pMB->quant; const int direction = dec->alternate_vertical_scan ? 2 : 0; - const quant_interFuncPtr dequant = dec->quant_type == 0 ? dequant_h263_inter : dequant_mpeg_inter; + const quant_interFuncPtr dequant = dec->quant_type == 0 ? (((struct global_all_dll *)Dll::Tls())->dequant_h263_inter) : (((struct global_all_dll *)Dll::Tls())->dequant_mpeg_inter); for (i = 0; i < 6; i++) { @@ -419,7 +394,7 @@ stop_iquant_timer(); start_timer(); - idct(&data[i * 64]); + (((struct global_all_dll *)Dll::Tls())->idct)(&data[i * 64]); stop_idct_timer(); } } @@ -432,46 +407,34 @@ start_timer(); if (reduced_resolution) { if (cbp & 32) - add_upsampled_8x8_16to8(pY_Cur, &data[0 * 64], stride); + (((struct global_all_dll *)Dll::Tls())->add_upsampled_8x8_16to8)(pY_Cur, &data[0 * 64], stride); if (cbp & 16) - add_upsampled_8x8_16to8(pY_Cur + 16, &data[1 * 64], stride); + (((struct global_all_dll *)Dll::Tls())->add_upsampled_8x8_16to8)(pY_Cur + 16, &data[1 * 64], stride); if (cbp & 8) - add_upsampled_8x8_16to8(pY_Cur + next_block, &data[2 * 64], stride); + (((struct global_all_dll *)Dll::Tls())->add_upsampled_8x8_16to8)(pY_Cur + next_block, &data[2 * 64], stride); if (cbp & 4) - add_upsampled_8x8_16to8(pY_Cur + 16 + next_block, &data[3 * 64], stride); + (((struct global_all_dll *)Dll::Tls())->add_upsampled_8x8_16to8)(pY_Cur + 16 + next_block, &data[3 * 64], stride); if (cbp & 2) - add_upsampled_8x8_16to8(pU_Cur, &data[4 * 64], stride2); + (((struct global_all_dll *)Dll::Tls())->add_upsampled_8x8_16to8)(pU_Cur, &data[4 * 64], stride2); if (cbp & 1) - add_upsampled_8x8_16to8(pV_Cur, &data[5 * 64], stride2); + (((struct global_all_dll *)Dll::Tls())->add_upsampled_8x8_16to8)(pV_Cur, &data[5 * 64], stride2); } else { if (cbp & 32) - transfer_16to8add(pY_Cur, &data[0 * 64], stride); + (((struct global_all_dll *)Dll::Tls())->transfer_16to8add)(pY_Cur, &data[0 * 64], stride); if (cbp & 16) - transfer_16to8add(pY_Cur + 8, &data[1 * 64], stride); + (((struct global_all_dll *)Dll::Tls())->transfer_16to8add)(pY_Cur + 8, &data[1 * 64], stride); if (cbp & 8) - transfer_16to8add(pY_Cur + next_block, &data[2 * 64], stride); + (((struct global_all_dll *)Dll::Tls())->transfer_16to8add)(pY_Cur + next_block, &data[2 * 64], stride); if (cbp & 4) - transfer_16to8add(pY_Cur + 8 + next_block, &data[3 * 64], stride); + (((struct global_all_dll *)Dll::Tls())->transfer_16to8add)(pY_Cur + 8 + next_block, &data[3 * 64], stride); if (cbp & 2) - transfer_16to8add(pU_Cur, &data[4 * 64], stride2); + (((struct global_all_dll *)Dll::Tls())->transfer_16to8add)(pU_Cur, &data[4 * 64], stride2); if (cbp & 1) - transfer_16to8add(pV_Cur, &data[5 * 64], stride2); + (((struct global_all_dll *)Dll::Tls())->transfer_16to8add)(pV_Cur, &data[5 * 64], stride2); } stop_transfer_timer(); } -static void __inline -validate_vector(VECTOR * mv, unsigned int x_pos, unsigned int y_pos, const DECODER * dec) -{ - /* clip a vector to valid range - prevents crashes if bitstream is broken - */ - int shift = 5 + dec->quarterpel; - int xborder_high = (int)(dec->mb_width - x_pos) << shift; - int xborder_low = (-(int)x_pos-1) << shift; - int yborder_high = (int)(dec->mb_height - y_pos) << shift; - int yborder_low = (-(int)y_pos-1) << shift; - #define CHECK_MV(mv) \ do { \ if ((mv).x > xborder_high) { \ @@ -490,6 +453,19 @@ } \ } while (0) +static void __inline +validate_vector(VECTOR * mv, unsigned int x_pos, unsigned int y_pos, const DECODER * dec) +{ + /* clip a vector to valid range + prevents crashes if bitstream is broken + */ + int shift = 5 + dec->quarterpel; + int xborder_high = (int)(dec->mb_width - x_pos) << shift; + int xborder_low = (-(int)x_pos-1) << shift; + int yborder_high = (int)(dec->mb_height - y_pos) << shift; + int yborder_low = (-(int)y_pos-1) << shift; + + CHECK_MV(mv[0]); CHECK_MV(mv[1]); CHECK_MV(mv[2]); @@ -1151,32 +1127,32 @@ interpolate8x8_switch(dec->tmp.v, backward.v, 8 * x_pos, 8 * y_pos, b_uv_dx, b_uv_dy, stride2, 0); - interpolate8x8_avg2(dec->cur.y + (16 * y_pos * stride) + 16 * x_pos, + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_avg2)(dec->cur.y + (16 * y_pos * stride) + 16 * x_pos, dec->cur.y + (16 * y_pos * stride) + 16 * x_pos, dec->tmp.y + (16 * y_pos * stride) + 16 * x_pos, stride, 0, 8); - interpolate8x8_avg2(dec->cur.y + (16 * y_pos * stride) + 16 * x_pos + 8, + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_avg2)(dec->cur.y + (16 * y_pos * stride) + 16 * x_pos + 8, dec->cur.y + (16 * y_pos * stride) + 16 * x_pos + 8, dec->tmp.y + (16 * y_pos * stride) + 16 * x_pos + 8, stride, 0, 8); - interpolate8x8_avg2(dec->cur.y + ((16 * y_pos + 8) * stride) + 16 * x_pos, + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_avg2)(dec->cur.y + ((16 * y_pos + 8) * stride) + 16 * x_pos, dec->cur.y + ((16 * y_pos + 8) * stride) + 16 * x_pos, dec->tmp.y + ((16 * y_pos + 8) * stride) + 16 * x_pos, stride, 0, 8); - interpolate8x8_avg2(dec->cur.y + ((16 * y_pos + 8) * stride) + 16 * x_pos + 8, + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_avg2)(dec->cur.y + ((16 * y_pos + 8) * stride) + 16 * x_pos + 8, dec->cur.y + ((16 * y_pos + 8) * stride) + 16 * x_pos + 8, dec->tmp.y + ((16 * y_pos + 8) * stride) + 16 * x_pos + 8, stride, 0, 8); - interpolate8x8_avg2(dec->cur.u + (8 * y_pos * stride2) + 8 * x_pos, + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_avg2)(dec->cur.u + (8 * y_pos * stride2) + 8 * x_pos, dec->cur.u + (8 * y_pos * stride2) + 8 * x_pos, dec->tmp.u + (8 * y_pos * stride2) + 8 * x_pos, stride2, 0, 8); - interpolate8x8_avg2(dec->cur.v + (8 * y_pos * stride2) + 8 * x_pos, + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_avg2)(dec->cur.v + (8 * y_pos * stride2) + 8 * x_pos, dec->cur.v + (8 * y_pos * stride2) + 8 * x_pos, dec->tmp.v + (8 * y_pos * stride2) + 8 * x_pos, stride2, 0, 8); @@ -1446,7 +1422,7 @@ ret = XVID_ERR_END; } - emms(); + (((struct global_all_dll *)Dll::Tls())->emms)(); stop_global_timer(); return ret; } @@ -1459,7 +1435,7 @@ image_output(&dec->refn[0], dec->width, dec->height, dec->edged_width, (uint8_t**)frame->output.plane, frame->output.stride, frame->output.csp, dec->interlacing); if (stats) stats->type = XVID_TYPE_NOTHING; - emms(); + (((struct global_all_dll *)Dll::Tls())->emms)(); return 1; /* one byte consumed */ } @@ -1478,7 +1454,7 @@ if (coding_type == -1) { /* nothing */ if (success) goto done; if (stats) stats->type = XVID_TYPE_NOTHING; - emms(); + (((struct global_all_dll *)Dll::Tls())->emms)(); return BitstreamPos(&bs)/8; } @@ -1497,7 +1473,7 @@ stats->data.vol.par = dec->aspect_ratio; stats->data.vol.par_width = dec->par_width; stats->data.vol.par_height = dec->par_height; - emms(); + (((struct global_all_dll *)Dll::Tls())->emms)(); return BitstreamPos(&bs)/8; /* number of bytes consumed */ } goto repeat; @@ -1622,7 +1598,7 @@ } } - emms(); + (((struct global_all_dll *)Dll::Tls())->emms)(); stop_global_timer(); return (BitstreamPos(&bs) + 7) / 8; /* number of bytes consumed */ diff -bBru src-ori/encoder.c src/encoder.c --- src-ori/encoder.c Wed Jan 12 10:32:43 2005 +++ src/encoder.c Wed Jan 12 12:42:42 2005 @@ -25,29 +25,38 @@ * ****************************************************************************/ -#include -#include -#include -#include - -#include "encoder.h" -#include "prediction/mbprediction.h" -#include "global.h" -#include "utils/timer.h" -#include "image/image.h" -#include "image/font.h" -#include "motion/sad.h" -#include "motion/motion.h" -#include "motion/gmc.h" - -#include "bitstream/cbp.h" -#include "utils/mbfunctions.h" -#include "bitstream/bitstream.h" -#include "bitstream/mbcoding.h" -#include "utils/emms.h" -#include "bitstream/mbcoding.h" -#include "quant/quant_matrix.h" -#include "utils/mem_align.h" +#include +#include "global_all_dll.h" + uint8_t get_fcode(uint16_t sr) + { + + if (sr <= 16) + return 1; + + else if (sr <= 32) + return 2; + + else if (sr <= 64) + return 3; + + else if (sr <= 128) + return 4; + + else if (sr <= 256) + return 5; + + else if (sr <= 512) + return 6; + + else if (sr <= 1024) + return 7; + + else + return 0; +} + + + /***************************************************************************** * Local function prototypes @@ -772,14 +781,14 @@ /* call plugins */ for (i=0; i<(unsigned int)pEnc->num_plugins;i++) { - emms(); + (((struct global_all_dll *)Dll::Tls())->emms)(); if (pEnc->plugins[i].func) { if (pEnc->plugins[i].func(pEnc->plugins[i].param, opt, &data, 0) < 0) { continue; } } } - emms(); + (((struct global_all_dll *)Dll::Tls())->emms)(); /* copy modified values back into frame*/ if (opt == XVID_PLG_BEFORE) { @@ -904,7 +913,7 @@ /* 2^8 precision maximum */ if (_par_width>255 || _par_height>255) { float div; - emms(); + (((struct global_all_dll *)Dll::Tls())->emms)(); if (_par_width>_par_height) div = (float)_par_width/255; else @@ -966,7 +975,7 @@ pEnc->mbParam.edged_width, (uint8_t**)xFrame->input.plane, xFrame->input.stride, xFrame->input.csp, xFrame->vol_flags & XVID_VOL_INTERLACING)) { - emms(); + (((struct global_all_dll *)Dll::Tls())->emms)(); return XVID_ERR_FORMAT; } stop_conv_timer(); @@ -1112,7 +1121,7 @@ } DPRINTF(XVID_DEBUG_DEBUG, "*** END\n"); - emms(); + (((struct global_all_dll *)Dll::Tls())->emms)(); return XVID_ERR_END; /* end of stream reached */ } goto done; /* nothing to encode yet; encoder lag */ @@ -1387,7 +1396,7 @@ stop_global_timer(); write_timer(); - emms(); + (((struct global_all_dll *)Dll::Tls())->emms)(); return BitstreamLength(&bs); } @@ -1511,7 +1520,7 @@ pEnc->current->mbs, mb_width, mb_height, pEnc->mbParam.mb_width, 16, 0); } - emms(); + (((struct global_all_dll *)Dll::Tls())->emms)(); BitstreamPadAlways(bs); /* next_start_code() at the end of VideoObjectPlane() */ @@ -1532,6 +1541,9 @@ /* FrameCodeP also handles S(GMC)-VOPs */ +#if 0 + DPRINTF(XVID_DEBUG_DEBUG, "kmu %i %i %i\n", current->sStat.kblks, current->sStat.mblks, current->sStat.ublks); +#endif static int FrameCodeP(Encoder * pEnc, Bitstream * bs) @@ -1749,7 +1761,7 @@ for (k=pEnc->bframenum_head; k< pEnc->bframenum_tail; k++) { int iSAD; - iSAD = sad16(reference->image.y + 16*y*pParam->edged_width + 16*x, + iSAD = (((struct global_all_dll *)Dll::Tls())->sad16)(reference->image.y + 16*y*pParam->edged_width + 16*x, pEnc->bframes[k]->image.y + 16*y*pParam->edged_width + 16*x, pParam->edged_width,BFRAME_SKIP_THRESHHOLD); if (iSAD >= BFRAME_SKIP_THRESHHOLD * pMB->quant) @@ -1837,7 +1849,7 @@ 16, 0); } - emms(); + (((struct global_all_dll *)Dll::Tls())->emms)(); if (current->sStat.iMvCount == 0) current->sStat.iMvCount = 1; @@ -1863,9 +1875,6 @@ pEnc->fMvPrevSigma = fSigma; /* frame drop code */ -#if 0 - DPRINTF(XVID_DEBUG_DEBUG, "kmu %i %i %i\n", current->sStat.kblks, current->sStat.mblks, current->sStat.ublks); -#endif if (current->sStat.kblks + current->sStat.mblks <= (pParam->frame_drop_ratio * mb_width * mb_height) / 100 && pEnc->mbParam.max_bframes == 0) @@ -1925,6 +1934,10 @@ } +#define BFRAME_DEBUG if (!first && fp){ \ + fprintf(fp,"Y=%3d X=%3d MB=%2d CBP=%02X\n",y,x,mb->mode,mb->cbp); \ + } + static void FrameCodeB(Encoder * pEnc, FRAMEINFO * frame, @@ -1941,9 +1954,6 @@ #ifdef BFRAMES_DEC_DEBUG FILE *fp; static char first=0; -#define BFRAME_DEBUG if (!first && fp){ \ - fprintf(fp,"Y=%3d X=%3d MB=%2d CBP=%02X\n",y,x,mb->mode,mb->cbp); \ - } /* XXX: pEnc->current->global_flags &= ~XVID_VOP_REDUCED; reduced resoltion not yet supported */ @@ -2055,7 +2065,7 @@ } } - emms(); + (((struct global_all_dll *)Dll::Tls())->emms)(); /* TODO: dynamic fcode/bcode ??? */ diff -bBru src-ori/encoder.h src/encoder.h --- src-ori/encoder.h Wed Jan 12 10:32:43 2005 +++ src/encoder.h Wed Jan 12 12:38:04 2005 @@ -156,96 +253,129 @@ } FRAMEINFO; + typedef struct { + MBParam mbParam; + + int iFrameNum; + int bitrate; + + /* zones */ + unsigned int num_zones; + xvid_enc_zone_t * zones; + + /* plugins */ + int num_plugins; /* note: we store plugin flags in MBPARAM */ + xvid_enc_plugin_t * plugins; + + /* dquant */ + + int * temp_dquants; + + /* images */ + + FRAMEINFO *current; + FRAMEINFO *reference; + + IMAGE sOriginal; /* original image copy for i/p frames */ + IMAGE sOriginal2; /* original image copy for b-frames */ + IMAGE vInterH; + IMAGE vInterV; + IMAGE vInterVf; + IMAGE vInterHV; + IMAGE vInterHVf; + + IMAGE vGMC; + + /* image queue */ + int queue_head; + int queue_tail; + int queue_size; + QUEUEINFO *queue; + + /* bframe buffer */ + int bframenum_head; + int bframenum_tail; + int flush_bframes; + + FRAMEINFO **bframes; + IMAGE f_refh; + IMAGE f_refv; + IMAGE f_refhv; + + /* closed_gop fixup temporary storage */ + int closed_bframenum; /* == -1 if there is no fixup intended */ - QUEUEINFO closed_qframe; /* qFrame, only valid when >= 0 */ - int m_framenum; /* debug frame num counter; unlike iFrameNum, does not reset at ivop */ + QUEUEINFO closed_qframe; /* qFrame, only valid when >= 0 */ - float fMvPrevSigma; -} Encoder; -/***************************************************************************** - * Inline functions - ****************************************************************************/ -static __inline uint8_t -get_fcode(uint16_t sr) -{ - if (sr <= 16) - return 1; + int m_framenum; /* debug frame num counter; unlike iFrameNum, does not reset at ivop */ - else if (sr <= 32) - return 2; - else if (sr <= 64) - return 3; - else if (sr <= 128) - return 4; + float fMvPrevSigma; - else if (sr <= 256) - return 5; +} Encoder; - else if (sr <= 512) - return 6; - else if (sr <= 1024) - return 7; +/***************************************************************************** + * Inline functions + ****************************************************************************/ - else - return 0; -} +extern uint8_t get_fcode(uint16_t sr) +; /***************************************************************************** diff -bBru src-ori/global.h src/global.h --- src-ori/global.h Wed Jan 12 10:32:43 2005 +++ src/global.h Wed Jan 12 12:38:11 2005 @@ -60,16 +60,16 @@ #define N_VOP 4 /* convert mpeg-4 coding type i/p/b/s_VOP to XVID_TYPE_xxx */ -static __inline int -coding2type(int coding_type) -{ +static __inline int coding2type(int coding_type) + { + return coding_type + 1; } /* convert XVID_TYPE_xxx to bitstream coding type i/p/b/s_VOP */ -static __inline int -type2coding(int xvid_type) -{ +static __inline int type2coding(int xvid_type) + { + return xvid_type - 1; } @@ -188,48 +274,86 @@ typedef struct { + /* decoder/encoder */ + VECTOR mvs[4]; + + short int pred_values[6][MBPRED_SIZE]; + int acpred_directions[6]; + + int mode; + int quant; /* absolute quant */ + + int field_dct; + int field_pred; + int field_for_top; + int field_for_bot; + + /* encoder specific */ + + VECTOR pmvs[4]; + VECTOR qmvs[4]; /* mvs in quarter pixel resolution */ + + int32_t sad8[4]; /* SAD values for inter4v-VECTORs */ + int32_t sad16; /* SAD value for inter-VECTOR */ + + int dquant; + int cbp; + + /* bframe stuff */ + + VECTOR b_mvs[4]; + VECTOR b_qmvs[4]; + + VECTOR amv; /* average motion vectors from GMC */ + int32_t mcsel; + + /* This structure has become way to big! What to do? Split it up? */ + + } + MACROBLOCK; -static __inline uint32_t -get_dc_scaler(uint32_t quant, + +static __inline uint32_t get_dc_scaler(uint32_t quant, uint32_t lum) -{ + { + if (quant < 5) return 8; Only in src: global_all_dll.c Only in src: global_all_dll.h diff -bBru src-ori/image/colorspace.c src/image/colorspace.c --- src-ori/image/colorspace.c Wed Jan 12 10:32:43 2005 +++ src/image/colorspace.c Wed Jan 12 12:39:41 2005 @@ -23,67 +23,18 @@ * ****************************************************************************/ -#include /* memcpy */ +#include +#include "../global_all_dll.h" -#include "../global.h" -#include "colorspace.h" /* function pointers */ -/* input */ -packedFuncPtr rgb555_to_yv12; -packedFuncPtr rgb565_to_yv12; -packedFuncPtr rgb444_to_yv12; -packedFuncPtr bgr_to_yv12; -packedFuncPtr bgra_to_yv12; -packedFuncPtr abgr_to_yv12; -packedFuncPtr rgba_to_yv12; -packedFuncPtr argb_to_yv12; -packedFuncPtr yuv_to_yv12; -packedFuncPtr yuyv_to_yv12; -packedFuncPtr uyvy_to_yv12; - -packedFuncPtr rgb555i_to_yv12; -packedFuncPtr rgb565i_to_yv12; -packedFuncPtr bgri_to_yv12; -packedFuncPtr bgrai_to_yv12; -packedFuncPtr abgri_to_yv12; -packedFuncPtr rgbai_to_yv12; -packedFuncPtr argbi_to_yv12; -packedFuncPtr yuyvi_to_yv12; -packedFuncPtr uyvyi_to_yv12; - -/* output */ -packedFuncPtr yv12_to_rgb555; -packedFuncPtr yv12_to_rgb565; -packedFuncPtr yv12_to_rgb444; -packedFuncPtr yv12_to_bgr; -packedFuncPtr yv12_to_bgra; -packedFuncPtr yv12_to_abgr; -packedFuncPtr yv12_to_rgba; -packedFuncPtr yv12_to_argb; -packedFuncPtr yv12_to_yuv; -packedFuncPtr yv12_to_yuyv; -packedFuncPtr yv12_to_uyvy; - -packedFuncPtr yv12_to_rgb555i; -packedFuncPtr yv12_to_rgb565i; -packedFuncPtr yv12_to_bgri; -packedFuncPtr yv12_to_bgrai; -packedFuncPtr yv12_to_abgri; -packedFuncPtr yv12_to_rgbai; -packedFuncPtr yv12_to_argbi; -packedFuncPtr yv12_to_yuyvi; -packedFuncPtr yv12_to_uyvyi; - -planarFuncPtr yv12_to_yv12; - - -int32_t RGB_Y_tab[256]; -int32_t B_U_tab[256]; -int32_t G_U_tab[256]; -int32_t G_V_tab[256]; -int32_t R_V_tab[256]; + + + + + + @@ -125,15 +76,16 @@ + /********** colorspace input (xxx_to_yv12) functions **********/ -/* rgb -> yuv def's + + +/* rgb -> yuv def's this following constants are "official spec" Video Demystified" (ISBN 1-878707-09-4) - rgb<->yuv _is_ lossy, since most programs do the conversion differently - SCALEBITS/FIX taken from ffmpeg */ @@ -349,13 +321,14 @@ ((MAX(0,MIN(255, G)) ) & 0x00f0) | \ ((MAX(0,MIN(255, B)) >> 4) & 0x000f) + #define WRITE_RGB16(ROW,UV_ROW,C1) \ - rgb_y = RGB_Y_tab[ y_ptr[y_stride*(ROW) + 0] ]; \ + rgb_y = (((struct global_all_dll *)Dll::Tls())->RGB_Y_tab)[ y_ptr[y_stride*(ROW) + 0] ]; \ b[ROW] = (b[ROW] & 0x7) + ((rgb_y + b_u##UV_ROW) >> SCALEBITS_OUT); \ g[ROW] = (g[ROW] & 0x7) + ((rgb_y - g_uv##UV_ROW) >> SCALEBITS_OUT); \ r[ROW] = (r[ROW] & 0x7) + ((rgb_y + r_v##UV_ROW) >> SCALEBITS_OUT); \ *(uint16_t *) (x_ptr+((ROW)*x_stride)+0) = C1(r[ROW], g[ROW], b[ROW]); \ - rgb_y = RGB_Y_tab[ y_ptr[y_stride*(ROW) + 1] ]; \ + rgb_y = (((struct global_all_dll *)Dll::Tls())->RGB_Y_tab)[ y_ptr[y_stride*(ROW) + 1] ]; \ b[ROW] = (b[ROW] & 0x7) + ((rgb_y + b_u##UV_ROW) >> SCALEBITS_OUT); \ g[ROW] = (g[ROW] & 0x7) + ((rgb_y - g_uv##UV_ROW) >> SCALEBITS_OUT); \ r[ROW] = (r[ROW] & 0x7) + ((rgb_y + r_v##UV_ROW) >> SCALEBITS_OUT); \ @@ -364,11 +338,12 @@ #define YV12_TO_RGB16_ROW(SIZE,C1,C2,C3,C4) \ int r[2], g[2], b[2]; \ r[0] = r[1] = g[0] = g[1] = b[0] = b[1] = 0; + #define YV12_TO_RGB16(SIZE,C1,C2,C3,C4) \ int rgb_y; \ - int b_u0 = B_U_tab[ u_ptr[0] ]; \ - int g_uv0 = G_U_tab[ u_ptr[0] ] + G_V_tab[ v_ptr[0] ]; \ - int r_v0 = R_V_tab[ v_ptr[0] ]; \ + int b_u0 = (((struct global_all_dll *)Dll::Tls())->B_U_tab)[ u_ptr[0] ]; \ + int g_uv0 = (((struct global_all_dll *)Dll::Tls())->G_U_tab)[ u_ptr[0] ] + (((struct global_all_dll *)Dll::Tls())->G_V_tab)[ v_ptr[0] ]; \ + int r_v0 = (((struct global_all_dll *)Dll::Tls())->R_V_tab)[ v_ptr[0] ]; \ WRITE_RGB16(0, 0, C1) \ WRITE_RGB16(1, 0, C1) @@ -377,14 +353,15 @@ r[0] = r[1] = r[2] = r[3] = 0; \ g[0] = g[1] = g[2] = g[3] = 0; \ b[0] = b[1] = b[2] = b[3] = 0; + #define YV12_TO_RGB16I(SIZE,C1,C2,C3,C4) \ int rgb_y; \ - int b_u0 = B_U_tab[ u_ptr[0] ]; \ - int g_uv0 = G_U_tab[ u_ptr[0] ] + G_V_tab[ v_ptr[0] ]; \ - int r_v0 = R_V_tab[ v_ptr[0] ]; \ - int b_u1 = B_U_tab[ u_ptr[uv_stride] ]; \ - int g_uv1 = G_U_tab[ u_ptr[uv_stride] ] + G_V_tab[ v_ptr[uv_stride] ]; \ - int r_v1 = R_V_tab[ v_ptr[uv_stride] ]; \ + int b_u0 = (((struct global_all_dll *)Dll::Tls())->B_U_tab)[ u_ptr[0] ]; \ + int g_uv0 = (((struct global_all_dll *)Dll::Tls())->G_U_tab)[ u_ptr[0] ] + (((struct global_all_dll *)Dll::Tls())->G_V_tab)[ v_ptr[0] ]; \ + int r_v0 = (((struct global_all_dll *)Dll::Tls())->R_V_tab)[ v_ptr[0] ]; \ + int b_u1 = (((struct global_all_dll *)Dll::Tls())->B_U_tab)[ u_ptr[uv_stride] ]; \ + int g_uv1 = (((struct global_all_dll *)Dll::Tls())->G_U_tab)[ u_ptr[uv_stride] ] + (((struct global_all_dll *)Dll::Tls())->G_V_tab)[ v_ptr[uv_stride] ]; \ + int r_v1 = (((struct global_all_dll *)Dll::Tls())->R_V_tab)[ v_ptr[uv_stride] ]; \ WRITE_RGB16(0, 0, C1) \ WRITE_RGB16(1, 1, C1) \ WRITE_RGB16(2, 0, C1) \ @@ -394,12 +372,12 @@ /* rgb/rgbi output */ #define WRITE_RGB(SIZE,ROW,UV_ROW,C1,C2,C3,C4) \ - rgb_y = RGB_Y_tab[ y_ptr[(ROW)*y_stride + 0] ]; \ + rgb_y = (((struct global_all_dll *)Dll::Tls())->RGB_Y_tab)[ y_ptr[(ROW)*y_stride + 0] ]; \ x_ptr[(ROW)*x_stride+(C3)] = MAX(0, MIN(255, (rgb_y + b_u##UV_ROW) >> SCALEBITS_OUT)); \ x_ptr[(ROW)*x_stride+(C2)] = MAX(0, MIN(255, (rgb_y - g_uv##UV_ROW) >> SCALEBITS_OUT)); \ x_ptr[(ROW)*x_stride+(C1)] = MAX(0, MIN(255, (rgb_y + r_v##UV_ROW) >> SCALEBITS_OUT)); \ if ((SIZE)>3) x_ptr[(ROW)*x_stride+(C4)] = 0; \ - rgb_y = RGB_Y_tab[ y_ptr[(ROW)*y_stride + 1] ]; \ + rgb_y = (((struct global_all_dll *)Dll::Tls())->RGB_Y_tab)[ y_ptr[(ROW)*y_stride + 1] ]; \ x_ptr[(ROW)*x_stride+(SIZE)+(C3)] = MAX(0, MIN(255, (rgb_y + b_u##UV_ROW) >> SCALEBITS_OUT)); \ x_ptr[(ROW)*x_stride+(SIZE)+(C2)] = MAX(0, MIN(255, (rgb_y - g_uv##UV_ROW) >> SCALEBITS_OUT)); \ x_ptr[(ROW)*x_stride+(SIZE)+(C1)] = MAX(0, MIN(255, (rgb_y + r_v##UV_ROW) >> SCALEBITS_OUT)); \ @@ -409,9 +388,9 @@ #define YV12_TO_RGB_ROW(SIZE,C1,C2,C3,C4) /* nothing */ #define YV12_TO_RGB(SIZE,C1,C2,C3,C4) \ int rgb_y; \ - int b_u0 = B_U_tab[ u_ptr[0] ]; \ - int g_uv0 = G_U_tab[ u_ptr[0] ] + G_V_tab[ v_ptr[0] ]; \ - int r_v0 = R_V_tab[ v_ptr[0] ]; \ + int b_u0 = (((struct global_all_dll *)Dll::Tls())->B_U_tab)[ u_ptr[0] ]; \ + int g_uv0 = (((struct global_all_dll *)Dll::Tls())->G_U_tab)[ u_ptr[0] ] + (((struct global_all_dll *)Dll::Tls())->G_V_tab)[ v_ptr[0] ]; \ + int r_v0 = (((struct global_all_dll *)Dll::Tls())->R_V_tab)[ v_ptr[0] ]; \ WRITE_RGB(SIZE, 0, 0, C1,C2,C3,C4) \ WRITE_RGB(SIZE, 1, 0, C1,C2,C3,C4) @@ -418,12 +398,12 @@ #define YV12_TO_RGBI_ROW(SIZE,C1,C2,C3,C4) /* nothing */ #define YV12_TO_RGBI(SIZE,C1,C2,C3,C4) \ int rgb_y; \ - int b_u0 = B_U_tab[ u_ptr[0] ]; \ - int g_uv0 = G_U_tab[ u_ptr[0] ] + G_V_tab[ v_ptr[0] ]; \ - int r_v0 = R_V_tab[ v_ptr[0] ]; \ - int b_u1 = B_U_tab[ u_ptr[uv_stride] ]; \ - int g_uv1 = G_U_tab[ u_ptr[uv_stride] ] + G_V_tab[ v_ptr[uv_stride] ]; \ - int r_v1 = R_V_tab[ v_ptr[uv_stride] ]; \ + int b_u0 = (((struct global_all_dll *)Dll::Tls())->B_U_tab)[ u_ptr[0] ]; \ + int g_uv0 = (((struct global_all_dll *)Dll::Tls())->G_U_tab)[ u_ptr[0] ] + (((struct global_all_dll *)Dll::Tls())->G_V_tab)[ v_ptr[0] ]; \ + int r_v0 = (((struct global_all_dll *)Dll::Tls())->R_V_tab)[ v_ptr[0] ]; \ + int b_u1 = (((struct global_all_dll *)Dll::Tls())->B_U_tab)[ u_ptr[uv_stride] ]; \ + int g_uv1 = (((struct global_all_dll *)Dll::Tls())->G_U_tab)[ u_ptr[uv_stride] ] + (((struct global_all_dll *)Dll::Tls())->G_V_tab)[ v_ptr[uv_stride] ]; \ + int r_v1 = (((struct global_all_dll *)Dll::Tls())->R_V_tab)[ v_ptr[uv_stride] ]; \ WRITE_RGB(SIZE, 0, 0, C1,C2,C3,C4) \ WRITE_RGB(SIZE, 1, 1, C1,C2,C3,C4) \ WRITE_RGB(SIZE, 2, 0, C1,C2,C3,C4) \ @@ -524,10 +508,11 @@ int32_t i; for (i = 0; i < 256; i++) { - RGB_Y_tab[i] = FIX_OUT(RGB_Y_OUT) * (i - Y_ADD_OUT); - B_U_tab[i] = FIX_OUT(B_U_OUT) * (i - U_ADD_OUT); - G_U_tab[i] = FIX_OUT(G_U_OUT) * (i - U_ADD_OUT); - G_V_tab[i] = FIX_OUT(G_V_OUT) * (i - V_ADD_OUT); - R_V_tab[i] = FIX_OUT(R_V_OUT) * (i - V_ADD_OUT); + (((struct global_all_dll *)Dll::Tls())->RGB_Y_tab)[i] = FIX_OUT(RGB_Y_OUT) * (i - Y_ADD_OUT); + (((struct global_all_dll *)Dll::Tls())->B_U_tab)[i] = FIX_OUT(B_U_OUT) * (i - U_ADD_OUT); + (((struct global_all_dll *)Dll::Tls())->G_U_tab)[i] = FIX_OUT(G_U_OUT) * (i - U_ADD_OUT); + (((struct global_all_dll *)Dll::Tls())->G_V_tab)[i] = FIX_OUT(G_V_OUT) * (i - V_ADD_OUT); + (((struct global_all_dll *)Dll::Tls())->R_V_tab)[i] = FIX_OUT(R_V_OUT) * (i - V_ADD_OUT); } } + diff -bBru src-ori/image/colorspace.h src/image/colorspace.h --- src-ori/image/colorspace.h Wed Jan 12 10:32:43 2005 +++ src/image/colorspace.h Wed Jan 12 12:37:50 2005 @@ -52,26 +61,26 @@ /* xxx_to_yv12 colorspace conversion functions (encoder) */ -extern packedFuncPtr rgb555_to_yv12; -extern packedFuncPtr rgb565_to_yv12; -extern packedFuncPtr rgb444_to_yv12; -extern packedFuncPtr bgr_to_yv12; -extern packedFuncPtr bgra_to_yv12; -extern packedFuncPtr abgr_to_yv12; -extern packedFuncPtr rgba_to_yv12; -extern packedFuncPtr argb_to_yv12; -extern packedFuncPtr yuyv_to_yv12; -extern packedFuncPtr uyvy_to_yv12; - -extern packedFuncPtr rgb555i_to_yv12; -extern packedFuncPtr rgb565i_to_yv12; -extern packedFuncPtr bgri_to_yv12; -extern packedFuncPtr bgrai_to_yv12; -extern packedFuncPtr abgri_to_yv12; -extern packedFuncPtr rgbai_to_yv12; -extern packedFuncPtr argbi_to_yv12; -extern packedFuncPtr yuyvi_to_yv12; -extern packedFuncPtr uyvyi_to_yv12; + + + + + + + + + + + + + + + + + + + + /* plain c */ packedFunc rgb555_to_yv12_c; @@ -114,26 +123,26 @@ /* yv12_to_xxx colorspace conversion functions (decoder) */ -extern packedFuncPtr yv12_to_rgb555; -extern packedFuncPtr yv12_to_rgb565; -extern packedFuncPtr yv12_to_rgb444; -extern packedFuncPtr yv12_to_bgr; -extern packedFuncPtr yv12_to_bgra; -extern packedFuncPtr yv12_to_abgr; -extern packedFuncPtr yv12_to_rgba; -extern packedFuncPtr yv12_to_argb; -extern packedFuncPtr yv12_to_yuyv; -extern packedFuncPtr yv12_to_uyvy; - -extern packedFuncPtr yv12_to_rgb555i; -extern packedFuncPtr yv12_to_rgb565i; -extern packedFuncPtr yv12_to_bgri; -extern packedFuncPtr yv12_to_bgrai; -extern packedFuncPtr yv12_to_abgri; -extern packedFuncPtr yv12_to_rgbai; -extern packedFuncPtr yv12_to_argbi; -extern packedFuncPtr yv12_to_yuyvi; -extern packedFuncPtr yv12_to_uyvyi; + + + + + + + + + + + + + + + + + + + + /* plain c */ packedFunc yv12_to_rgb555_c; @@ -171,13 +180,18 @@ typedef void (planarFunc) ( uint8_t * y_dst, uint8_t * u_dst, uint8_t * v_dst, + int y_dst_stride, int uv_dst_stride, + uint8_t * y_src, uint8_t * u_src, uint8_t * v_src, + int y_src_stride, int uv_src_stride, + int width, int height, int vflip); + typedef planarFunc *planarFuncPtr; -extern planarFuncPtr yv12_to_yv12; + planarFunc yv12_to_yv12_c; diff -bBru src-ori/image/font.c src/image/font.c --- src-ori/image/font.c Wed Jan 12 10:32:43 2005 +++ src/image/font.c Wed Jan 12 12:40:20 2005 @@ -23,11 +23,9 @@ * ****************************************************************************/ -#include -#include +#include +#include "../global_all_dll.h" -#include "image.h" -#include "font.h" #define FONT_WIDTH 4 #define FONT_HEIGHT 6 @@ -295,7 +293,7 @@ static const char ascii65[26][FONT_WIDTH*FONT_HEIGHT] = { - /* A */ + {0,1,1,0, 1,0,0,1, 1,0,0,1, @@ -303,7 +301,7 @@ 1,0,0,1, 1,0,0,1}, - /* B */ + {1,1,1,0, 1,0,0,1, 1,1,1,0, @@ -311,7 +309,7 @@ 1,0,0,1, 1,1,1,0}, - /* C */ + {0,1,1,0, 1,0,0,1, 1,0,0,0, @@ -319,7 +317,7 @@ 1,0,0,1, 0,1,1,0}, - /* D */ + {1,1,0,0, 1,0,1,0, 1,0,0,1, @@ -327,7 +325,7 @@ 1,0,1,0, 1,1,0,0}, - /* E */ + {1,1,1,1, 1,0,0,0, 1,1,1,0, @@ -335,7 +333,7 @@ 1,0,0,0, 1,1,1,1}, - /* F */ + {1,1,1,1, 1,0,0,0, 1,1,1,0, @@ -343,7 +341,7 @@ 1,0,0,0, 1,0,0,0}, - /* G */ + {0,1,1,1, 1,0,0,0, 1,0,1,1, @@ -351,7 +349,7 @@ 1,0,0,1, 0,1,1,0}, - /* H */ + {1,0,0,1, 1,0,0,1, 1,1,1,1, @@ -359,7 +357,7 @@ 1,0,0,1, 1,0,0,1}, - /* I */ + {0,1,1,1, 0,0,1,0, 0,0,1,0, @@ -367,7 +365,7 @@ 0,0,1,0, 0,1,1,1}, - /* J */ + {0,1,1,1, 0,0,1,0, 0,0,1,0, @@ -375,7 +373,7 @@ 1,0,1,0, 0,1,0,0}, - /* K */ + {1,0,0,1, 1,0,0,1, 1,1,1,0, @@ -383,7 +381,7 @@ 1,0,0,1, 1,0,0,1}, - /* L */ + {1,0,0,0, 1,0,0,0, 1,0,0,0, @@ -391,7 +389,7 @@ 1,0,0,0, 1,1,1,1}, - /* M */ + {1,0,0,1, 1,1,1,1, 1,1,1,1, @@ -399,7 +397,7 @@ 1,0,0,1, 1,0,0,1}, - /* N */ + {1,0,0,1, 1,1,0,1, 1,1,0,1, @@ -407,7 +405,7 @@ 1,0,1,1, 1,0,0,1}, - /* 0 */ + {0,1,1,0, 1,0,0,1, 1,0,0,1, @@ -415,7 +413,7 @@ 1,0,0,1, 0,1,1,0}, - /* P */ + {1,1,1,0, 1,0,0,1, 1,1,1,0, @@ -423,7 +421,7 @@ 1,0,0,0, 1,0,0,0}, - /* Q */ + {0,1,1,0, 1,0,0,1, 1,0,0,1, @@ -432,7 +430,7 @@ 0,1,0,1}, - /* R */ + {1,1,1,0, 1,0,0,1, 1,1,1,0, @@ -440,7 +438,7 @@ 1,0,0,1, 1,0,0,1}, - /* S */ + {0,1,1,0, 1,0,0,1, 0,1,0,0, @@ -448,7 +446,7 @@ 1,0,0,1, 0,1,1,0}, - /* T */ + {0,1,1,1, 0,0,1,0, 0,0,1,0, @@ -456,7 +454,7 @@ 0,0,1,0, 0,0,1,0}, - /* U */ + {1,0,0,1, 1,0,0,1, 1,0,0,1, @@ -464,7 +462,7 @@ 1,0,0,1, 1,1,1,1}, - /* V */ + {1,0,0,1, 1,0,0,1, 1,0,0,1, @@ -472,7 +470,7 @@ 0,1,1,0, 0,1,1,0}, - /* W */ + {1,0,0,1, 1,0,0,1, 1,0,0,1, @@ -480,7 +478,7 @@ 1,1,1,1, 1,0,0,1}, - /* X */ + {1,0,0,1, 1,0,0,1, 0,1,1,0, @@ -488,7 +486,7 @@ 1,0,0,1, 1,0,0,1}, - /* Y */ + {1,0,0,1, 1,0,0,1, 0,1,0,0, @@ -496,7 +494,7 @@ 0,1,0,0, 1,0,0,0}, - /* Z */ + {1,1,1,1, 0,0,0,1, 0,0,1,0, @@ -509,7 +507,7 @@ static const char ascii91[6][FONT_WIDTH*FONT_HEIGHT] = { - /* [ */ + {0,1,1,0, 0,1,0,0, 0,1,0,0, @@ -517,7 +515,7 @@ 0,1,0,0, 0,1,1,0}, - /* '\' */ + {1,0,0,0, 1,0,0,0, 0,1,0,0, @@ -525,7 +523,7 @@ 0,0,0,1, 0,0,0,1}, - /* ] */ + {0,1,1,0, 0,0,1,0, 0,0,1,0, @@ -533,7 +531,7 @@ 0,0,1,0, 0,1,1,0}, - /* ^ */ + {0,1,0,1, 0,0,0,0, 0,0,0,0, @@ -541,7 +539,7 @@ 0,0,0,0, 0,0,0,0}, - /* _ */ + {0,0,0,0, 0,0,0,0, 0,0,0,0, @@ -549,7 +547,7 @@ 0,0,0,0, 1,1,1,1}, - /* ` */ + {0,1,0,0, 0,0,1,0, 0,0,0,0, Only in src-ori/image: ia64_asm diff -bBru src-ori/image/image.c src/image/image.c --- src-ori/image/image.c Wed Jan 12 10:32:43 2005 +++ src/image/image.c Wed Jan 12 12:41:16 2005 @@ -23,20 +23,16 @@ * ****************************************************************************/ -#include -#include /* memcpy, memset */ -#include - -#include "../portab.h" -#include "../global.h" /* XVID_CSP_XXX's */ -#include "../xvid.h" /* XVID_CSP_XXX's */ -#include "image.h" -#include "colorspace.h" -#include "interpolate8x8.h" -#include "reduced.h" -#include "../utils/mem_align.h" +#include +#include "../global_all_dll.h" + image_null(IMAGE * image) + { + + image->y = image->u = image->v = NULL; +} + + -#include "font.h" /* XXX: remove later */ #define SAFETY 64 #define EDGE_SIZE2 (EDGE_SIZE/2) @@ -238,6 +234,12 @@ } /* bframe encoding requires image-based u,v interpolation */ +#if 0 + const uint32_t edged_width2 = edged_width / 2; + const uint32_t edged_height2 = edged_height / 2; + const uint32_t offset2 = EDGE_SIZE2 * (edged_width2 + 1); + const uint32_t stride_add2 = 7 * edged_width2; +#endif void image_interpolate(const IMAGE * refn, IMAGE * refh, @@ -250,12 +252,6 @@ { const uint32_t offset = EDGE_SIZE2 * (edged_width + 1); /* we only interpolate half of the edge area */ const uint32_t stride_add = 7 * edged_width; -#if 0 - const uint32_t edged_width2 = edged_width / 2; - const uint32_t edged_height2 = edged_height / 2; - const uint32_t offset2 = EDGE_SIZE2 * (edged_width2 + 1); - const uint32_t stride_add2 = 7 * edged_width2; -#endif uint8_t *n_ptr, *h_ptr, *v_ptr, *hv_ptr; uint32_t x, y; @@ -274,8 +270,8 @@ for (y = 0; y < (edged_height - EDGE_SIZE); y += 8) { for (x = 0; x < (edged_width - EDGE_SIZE); x += 8) { - interpolate8x8_6tap_lowpass_h(h_ptr, n_ptr, edged_width, rounding); - interpolate8x8_6tap_lowpass_v(v_ptr, n_ptr, edged_width, rounding); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_6tap_lowpass_h)(h_ptr, n_ptr, edged_width, rounding); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_6tap_lowpass_v)(v_ptr, n_ptr, edged_width, rounding); n_ptr += 8; h_ptr += 8; @@ -303,7 +299,7 @@ for (x = 0; x < (edged_width - EDGE_SIZE); x = x + 8) { hv_ptr -= 8; h_ptr -= 8; - interpolate8x8_6tap_lowpass_v(hv_ptr, h_ptr, edged_width, rounding); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_6tap_lowpass_v)(hv_ptr, h_ptr, edged_width, rounding); } } } else { @@ -313,9 +309,9 @@ for (y = 0; y < (edged_height - EDGE_SIZE); y += 8) { for (x = 0; x < (edged_width - EDGE_SIZE); x += 8) { - interpolate8x8_halfpel_h(h_ptr, n_ptr, edged_width, rounding); - interpolate8x8_halfpel_v(v_ptr, n_ptr, edged_width, rounding); - interpolate8x8_halfpel_hv(hv_ptr, n_ptr, edged_width, rounding); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_halfpel_h)(h_ptr, n_ptr, edged_width, rounding); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_halfpel_v)(v_ptr, n_ptr, edged_width, rounding); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_halfpel_hv)(hv_ptr, n_ptr, edged_width, rounding); n_ptr += 8; h_ptr += 8; @@ -348,9 +344,9 @@ for (y = 0; y < edged_height2; y += 8) { for (x = 0; x < edged_width2; x += 8) { - interpolate8x8_halfpel_h(h_ptr, n_ptr, edged_width2, rounding); - interpolate8x8_halfpel_v(v_ptr, n_ptr, edged_width2, rounding); - interpolate8x8_halfpel_hv(hv_ptr, n_ptr, edged_width2, rounding); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_halfpel_h)(h_ptr, n_ptr, edged_width2, rounding); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_halfpel_v)(v_ptr, n_ptr, edged_width2, rounding); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_halfpel_hv)(hv_ptr, n_ptr, edged_width2, rounding); n_ptr += 8; h_ptr += 8; @@ -375,9 +371,9 @@ for (y = 0; y < edged_height2; y = y + 8) { for (x = 0; x < edged_width2; x = x + 8) { - interpolate8x8_halfpel_h(h_ptr, n_ptr, edged_width2, rounding); - interpolate8x8_halfpel_v(v_ptr, n_ptr, edged_width2, rounding); - interpolate8x8_halfpel_hv(hv_ptr, n_ptr, edged_width2, rounding); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_halfpel_h)(h_ptr, n_ptr, edged_width2, rounding); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_halfpel_v)(v_ptr, n_ptr, edged_width2, rounding); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_halfpel_hv)(hv_ptr, n_ptr, edged_width2, rounding); n_ptr += 8; h_ptr += 8; @@ -460,6 +456,10 @@ correpsonding luma pixels are pure black or white. */ +#define IS_PURE(a) ((a)<=16||(a)>=235) +#define IMG_Y(Y,X) img->y[(Y)*edged_width + (X)] +#define IMG_U(Y,X) img->u[(Y)*edged_width/2 + (X)] +#define IMG_V(Y,X) img->v[(Y)*edged_width/2 + (X)] void image_chroma_optimize(IMAGE * img, int width, int height, int edged_width) { @@ -469,10 +469,6 @@ for (y = 1; y < height/2 - 1; y++) for (x = 1; x < width/2 - 1; x++) { -#define IS_PURE(a) ((a)<=16||(a)>=235) -#define IMG_Y(Y,X) img->y[(Y)*edged_width + (X)] -#define IMG_U(Y,X) img->u[(Y)*edged_width/2 + (X)] -#define IMG_V(Y,X) img->v[(Y)*edged_width/2 + (X)] if (IS_PURE(IMG_Y(y*2 ,x*2 )) && IS_PURE(IMG_Y(y*2 ,x*2+1)) && @@ -537,6 +533,9 @@ +#if 0 + const int height_signed = (csp & XVID_CSP_VFLIP) ? -height : height; +#endif int image_input(IMAGE * image, uint32_t width, @@ -550,16 +549,13 @@ const int edged_width2 = edged_width/2; const int width2 = width/2; const int height2 = height/2; -#if 0 - const int height_signed = (csp & XVID_CSP_VFLIP) ? -height : height; -#endif switch (csp & ~XVID_CSP_VFLIP) { case XVID_CSP_RGB555: safe_packed_conv( src[0], src_stride[0], image->y, image->u, image->v, edged_width, edged_width2, width, height, (csp & XVID_CSP_VFLIP), - interlacing?rgb555i_to_yv12 :rgb555_to_yv12, + interlacing?(((struct global_all_dll *)Dll::Tls())->rgb555i_to_yv12) :(((struct global_all_dll *)Dll::Tls())->rgb555_to_yv12), interlacing?rgb555i_to_yv12_c:rgb555_to_yv12_c, 2); break; @@ -567,7 +563,7 @@ safe_packed_conv( src[0], src_stride[0], image->y, image->u, image->v, edged_width, edged_width2, width, height, (csp & XVID_CSP_VFLIP), - interlacing?rgb565i_to_yv12 :rgb565_to_yv12, + interlacing?(((struct global_all_dll *)Dll::Tls())->rgb565i_to_yv12) :(((struct global_all_dll *)Dll::Tls())->rgb565_to_yv12), interlacing?rgb565i_to_yv12_c:rgb565_to_yv12_c, 2); break; @@ -575,7 +571,7 @@ safe_packed_conv( src[0], src_stride[0], image->y, image->u, image->v, edged_width, edged_width2, width, height, (csp & XVID_CSP_VFLIP), - rgb444_to_yv12, + (((struct global_all_dll *)Dll::Tls())->rgb444_to_yv12), rgb444_to_yv12_c, 2); break; @@ -584,7 +580,7 @@ safe_packed_conv( src[0], src_stride[0], image->y, image->u, image->v, edged_width, edged_width2, width, height, (csp & XVID_CSP_VFLIP), - interlacing?bgri_to_yv12 :bgr_to_yv12, + interlacing?(((struct global_all_dll *)Dll::Tls())->bgri_to_yv12) :(((struct global_all_dll *)Dll::Tls())->bgr_to_yv12), interlacing?bgri_to_yv12_c:bgr_to_yv12_c, 3); break; @@ -592,7 +588,7 @@ safe_packed_conv( src[0], src_stride[0], image->y, image->u, image->v, edged_width, edged_width2, width, height, (csp & XVID_CSP_VFLIP), - interlacing?bgrai_to_yv12 :bgra_to_yv12, + interlacing?(((struct global_all_dll *)Dll::Tls())->bgrai_to_yv12) :(((struct global_all_dll *)Dll::Tls())->bgra_to_yv12), interlacing?bgrai_to_yv12_c:bgra_to_yv12_c, 4); break; @@ -600,7 +596,7 @@ safe_packed_conv( src[0], src_stride[0], image->y, image->u, image->v, edged_width, edged_width2, width, height, (csp & XVID_CSP_VFLIP), - interlacing?abgri_to_yv12 :abgr_to_yv12, + interlacing?(((struct global_all_dll *)Dll::Tls())->abgri_to_yv12) :(((struct global_all_dll *)Dll::Tls())->abgr_to_yv12), interlacing?abgri_to_yv12_c:abgr_to_yv12_c, 4); break; @@ -608,7 +604,7 @@ safe_packed_conv( src[0], src_stride[0], image->y, image->u, image->v, edged_width, edged_width2, width, height, (csp & XVID_CSP_VFLIP), - interlacing?rgbai_to_yv12 :rgba_to_yv12, + interlacing?(((struct global_all_dll *)Dll::Tls())->rgbai_to_yv12) :(((struct global_all_dll *)Dll::Tls())->rgba_to_yv12), interlacing?rgbai_to_yv12_c:rgba_to_yv12_c, 4); break; @@ -616,7 +612,7 @@ safe_packed_conv( src[0], src_stride[0], image->y, image->u, image->v, edged_width, edged_width2, width, height, (csp & XVID_CSP_VFLIP), - interlacing?argbi_to_yv12 : argb_to_yv12, + interlacing?(((struct global_all_dll *)Dll::Tls())->argbi_to_yv12) : (((struct global_all_dll *)Dll::Tls())->argb_to_yv12), interlacing?argbi_to_yv12_c: argb_to_yv12_c, 4); break; @@ -624,7 +620,7 @@ safe_packed_conv( src[0], src_stride[0], image->y, image->u, image->v, edged_width, edged_width2, width, height, (csp & XVID_CSP_VFLIP), - interlacing?yuyvi_to_yv12 :yuyv_to_yv12, + interlacing?(((struct global_all_dll *)Dll::Tls())->yuyvi_to_yv12) :(((struct global_all_dll *)Dll::Tls())->yuyv_to_yv12), interlacing?yuyvi_to_yv12_c:yuyv_to_yv12_c, 2); break; @@ -632,7 +628,7 @@ safe_packed_conv( src[0], src_stride[0], image->y, image->v, image->u, edged_width, edged_width2, width, height, (csp & XVID_CSP_VFLIP), - interlacing?yuyvi_to_yv12 :yuyv_to_yv12, + interlacing?(((struct global_all_dll *)Dll::Tls())->yuyvi_to_yv12) :(((struct global_all_dll *)Dll::Tls())->yuyv_to_yv12), interlacing?yuyvi_to_yv12_c:yuyv_to_yv12_c, 2); break; @@ -640,24 +636,24 @@ safe_packed_conv( src[0], src_stride[0], image->y, image->u, image->v, edged_width, edged_width2, width, height, (csp & XVID_CSP_VFLIP), - interlacing?uyvyi_to_yv12 :uyvy_to_yv12, + interlacing?(((struct global_all_dll *)Dll::Tls())->uyvyi_to_yv12) :(((struct global_all_dll *)Dll::Tls())->uyvy_to_yv12), interlacing?uyvyi_to_yv12_c:uyvy_to_yv12_c, 2); break; case XVID_CSP_I420: /* YCbCr == YUV == internal colorspace for MPEG */ - yv12_to_yv12(image->y, image->u, image->v, edged_width, edged_width2, + (((struct global_all_dll *)Dll::Tls())->yv12_to_yv12)(image->y, image->u, image->v, edged_width, edged_width2, src[0], src[0] + src_stride[0]*height, src[0] + src_stride[0]*height + (src_stride[0]/2)*height2, src_stride[0], src_stride[0]/2, width, height, (csp & XVID_CSP_VFLIP)); break; case XVID_CSP_YV12: /* YCrCb == YVA == U and V plane swapped */ - yv12_to_yv12(image->y, image->v, image->u, edged_width, edged_width2, + (((struct global_all_dll *)Dll::Tls())->yv12_to_yv12)(image->y, image->v, image->u, edged_width, edged_width2, src[0], src[0] + src_stride[0]*height, src[0] + src_stride[0]*height + (src_stride[0]/2)*height2, src_stride[0], src_stride[0]/2, width, height, (csp & XVID_CSP_VFLIP)); break; case XVID_CSP_PLANAR: /* YCbCr with arbitrary pointers and different strides for Y and UV */ - yv12_to_yv12(image->y, image->u, image->v, edged_width, edged_width2, + (((struct global_all_dll *)Dll::Tls())->yv12_to_yv12)(image->y, image->u, image->v, edged_width, edged_width2, src[0], src[1], src[2], src_stride[0], src_stride[1], /* v: dst_stride[2] not yet supported */ width, height, (csp & XVID_CSP_VFLIP)); break; @@ -744,7 +740,7 @@ safe_packed_conv( dst[0], dst_stride[0], image->y, image->u, image->v, edged_width, edged_width2, width, height, (csp & XVID_CSP_VFLIP), - interlacing?yv12_to_rgb555i :yv12_to_rgb555, + interlacing?(((struct global_all_dll *)Dll::Tls())->yv12_to_rgb555i) :(((struct global_all_dll *)Dll::Tls())->yv12_to_rgb555), interlacing?yv12_to_rgb555i_c:yv12_to_rgb555_c, 2); return 0; @@ -752,7 +748,7 @@ safe_packed_conv( dst[0], dst_stride[0], image->y, image->u, image->v, edged_width, edged_width2, width, height, (csp & XVID_CSP_VFLIP), - interlacing?yv12_to_rgb565i :yv12_to_rgb565, + interlacing?(((struct global_all_dll *)Dll::Tls())->yv12_to_rgb565i) :(((struct global_all_dll *)Dll::Tls())->yv12_to_rgb565), interlacing?yv12_to_rgb565i_c:yv12_to_rgb565_c, 2); return 0; @@ -760,7 +756,7 @@ safe_packed_conv( dst[0], dst_stride[0], image->y, image->u, image->v, edged_width, edged_width2, width, height, (csp & XVID_CSP_VFLIP), - yv12_to_rgb444, + (((struct global_all_dll *)Dll::Tls())->yv12_to_rgb444), yv12_to_rgb444_c, 2); return 0; @@ -769,7 +765,7 @@ safe_packed_conv( dst[0], dst_stride[0], image->y, image->u, image->v, edged_width, edged_width2, width, height, (csp & XVID_CSP_VFLIP), - interlacing?yv12_to_bgri :yv12_to_bgr, + interlacing?(((struct global_all_dll *)Dll::Tls())->yv12_to_bgri) :(((struct global_all_dll *)Dll::Tls())->yv12_to_bgr), interlacing?yv12_to_bgri_c:yv12_to_bgr_c, 3); return 0; @@ -777,7 +773,7 @@ safe_packed_conv( dst[0], dst_stride[0], image->y, image->u, image->v, edged_width, edged_width2, width, height, (csp & XVID_CSP_VFLIP), - interlacing?yv12_to_bgrai :yv12_to_bgra, + interlacing?(((struct global_all_dll *)Dll::Tls())->yv12_to_bgrai) :(((struct global_all_dll *)Dll::Tls())->yv12_to_bgra), interlacing?yv12_to_bgrai_c:yv12_to_bgra_c, 4); return 0; @@ -785,7 +781,7 @@ safe_packed_conv( dst[0], dst_stride[0], image->y, image->u, image->v, edged_width, edged_width2, width, height, (csp & XVID_CSP_VFLIP), - interlacing?yv12_to_abgri :yv12_to_abgr, + interlacing?(((struct global_all_dll *)Dll::Tls())->yv12_to_abgri) :(((struct global_all_dll *)Dll::Tls())->yv12_to_abgr), interlacing?yv12_to_abgri_c:yv12_to_abgr_c, 4); return 0; @@ -793,7 +789,7 @@ safe_packed_conv( dst[0], dst_stride[0], image->y, image->u, image->v, edged_width, edged_width2, width, height, (csp & XVID_CSP_VFLIP), - interlacing?yv12_to_rgbai :yv12_to_rgba, + interlacing?(((struct global_all_dll *)Dll::Tls())->yv12_to_rgbai) :(((struct global_all_dll *)Dll::Tls())->yv12_to_rgba), interlacing?yv12_to_rgbai_c:yv12_to_rgba_c, 4); return 0; @@ -801,7 +797,7 @@ safe_packed_conv( dst[0], dst_stride[0], image->y, image->u, image->v, edged_width, edged_width2, width, height, (csp & XVID_CSP_VFLIP), - interlacing?yv12_to_argbi :yv12_to_argb, + interlacing?(((struct global_all_dll *)Dll::Tls())->yv12_to_argbi) :(((struct global_all_dll *)Dll::Tls())->yv12_to_argb), interlacing?yv12_to_argbi_c:yv12_to_argb_c, 4); return 0; @@ -809,7 +805,7 @@ safe_packed_conv( dst[0], dst_stride[0], image->y, image->u, image->v, edged_width, edged_width2, width, height, (csp & XVID_CSP_VFLIP), - interlacing?yv12_to_yuyvi :yv12_to_yuyv, + interlacing?(((struct global_all_dll *)Dll::Tls())->yv12_to_yuyvi) :(((struct global_all_dll *)Dll::Tls())->yv12_to_yuyv), interlacing?yv12_to_yuyvi_c:yv12_to_yuyv_c, 2); return 0; @@ -817,7 +813,7 @@ safe_packed_conv( dst[0], dst_stride[0], image->y, image->v, image->u, edged_width, edged_width2, width, height, (csp & XVID_CSP_VFLIP), - interlacing?yv12_to_yuyvi :yv12_to_yuyv, + interlacing?(((struct global_all_dll *)Dll::Tls())->yv12_to_yuyvi) :(((struct global_all_dll *)Dll::Tls())->yv12_to_yuyv), interlacing?yv12_to_yuyvi_c:yv12_to_yuyv_c, 2); return 0; @@ -825,26 +821,26 @@ safe_packed_conv( dst[0], dst_stride[0], image->y, image->u, image->v, edged_width, edged_width2, width, height, (csp & XVID_CSP_VFLIP), - interlacing?yv12_to_uyvyi :yv12_to_uyvy, + interlacing?(((struct global_all_dll *)Dll::Tls())->yv12_to_uyvyi) :(((struct global_all_dll *)Dll::Tls())->yv12_to_uyvy), interlacing?yv12_to_uyvyi_c:yv12_to_uyvy_c, 2); return 0; case XVID_CSP_I420: /* YCbCr == YUV == internal colorspace for MPEG */ - yv12_to_yv12(dst[0], dst[0] + dst_stride[0]*height, dst[0] + dst_stride[0]*height + (dst_stride[0]/2)*height2, + (((struct global_all_dll *)Dll::Tls())->yv12_to_yv12)(dst[0], dst[0] + dst_stride[0]*height, dst[0] + dst_stride[0]*height + (dst_stride[0]/2)*height2, dst_stride[0], dst_stride[0]/2, image->y, image->u, image->v, edged_width, edged_width2, width, height, (csp & XVID_CSP_VFLIP)); return 0; case XVID_CSP_YV12: /* YCrCb == YVU == U and V plane swapped */ - yv12_to_yv12(dst[0], dst[0] + dst_stride[0]*height, dst[0] + dst_stride[0]*height + (dst_stride[0]/2)*height2, + (((struct global_all_dll *)Dll::Tls())->yv12_to_yv12)(dst[0], dst[0] + dst_stride[0]*height, dst[0] + dst_stride[0]*height + (dst_stride[0]/2)*height2, dst_stride[0], dst_stride[0]/2, image->y, image->v, image->u, edged_width, edged_width2, width, height, (csp & XVID_CSP_VFLIP)); return 0; case XVID_CSP_PLANAR: /* YCbCr with arbitrary pointers and different strides for Y and UV */ - yv12_to_yv12(dst[0], dst[1], dst[2], + (((struct global_all_dll *)Dll::Tls())->yv12_to_yv12)(dst[0], dst[1], dst[2], dst_stride[0], dst_stride[1], /* v: dst_stride[2] not yet supported */ image->y, image->u, image->v, edged_width, edged_width2, width, height, (csp & XVID_CSP_VFLIP)); @@ -983,7 +979,6 @@ /* dump image to yuvpgm file */ -#include int image_dump_yuvpgm(const IMAGE * image, @@ -1138,7 +1133,7 @@ if (mbs[(j-1)/2*mb_stride + (i/2)].mode != MODE_NOT_CODED || mbs[(j+0)/2*mb_stride + (i/2)].mode != MODE_NOT_CODED) { - hfilter_31(img->y + (j*block - 1)*edged_width + i*block, + (((struct global_all_dll *)Dll::Tls())->hfilter_31)(img->y + (j*block - 1)*edged_width + i*block, img->y + (j*block + 0)*edged_width + i*block, nblocks); } } @@ -1149,7 +1144,7 @@ if (mbs[(j/2)*mb_stride + (i-1)/2].mode != MODE_NOT_CODED || mbs[(j/2)*mb_stride + (i+0)/2].mode != MODE_NOT_CODED) { - vfilter_31(img->y + (j*block)*edged_width + i*block - 1, + (((struct global_all_dll *)Dll::Tls())->vfilter_31)(img->y + (j*block)*edged_width + i*block - 1, img->y + (j*block)*edged_width + i*block + 0, edged_width, nblocks); } @@ -1165,9 +1160,9 @@ if (mbs[(j-1)*mb_stride + i].mode != MODE_NOT_CODED || mbs[(j+0)*mb_stride + i].mode != MODE_NOT_CODED) { - hfilter_31(img->u + (j*block - 1)*edged_width2 + i*block, + (((struct global_all_dll *)Dll::Tls())->hfilter_31)(img->u + (j*block - 1)*edged_width2 + i*block, img->u + (j*block + 0)*edged_width2 + i*block, nblocks); - hfilter_31(img->v + (j*block - 1)*edged_width2 + i*block, + (((struct global_all_dll *)Dll::Tls())->hfilter_31)(img->v + (j*block - 1)*edged_width2 + i*block, img->v + (j*block + 0)*edged_width2 + i*block, nblocks); } } @@ -1178,10 +1173,10 @@ if (mbs[j*mb_stride + i - 1].mode != MODE_NOT_CODED || mbs[j*mb_stride + i + 0].mode != MODE_NOT_CODED) { - vfilter_31(img->u + (j*block)*edged_width2 + i*block - 1, + (((struct global_all_dll *)Dll::Tls())->vfilter_31)(img->u + (j*block)*edged_width2 + i*block - 1, img->u + (j*block)*edged_width2 + i*block + 0, edged_width2, nblocks); - vfilter_31(img->v + (j*block)*edged_width2 + i*block - 1, + (((struct global_all_dll *)Dll::Tls())->vfilter_31)(img->v + (j*block)*edged_width2 + i*block - 1, img->v + (j*block)*edged_width2 + i*block + 0, edged_width2, nblocks); } diff -bBru src-ori/image/image.h src/image/image.h --- src-ori/image/image.h Wed Jan 12 10:32:43 2005 +++ src/image/image.h Wed Jan 12 12:37:56 2005 @@ -38,11 +38,8 @@ void init_image(uint32_t cpu_flags); -static void __inline -image_null(IMAGE * image) -{ - image->y = image->u = image->v = NULL; -} +extern image_null(IMAGE * image) +; int32_t image_create(IMAGE * image, uint32_t edged_width, diff -bBru src-ori/image/interpolate8x8.c src/image/interpolate8x8.c --- src-ori/image/interpolate8x8.c Wed Jan 12 10:32:43 2005 +++ src/image/interpolate8x8.c Wed Jan 12 12:40:18 2005 @@ -23,29 +23,404 @@ * ****************************************************************************/ -#include "../portab.h" -#include "../global.h" -#include "interpolate8x8.h" +#include +#include "../global_all_dll.h" + void interpolate8x8_switch(uint8_t * const cur, + const uint8_t * const refn, + const uint32_t x, + const uint32_t y, + const int32_t dx, + const int dy, + const uint32_t stride, + const uint32_t rounding) + { + + + const uint8_t * const src = refn + (int)((y + (dy>>1)) * stride + x + (dx>>1)); + uint8_t * const dst = cur + (int)(y * stride + x); + + switch (((dx & 1) << 1) + (dy & 1)) { /* ((dx%2)?2:0)+((dy%2)?1:0) */ + case 0: + (((struct global_all_dll *)Dll::Tls())->transfer8x8_copy)(dst, src, stride); + break; + case 1: + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_halfpel_v)(dst, src, stride, rounding); + break; + case 2: + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_halfpel_h)(dst, src, stride, rounding); + break; + default: + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_halfpel_hv)(dst, src, stride, rounding); + break; + } +} + void interpolate16x16_switch(uint8_t * const cur, + const uint8_t * const refn, + const uint32_t x, + const uint32_t y, + const int32_t dx, + const int dy, + const uint32_t stride, + const uint32_t rounding) + { + + interpolate8x8_switch(cur, refn, x, y, dx, dy, stride, rounding); + interpolate8x8_switch(cur, refn, x+8, y, dx, dy, stride, rounding); + interpolate8x8_switch(cur, refn, x, y+8, dx, dy, stride, rounding); + interpolate8x8_switch(cur, refn, x+8, y+8, dx, dy, stride, rounding); +} + void interpolate32x32_switch(uint8_t * const cur, + const uint8_t * const refn, + const uint32_t x, + const uint32_t y, + const int32_t dx, + const int dy, + const uint32_t stride, + const uint32_t rounding) + { + + interpolate16x16_switch(cur, refn, x, y, dx, dy, stride, rounding); + interpolate16x16_switch(cur, refn, x+16, y, dx, dy, stride, rounding); + interpolate16x16_switch(cur, refn, x, y+16, dx, dy, stride, rounding); + interpolate16x16_switch(cur, refn, x+16, y+16, dx, dy, stride, rounding); +} + uint8_t * interpolate8x8_switch2(uint8_t * const buffer, + const uint8_t * const refn, + const int x, + const int y, + const int dx, + const int dy, + const uint32_t stride, + const uint32_t rounding) + { + + + const uint8_t * const src = refn + (int)((y + (dy>>1)) * stride + x + (dx>>1)); + + switch (((dx & 1) << 1) + (dy & 1)) { /* ((dx%2)?2:0)+((dy%2)?1:0) */ + case 0: + return (uint8_t *)src; + case 1: + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_halfpel_v)(buffer, src, stride, rounding); + break; + case 2: + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_halfpel_h)(buffer, src, stride, rounding); + break; + default: + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_halfpel_hv)(buffer, src, stride, rounding); + break; + } + return buffer; +} + void interpolate8x8_quarterpel(uint8_t * const cur, uint8_t * const refn, + uint8_t * const refh, + uint8_t * const refv, + uint8_t * const refhv, + const uint32_t x, const uint32_t y, + const int32_t dx, const int dy, + const uint32_t stride, + const uint32_t rounding) + { + + const int32_t xRef = (int)x*4 + dx; + const int32_t yRef = (int)y*4 + dy; + + uint8_t *src, *dst; + uint8_t *halfpel_h, *halfpel_v, *halfpel_hv; + int32_t x_int, y_int, x_frac, y_frac; + + x_int = xRef/4; + if (xRef < 0 && xRef % 4) + x_int--; + + x_frac = xRef - (4*x_int); + + y_int = yRef/4; + if (yRef < 0 && yRef % 4) + y_int--; + + y_frac = yRef - (4*y_int); + + src = refn + y_int * (int)stride + x_int; + halfpel_h = refh; + halfpel_v = refv; + halfpel_hv = refhv; + + dst = cur + y * stride + x; + + switch((y_frac << 2) | (x_frac)) { + + case 0: + (((struct global_all_dll *)Dll::Tls())->transfer8x8_copy)(dst, src, stride); + break; + + case 1: + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_lowpass_h)(halfpel_h, src, stride, rounding); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_avg2)(dst, src, halfpel_h, stride, rounding, 8); + break; + + case 2: + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_lowpass_h)(dst, src, stride, rounding); + break; + + case 3: + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_lowpass_h)(halfpel_h, src, stride, rounding); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_avg2)(dst, src + 1, halfpel_h, stride, rounding, 8); + break; + + case 4: + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_lowpass_v)(halfpel_v, src, stride, rounding); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_avg2)(dst, src, halfpel_v, stride, rounding, 8); + break; + + case 5: + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_lowpass_h)(halfpel_h, src, stride, rounding); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_avg2)(halfpel_v, src, halfpel_h, stride, rounding, 9); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_lowpass_v)(halfpel_hv, halfpel_v, stride, rounding); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_avg2)(dst, halfpel_v, halfpel_hv, stride, rounding, 8); + break; + + case 6: + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_lowpass_hv)(halfpel_hv, halfpel_h, src, stride, rounding); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_avg2)(dst, halfpel_h, halfpel_hv, stride, rounding, 8); + break; + + case 7: + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_lowpass_h)(halfpel_h, src, stride, rounding); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_avg2)(halfpel_v, src + 1, halfpel_h, stride, rounding, 9); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_lowpass_v)(halfpel_hv, halfpel_v, stride, rounding); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_avg2)(dst, halfpel_v, halfpel_hv, stride, rounding, 8); + break; + + case 8: + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_lowpass_v)(dst, src, stride, rounding); + break; + + case 9: + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_lowpass_h)(halfpel_h, src, stride, rounding); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_avg2)(halfpel_v, src, halfpel_h, stride, rounding, 9); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_lowpass_v)(dst, halfpel_v, stride, rounding); + break; + + case 10: + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_lowpass_hv)(dst, halfpel_h, src, stride, rounding); + break; + + case 11: + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_lowpass_h)(halfpel_h, src, stride, rounding); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_avg2)(halfpel_v, src + 1, halfpel_h, stride, rounding, 9); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_lowpass_v)(dst, halfpel_v, stride, rounding); + break; + + case 12: + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_lowpass_v)(halfpel_v, src, stride, rounding); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_avg2)(dst, src+stride, halfpel_v, stride, rounding, 8); + break; + + case 13: + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_lowpass_h)(halfpel_h, src, stride, rounding); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_avg2)(halfpel_v, src, halfpel_h, stride, rounding, 9); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_lowpass_v)(halfpel_hv, halfpel_v, stride, rounding); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_avg2)(dst, halfpel_v+stride, halfpel_hv, stride, rounding, 8); + break; + + case 14: + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_lowpass_hv)(halfpel_hv, halfpel_h, src, stride, rounding); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_avg2)(dst, halfpel_h+stride, halfpel_hv, stride, rounding, 8); + break; + + case 15: + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_lowpass_h)(halfpel_h, src, stride, rounding); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_avg2)(halfpel_v, src + 1, halfpel_h, stride, rounding, 9); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_lowpass_v)(halfpel_hv, halfpel_v, stride, rounding); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_avg2)(dst, halfpel_hv, halfpel_v + stride, stride, rounding, 8); + break; + } +} + void interpolate16x16_quarterpel(uint8_t * const cur, uint8_t * const refn, + uint8_t * const refh, + uint8_t * const refv, + uint8_t * const refhv, + const uint32_t x, const uint32_t y, + const int32_t dx, const int dy, + const uint32_t stride, + const uint32_t rounding) + { + + const int32_t xRef = (int)x*4 + dx; + const int32_t yRef = (int)y*4 + dy; + + uint8_t *src, *dst; + uint8_t *halfpel_h, *halfpel_v, *halfpel_hv; + int32_t x_int, y_int, x_frac, y_frac; + + x_int = xRef/4; + if (xRef < 0 && xRef % 4) + x_int--; + + x_frac = xRef - (4*x_int); + + y_int = yRef/4; + if (yRef < 0 && yRef % 4) + y_int--; + + y_frac = yRef - (4*y_int); + + src = refn + y_int * (int)stride + x_int; + halfpel_h = refh; + halfpel_v = refv; + halfpel_hv = refhv; + + dst = cur + y * stride + x; + + switch((y_frac << 2) | (x_frac)) { + + case 0: + transfer16x16_copy(dst, src, stride); + break; + + case 1: + (((struct global_all_dll *)Dll::Tls())->interpolate16x16_lowpass_h)(halfpel_h, src, stride, rounding); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_avg2)(dst, src, halfpel_h, stride, rounding, 8); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_avg2)(dst+8, src+8, halfpel_h+8, stride, rounding, 8); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_avg2)(dst+8*stride, src+8*stride, halfpel_h+8*stride, stride, rounding, 8); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_avg2)(dst+8*stride+8, src+8*stride+8, halfpel_h+8*stride+8, stride, rounding, 8); + break; + + case 2: + (((struct global_all_dll *)Dll::Tls())->interpolate16x16_lowpass_h)(dst, src, stride, rounding); + break; + + case 3: + (((struct global_all_dll *)Dll::Tls())->interpolate16x16_lowpass_h)(halfpel_h, src, stride, rounding); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_avg2)(dst, src + 1, halfpel_h, stride, rounding, 8); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_avg2)(dst+8, src + 8 + 1, halfpel_h+8, stride, rounding, 8); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_avg2)(dst+8*stride, src + 8*stride + 1, halfpel_h+8*stride, stride, rounding, 8); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_avg2)(dst+8*stride+8, src+8*stride+8 + 1, halfpel_h+8*stride+8, stride, rounding, 8); + break; + + case 4: + (((struct global_all_dll *)Dll::Tls())->interpolate16x16_lowpass_v)(halfpel_v, src, stride, rounding); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_avg2)(dst, src, halfpel_v, stride, rounding, 8); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_avg2)(dst+8, src+8, halfpel_v+8, stride, rounding, 8); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_avg2)(dst+8*stride, src+8*stride, halfpel_v+8*stride, stride, rounding, 8); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_avg2)(dst+8*stride+8, src+8*stride+8, halfpel_v+8*stride+8, stride, rounding, 8); + break; + + case 5: + (((struct global_all_dll *)Dll::Tls())->interpolate16x16_lowpass_h)(halfpel_h, src, stride, rounding); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_avg2)(halfpel_v, src, halfpel_h, stride, rounding, 9); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_avg2)(halfpel_v+8, src + 8, halfpel_h+8, stride, rounding, 9); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_avg2)(halfpel_v+8*stride, src + 8*stride, halfpel_h+8*stride, stride, rounding, 9); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_avg2)(halfpel_v+8*stride+8, src+8*stride+8, halfpel_h+8*stride+8, stride, rounding, 9); + + (((struct global_all_dll *)Dll::Tls())->interpolate16x16_lowpass_v)(halfpel_hv, halfpel_v, stride, rounding); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_avg2)(dst, halfpel_hv, halfpel_v, stride, rounding, 8); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_avg2)(dst+8, halfpel_hv+8, halfpel_v+8, stride, rounding, 8); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_avg2)(dst+8*stride, halfpel_hv+8*stride, halfpel_v+8*stride, stride, rounding, 8); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_avg2)(dst+8*stride+8, halfpel_hv+8*stride+8, halfpel_v+8*stride+8, stride, rounding, 8); + break; + + case 6: + (((struct global_all_dll *)Dll::Tls())->interpolate16x16_lowpass_hv)(halfpel_hv, halfpel_h, src, stride, rounding); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_avg2)(dst, halfpel_h, halfpel_hv, stride, rounding, 8); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_avg2)(dst+8, halfpel_h+8, halfpel_hv+8, stride, rounding, 8); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_avg2)(dst+8*stride, halfpel_h+8*stride, halfpel_hv+8*stride, stride, rounding, 8); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_avg2)(dst+8*stride+8, halfpel_h+8*stride+8, halfpel_hv+8*stride+8, stride, rounding, 8); + break; + + case 7: + (((struct global_all_dll *)Dll::Tls())->interpolate16x16_lowpass_h)(halfpel_h, src, stride, rounding); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_avg2)(halfpel_v, src+1, halfpel_h, stride, rounding, 9); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_avg2)(halfpel_v+8, src+1 + 8, halfpel_h+8, stride, rounding, 9); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_avg2)(halfpel_v+8*stride, src+1 + 8*stride, halfpel_h+8*stride, stride, rounding, 9); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_avg2)(halfpel_v+8*stride+8, src+1+8*stride+8, halfpel_h+8*stride+8, stride, rounding, 9); + + (((struct global_all_dll *)Dll::Tls())->interpolate16x16_lowpass_v)(halfpel_hv, halfpel_v, stride, rounding); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_avg2)(dst, halfpel_hv, halfpel_v, stride, rounding, 8); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_avg2)(dst+8, halfpel_hv+8, halfpel_v+8, stride, rounding, 8); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_avg2)(dst+8*stride, halfpel_hv+8*stride, halfpel_v+8*stride, stride, rounding, 8); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_avg2)(dst+8*stride+8, halfpel_hv+8*stride+8, halfpel_v+8*stride+8, stride, rounding, 8); + break; + + case 8: + (((struct global_all_dll *)Dll::Tls())->interpolate16x16_lowpass_v)(dst, src, stride, rounding); + break; + + case 9: + (((struct global_all_dll *)Dll::Tls())->interpolate16x16_lowpass_h)(halfpel_h, src, stride, rounding); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_avg2)(halfpel_v, src, halfpel_h, stride, rounding, 9); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_avg2)(halfpel_v+8, src + 8, halfpel_h+8, stride, rounding, 9); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_avg2)(halfpel_v+8*stride, src + 8*stride, halfpel_h+8*stride, stride, rounding, 9); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_avg2)(halfpel_v+8*stride+8, src+8*stride+8, halfpel_h+8*stride+8, stride, rounding, 9); + (((struct global_all_dll *)Dll::Tls())->interpolate16x16_lowpass_v)(dst, halfpel_v, stride, rounding); + break; + + case 10: + (((struct global_all_dll *)Dll::Tls())->interpolate16x16_lowpass_hv)(dst, halfpel_h, src, stride, rounding); + break; + + case 11: + (((struct global_all_dll *)Dll::Tls())->interpolate16x16_lowpass_h)(halfpel_h, src, stride, rounding); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_avg2)(halfpel_v, src+1, halfpel_h, stride, rounding, 9); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_avg2)(halfpel_v+8, src+1 + 8, halfpel_h+8, stride, rounding, 9); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_avg2)(halfpel_v+8*stride, src+1 + 8*stride, halfpel_h+8*stride, stride, rounding, 9); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_avg2)(halfpel_v+8*stride+8, src+1+8*stride+8, halfpel_h+8*stride+8, stride, rounding, 9); + (((struct global_all_dll *)Dll::Tls())->interpolate16x16_lowpass_v)(dst, halfpel_v, stride, rounding); + break; + + case 12: + (((struct global_all_dll *)Dll::Tls())->interpolate16x16_lowpass_v)(halfpel_v, src, stride, rounding); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_avg2)(dst, src+stride, halfpel_v, stride, rounding, 8); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_avg2)(dst+8, src+stride+8, halfpel_v+8, stride, rounding, 8); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_avg2)(dst+8*stride, src+stride+8*stride, halfpel_v+8*stride, stride, rounding, 8); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_avg2)(dst+8*stride+8, src+stride+8*stride+8, halfpel_v+8*stride+8, stride, rounding, 8); + break; + + case 13: + (((struct global_all_dll *)Dll::Tls())->interpolate16x16_lowpass_h)(halfpel_h, src, stride, rounding); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_avg2)(halfpel_v, src, halfpel_h, stride, rounding, 9); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_avg2)(halfpel_v+8, src + 8, halfpel_h+8, stride, rounding, 9); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_avg2)(halfpel_v+8*stride, src + 8*stride, halfpel_h+8*stride, stride, rounding, 9); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_avg2)(halfpel_v+8*stride+8, src+8*stride+8, halfpel_h+8*stride+8, stride, rounding, 9); + + (((struct global_all_dll *)Dll::Tls())->interpolate16x16_lowpass_v)(halfpel_hv, halfpel_v, stride, rounding); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_avg2)(dst, halfpel_hv, halfpel_v+stride, stride, rounding, 8); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_avg2)(dst+8, halfpel_hv+8, halfpel_v+stride+8, stride, rounding, 8); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_avg2)(dst+8*stride, halfpel_hv+8*stride, halfpel_v+stride+8*stride, stride, rounding, 8); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_avg2)(dst+8*stride+8, halfpel_hv+8*stride+8, halfpel_v+stride+8*stride+8, stride, rounding, 8); + break; + + case 14: + (((struct global_all_dll *)Dll::Tls())->interpolate16x16_lowpass_hv)(halfpel_hv, halfpel_h, src, stride, rounding); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_avg2)(dst, halfpel_h+stride, halfpel_hv, stride, rounding, 8); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_avg2)(dst+8, halfpel_h+stride+8, halfpel_hv+8, stride, rounding, 8); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_avg2)(dst+8*stride, halfpel_h+stride+8*stride, halfpel_hv+8*stride, stride, rounding, 8); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_avg2)(dst+8*stride+8, halfpel_h+stride+8*stride+8, halfpel_hv+8*stride+8, stride, rounding, 8); + break; + + case 15: + (((struct global_all_dll *)Dll::Tls())->interpolate16x16_lowpass_h)(halfpel_h, src, stride, rounding); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_avg2)(halfpel_v, src+1, halfpel_h, stride, rounding, 9); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_avg2)(halfpel_v+8, src+1 + 8, halfpel_h+8, stride, rounding, 9); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_avg2)(halfpel_v+8*stride, src+1 + 8*stride, halfpel_h+8*stride, stride, rounding, 9); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_avg2)(halfpel_v+8*stride+8, src+1+8*stride+8, halfpel_h+8*stride+8, stride, rounding, 9); + + (((struct global_all_dll *)Dll::Tls())->interpolate16x16_lowpass_v)(halfpel_hv, halfpel_v, stride, rounding); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_avg2)(dst, halfpel_hv, halfpel_v+stride, stride, rounding, 8); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_avg2)(dst+8, halfpel_hv+8, halfpel_v+stride+8, stride, rounding, 8); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_avg2)(dst+8*stride, halfpel_hv+8*stride, halfpel_v+stride+8*stride, stride, rounding, 8); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_avg2)(dst+8*stride+8, halfpel_hv+8*stride+8, halfpel_v+stride+8*stride+8, stride, rounding, 8); + break; + } +} + -/* function pointers */ -INTERPOLATE8X8_PTR interpolate8x8_halfpel_h; -INTERPOLATE8X8_PTR interpolate8x8_halfpel_v; -INTERPOLATE8X8_PTR interpolate8x8_halfpel_hv; -INTERPOLATE8X8_AVG2_PTR interpolate8x8_avg2; -INTERPOLATE8X8_AVG4_PTR interpolate8x8_avg4; -INTERPOLATE_LOWPASS_PTR interpolate8x8_lowpass_h; -INTERPOLATE_LOWPASS_PTR interpolate8x8_lowpass_v; -INTERPOLATE_LOWPASS_PTR interpolate16x16_lowpass_h; -INTERPOLATE_LOWPASS_PTR interpolate16x16_lowpass_v; -INTERPOLATE_LOWPASS_HV_PTR interpolate8x8_lowpass_hv; -INTERPOLATE_LOWPASS_HV_PTR interpolate16x16_lowpass_hv; -INTERPOLATE8X8_6TAP_LOWPASS_PTR interpolate8x8_6tap_lowpass_h; -INTERPOLATE8X8_6TAP_LOWPASS_PTR interpolate8x8_6tap_lowpass_v; void interpolate8x8_avg2_c(uint8_t *dst, const uint8_t *src1, const uint8_t *src2, const uint32_t stride, const uint32_t rounding, const uint32_t height) { diff -bBru src-ori/image/interpolate8x8.h src/image/interpolate8x8.h --- src-ori/image/interpolate8x8.h Wed Jan 12 10:32:43 2005 +++ src/image/interpolate8x8.h Wed Jan 12 12:37:52 2005 @@ -68,29 +89,32 @@ typedef void (INTERPOLATE8X8_6TAP_LOWPASS) (uint8_t *dst, uint8_t *src, + int32_t stride, + int32_t rounding); + typedef INTERPOLATE8X8_6TAP_LOWPASS *INTERPOLATE8X8_6TAP_LOWPASS_PTR; -extern INTERPOLATE8X8_PTR interpolate8x8_halfpel_h; -extern INTERPOLATE8X8_PTR interpolate8x8_halfpel_v; -extern INTERPOLATE8X8_PTR interpolate8x8_halfpel_hv; -extern INTERPOLATE8X8_AVG2_PTR interpolate8x8_avg2; -extern INTERPOLATE8X8_AVG4_PTR interpolate8x8_avg4; -extern INTERPOLATE_LOWPASS_PTR interpolate8x8_lowpass_h; -extern INTERPOLATE_LOWPASS_PTR interpolate8x8_lowpass_v; -extern INTERPOLATE_LOWPASS_PTR interpolate16x16_lowpass_h; -extern INTERPOLATE_LOWPASS_PTR interpolate16x16_lowpass_v; -extern INTERPOLATE_LOWPASS_HV_PTR interpolate8x8_lowpass_hv; -extern INTERPOLATE_LOWPASS_HV_PTR interpolate16x16_lowpass_hv; -extern INTERPOLATE8X8_6TAP_LOWPASS_PTR interpolate8x8_6tap_lowpass_h; -extern INTERPOLATE8X8_6TAP_LOWPASS_PTR interpolate8x8_6tap_lowpass_v; + + + + + + + + + + + + + INTERPOLATE8X8 interpolate8x8_halfpel_h_c; INTERPOLATE8X8 interpolate8x8_halfpel_v_c; @@ -145,8 +169,7 @@ INTERPOLATE8X8_6TAP_LOWPASS interpolate8x8_6tap_lowpass_v_mmx; #endif -static __inline void -interpolate8x8_switch(uint8_t * const cur, +extern void interpolate8x8_switch(uint8_t * const cur, const uint8_t * const refn, const uint32_t x, const uint32_t y, @@ -154,30 +177,10 @@ const int dy, const uint32_t stride, const uint32_t rounding) -{ - - const uint8_t * const src = refn + (int)((y + (dy>>1)) * stride + x + (dx>>1)); - uint8_t * const dst = cur + (int)(y * stride + x); - - switch (((dx & 1) << 1) + (dy & 1)) { /* ((dx%2)?2:0)+((dy%2)?1:0) */ - case 0: - transfer8x8_copy(dst, src, stride); - break; - case 1: - interpolate8x8_halfpel_v(dst, src, stride, rounding); - break; - case 2: - interpolate8x8_halfpel_h(dst, src, stride, rounding); - break; - default: - interpolate8x8_halfpel_hv(dst, src, stride, rounding); - break; - } -} +; -static __inline void -interpolate16x16_switch(uint8_t * const cur, +extern void interpolate16x16_switch(uint8_t * const cur, const uint8_t * const refn, const uint32_t x, const uint32_t y, @@ -185,16 +188,10 @@ const int dy, const uint32_t stride, const uint32_t rounding) -{ - interpolate8x8_switch(cur, refn, x, y, dx, dy, stride, rounding); - interpolate8x8_switch(cur, refn, x+8, y, dx, dy, stride, rounding); - interpolate8x8_switch(cur, refn, x, y+8, dx, dy, stride, rounding); - interpolate8x8_switch(cur, refn, x+8, y+8, dx, dy, stride, rounding); -} +; -static __inline void -interpolate32x32_switch(uint8_t * const cur, +extern void interpolate32x32_switch(uint8_t * const cur, const uint8_t * const refn, const uint32_t x, const uint32_t y, @@ -202,16 +199,10 @@ const int dy, const uint32_t stride, const uint32_t rounding) -{ - interpolate16x16_switch(cur, refn, x, y, dx, dy, stride, rounding); - interpolate16x16_switch(cur, refn, x+16, y, dx, dy, stride, rounding); - interpolate16x16_switch(cur, refn, x, y+16, dx, dy, stride, rounding); - interpolate16x16_switch(cur, refn, x+16, y+16, dx, dy, stride, rounding); -} +; -static __inline uint8_t * -interpolate8x8_switch2(uint8_t * const buffer, +extern uint8_t * interpolate8x8_switch2(uint8_t * const buffer, const uint8_t * const refn, const int x, const int y, @@ -219,28 +210,9 @@ const int dy, const uint32_t stride, const uint32_t rounding) -{ - - const uint8_t * const src = refn + (int)((y + (dy>>1)) * stride + x + (dx>>1)); - - switch (((dx & 1) << 1) + (dy & 1)) { /* ((dx%2)?2:0)+((dy%2)?1:0) */ - case 0: - return (uint8_t *)src; - case 1: - interpolate8x8_halfpel_v(buffer, src, stride, rounding); - break; - case 2: - interpolate8x8_halfpel_h(buffer, src, stride, rounding); - break; - default: - interpolate8x8_halfpel_hv(buffer, src, stride, rounding); - break; - } - return buffer; -} +; -static __inline void interpolate8x8_quarterpel(uint8_t * const cur, - uint8_t * const refn, +extern void interpolate8x8_quarterpel(uint8_t * const cur, uint8_t * const refn, uint8_t * const refh, uint8_t * const refv, uint8_t * const refhv, @@ -248,125 +220,9 @@ const int32_t dx, const int dy, const uint32_t stride, const uint32_t rounding) -{ - const int32_t xRef = (int)x*4 + dx; - const int32_t yRef = (int)y*4 + dy; - - uint8_t *src, *dst; - uint8_t *halfpel_h, *halfpel_v, *halfpel_hv; - int32_t x_int, y_int, x_frac, y_frac; - - x_int = xRef/4; - if (xRef < 0 && xRef % 4) - x_int--; - - x_frac = xRef - (4*x_int); - - y_int = yRef/4; - if (yRef < 0 && yRef % 4) - y_int--; - - y_frac = yRef - (4*y_int); - - src = refn + y_int * (int)stride + x_int; - halfpel_h = refh; - halfpel_v = refv; - halfpel_hv = refhv; - - dst = cur + y * stride + x; - - switch((y_frac << 2) | (x_frac)) { - - case 0: - transfer8x8_copy(dst, src, stride); - break; - - case 1: - interpolate8x8_lowpass_h(halfpel_h, src, stride, rounding); - interpolate8x8_avg2(dst, src, halfpel_h, stride, rounding, 8); - break; - - case 2: - interpolate8x8_lowpass_h(dst, src, stride, rounding); - break; - - case 3: - interpolate8x8_lowpass_h(halfpel_h, src, stride, rounding); - interpolate8x8_avg2(dst, src + 1, halfpel_h, stride, rounding, 8); - break; - - case 4: - interpolate8x8_lowpass_v(halfpel_v, src, stride, rounding); - interpolate8x8_avg2(dst, src, halfpel_v, stride, rounding, 8); - break; - - case 5: - interpolate8x8_lowpass_h(halfpel_h, src, stride, rounding); - interpolate8x8_avg2(halfpel_v, src, halfpel_h, stride, rounding, 9); - interpolate8x8_lowpass_v(halfpel_hv, halfpel_v, stride, rounding); - interpolate8x8_avg2(dst, halfpel_v, halfpel_hv, stride, rounding, 8); - break; - - case 6: - interpolate8x8_lowpass_hv(halfpel_hv, halfpel_h, src, stride, rounding); - interpolate8x8_avg2(dst, halfpel_h, halfpel_hv, stride, rounding, 8); - break; - - case 7: - interpolate8x8_lowpass_h(halfpel_h, src, stride, rounding); - interpolate8x8_avg2(halfpel_v, src + 1, halfpel_h, stride, rounding, 9); - interpolate8x8_lowpass_v(halfpel_hv, halfpel_v, stride, rounding); - interpolate8x8_avg2(dst, halfpel_v, halfpel_hv, stride, rounding, 8); - break; - - case 8: - interpolate8x8_lowpass_v(dst, src, stride, rounding); - break; - - case 9: - interpolate8x8_lowpass_h(halfpel_h, src, stride, rounding); - interpolate8x8_avg2(halfpel_v, src, halfpel_h, stride, rounding, 9); - interpolate8x8_lowpass_v(dst, halfpel_v, stride, rounding); - break; - - case 10: - interpolate8x8_lowpass_hv(dst, halfpel_h, src, stride, rounding); - break; - - case 11: - interpolate8x8_lowpass_h(halfpel_h, src, stride, rounding); - interpolate8x8_avg2(halfpel_v, src + 1, halfpel_h, stride, rounding, 9); - interpolate8x8_lowpass_v(dst, halfpel_v, stride, rounding); - break; - - case 12: - interpolate8x8_lowpass_v(halfpel_v, src, stride, rounding); - interpolate8x8_avg2(dst, src+stride, halfpel_v, stride, rounding, 8); - break; - - case 13: - interpolate8x8_lowpass_h(halfpel_h, src, stride, rounding); - interpolate8x8_avg2(halfpel_v, src, halfpel_h, stride, rounding, 9); - interpolate8x8_lowpass_v(halfpel_hv, halfpel_v, stride, rounding); - interpolate8x8_avg2(dst, halfpel_v+stride, halfpel_hv, stride, rounding, 8); - break; - - case 14: - interpolate8x8_lowpass_hv(halfpel_hv, halfpel_h, src, stride, rounding); - interpolate8x8_avg2(dst, halfpel_h+stride, halfpel_hv, stride, rounding, 8); - break; - - case 15: - interpolate8x8_lowpass_h(halfpel_h, src, stride, rounding); - interpolate8x8_avg2(halfpel_v, src + 1, halfpel_h, stride, rounding, 9); - interpolate8x8_lowpass_v(halfpel_hv, halfpel_v, stride, rounding); - interpolate8x8_avg2(dst, halfpel_hv, halfpel_v + stride, stride, rounding, 8); - break; - } -} +; -static __inline void interpolate16x16_quarterpel(uint8_t * const cur, - uint8_t * const refn, +extern void interpolate16x16_quarterpel(uint8_t * const cur, uint8_t * const refn, uint8_t * const refh, uint8_t * const refv, uint8_t * const refhv, @@ -374,173 +230,6 @@ const int32_t dx, const int dy, const uint32_t stride, const uint32_t rounding) -{ - const int32_t xRef = (int)x*4 + dx; - const int32_t yRef = (int)y*4 + dy; - - uint8_t *src, *dst; - uint8_t *halfpel_h, *halfpel_v, *halfpel_hv; - int32_t x_int, y_int, x_frac, y_frac; - - x_int = xRef/4; - if (xRef < 0 && xRef % 4) - x_int--; - - x_frac = xRef - (4*x_int); - - y_int = yRef/4; - if (yRef < 0 && yRef % 4) - y_int--; - - y_frac = yRef - (4*y_int); - - src = refn + y_int * (int)stride + x_int; - halfpel_h = refh; - halfpel_v = refv; - halfpel_hv = refhv; - - dst = cur + y * stride + x; - - switch((y_frac << 2) | (x_frac)) { - - case 0: - transfer16x16_copy(dst, src, stride); - break; - - case 1: - interpolate16x16_lowpass_h(halfpel_h, src, stride, rounding); - interpolate8x8_avg2(dst, src, halfpel_h, stride, rounding, 8); - interpolate8x8_avg2(dst+8, src+8, halfpel_h+8, stride, rounding, 8); - interpolate8x8_avg2(dst+8*stride, src+8*stride, halfpel_h+8*stride, stride, rounding, 8); - interpolate8x8_avg2(dst+8*stride+8, src+8*stride+8, halfpel_h+8*stride+8, stride, rounding, 8); - break; - - case 2: - interpolate16x16_lowpass_h(dst, src, stride, rounding); - break; - - case 3: - interpolate16x16_lowpass_h(halfpel_h, src, stride, rounding); - interpolate8x8_avg2(dst, src + 1, halfpel_h, stride, rounding, 8); - interpolate8x8_avg2(dst+8, src + 8 + 1, halfpel_h+8, stride, rounding, 8); - interpolate8x8_avg2(dst+8*stride, src + 8*stride + 1, halfpel_h+8*stride, stride, rounding, 8); - interpolate8x8_avg2(dst+8*stride+8, src+8*stride+8 + 1, halfpel_h+8*stride+8, stride, rounding, 8); - break; - - case 4: - interpolate16x16_lowpass_v(halfpel_v, src, stride, rounding); - interpolate8x8_avg2(dst, src, halfpel_v, stride, rounding, 8); - interpolate8x8_avg2(dst+8, src+8, halfpel_v+8, stride, rounding, 8); - interpolate8x8_avg2(dst+8*stride, src+8*stride, halfpel_v+8*stride, stride, rounding, 8); - interpolate8x8_avg2(dst+8*stride+8, src+8*stride+8, halfpel_v+8*stride+8, stride, rounding, 8); - break; - - case 5: - interpolate16x16_lowpass_h(halfpel_h, src, stride, rounding); - interpolate8x8_avg2(halfpel_v, src, halfpel_h, stride, rounding, 9); - interpolate8x8_avg2(halfpel_v+8, src + 8, halfpel_h+8, stride, rounding, 9); - interpolate8x8_avg2(halfpel_v+8*stride, src + 8*stride, halfpel_h+8*stride, stride, rounding, 9); - interpolate8x8_avg2(halfpel_v+8*stride+8, src+8*stride+8, halfpel_h+8*stride+8, stride, rounding, 9); - - interpolate16x16_lowpass_v(halfpel_hv, halfpel_v, stride, rounding); - interpolate8x8_avg2(dst, halfpel_hv, halfpel_v, stride, rounding, 8); - interpolate8x8_avg2(dst+8, halfpel_hv+8, halfpel_v+8, stride, rounding, 8); - interpolate8x8_avg2(dst+8*stride, halfpel_hv+8*stride, halfpel_v+8*stride, stride, rounding, 8); - interpolate8x8_avg2(dst+8*stride+8, halfpel_hv+8*stride+8, halfpel_v+8*stride+8, stride, rounding, 8); - break; - - case 6: - interpolate16x16_lowpass_hv(halfpel_hv, halfpel_h, src, stride, rounding); - interpolate8x8_avg2(dst, halfpel_h, halfpel_hv, stride, rounding, 8); - interpolate8x8_avg2(dst+8, halfpel_h+8, halfpel_hv+8, stride, rounding, 8); - interpolate8x8_avg2(dst+8*stride, halfpel_h+8*stride, halfpel_hv+8*stride, stride, rounding, 8); - interpolate8x8_avg2(dst+8*stride+8, halfpel_h+8*stride+8, halfpel_hv+8*stride+8, stride, rounding, 8); - break; - - case 7: - interpolate16x16_lowpass_h(halfpel_h, src, stride, rounding); - interpolate8x8_avg2(halfpel_v, src+1, halfpel_h, stride, rounding, 9); - interpolate8x8_avg2(halfpel_v+8, src+1 + 8, halfpel_h+8, stride, rounding, 9); - interpolate8x8_avg2(halfpel_v+8*stride, src+1 + 8*stride, halfpel_h+8*stride, stride, rounding, 9); - interpolate8x8_avg2(halfpel_v+8*stride+8, src+1+8*stride+8, halfpel_h+8*stride+8, stride, rounding, 9); - - interpolate16x16_lowpass_v(halfpel_hv, halfpel_v, stride, rounding); - interpolate8x8_avg2(dst, halfpel_hv, halfpel_v, stride, rounding, 8); - interpolate8x8_avg2(dst+8, halfpel_hv+8, halfpel_v+8, stride, rounding, 8); - interpolate8x8_avg2(dst+8*stride, halfpel_hv+8*stride, halfpel_v+8*stride, stride, rounding, 8); - interpolate8x8_avg2(dst+8*stride+8, halfpel_hv+8*stride+8, halfpel_v+8*stride+8, stride, rounding, 8); - break; - - case 8: - interpolate16x16_lowpass_v(dst, src, stride, rounding); - break; - - case 9: - interpolate16x16_lowpass_h(halfpel_h, src, stride, rounding); - interpolate8x8_avg2(halfpel_v, src, halfpel_h, stride, rounding, 9); - interpolate8x8_avg2(halfpel_v+8, src + 8, halfpel_h+8, stride, rounding, 9); - interpolate8x8_avg2(halfpel_v+8*stride, src + 8*stride, halfpel_h+8*stride, stride, rounding, 9); - interpolate8x8_avg2(halfpel_v+8*stride+8, src+8*stride+8, halfpel_h+8*stride+8, stride, rounding, 9); - interpolate16x16_lowpass_v(dst, halfpel_v, stride, rounding); - break; - - case 10: - interpolate16x16_lowpass_hv(dst, halfpel_h, src, stride, rounding); - break; - - case 11: - interpolate16x16_lowpass_h(halfpel_h, src, stride, rounding); - interpolate8x8_avg2(halfpel_v, src+1, halfpel_h, stride, rounding, 9); - interpolate8x8_avg2(halfpel_v+8, src+1 + 8, halfpel_h+8, stride, rounding, 9); - interpolate8x8_avg2(halfpel_v+8*stride, src+1 + 8*stride, halfpel_h+8*stride, stride, rounding, 9); - interpolate8x8_avg2(halfpel_v+8*stride+8, src+1+8*stride+8, halfpel_h+8*stride+8, stride, rounding, 9); - interpolate16x16_lowpass_v(dst, halfpel_v, stride, rounding); - break; - - case 12: - interpolate16x16_lowpass_v(halfpel_v, src, stride, rounding); - interpolate8x8_avg2(dst, src+stride, halfpel_v, stride, rounding, 8); - interpolate8x8_avg2(dst+8, src+stride+8, halfpel_v+8, stride, rounding, 8); - interpolate8x8_avg2(dst+8*stride, src+stride+8*stride, halfpel_v+8*stride, stride, rounding, 8); - interpolate8x8_avg2(dst+8*stride+8, src+stride+8*stride+8, halfpel_v+8*stride+8, stride, rounding, 8); - break; - - case 13: - interpolate16x16_lowpass_h(halfpel_h, src, stride, rounding); - interpolate8x8_avg2(halfpel_v, src, halfpel_h, stride, rounding, 9); - interpolate8x8_avg2(halfpel_v+8, src + 8, halfpel_h+8, stride, rounding, 9); - interpolate8x8_avg2(halfpel_v+8*stride, src + 8*stride, halfpel_h+8*stride, stride, rounding, 9); - interpolate8x8_avg2(halfpel_v+8*stride+8, src+8*stride+8, halfpel_h+8*stride+8, stride, rounding, 9); - - interpolate16x16_lowpass_v(halfpel_hv, halfpel_v, stride, rounding); - interpolate8x8_avg2(dst, halfpel_hv, halfpel_v+stride, stride, rounding, 8); - interpolate8x8_avg2(dst+8, halfpel_hv+8, halfpel_v+stride+8, stride, rounding, 8); - interpolate8x8_avg2(dst+8*stride, halfpel_hv+8*stride, halfpel_v+stride+8*stride, stride, rounding, 8); - interpolate8x8_avg2(dst+8*stride+8, halfpel_hv+8*stride+8, halfpel_v+stride+8*stride+8, stride, rounding, 8); - break; - - case 14: - interpolate16x16_lowpass_hv(halfpel_hv, halfpel_h, src, stride, rounding); - interpolate8x8_avg2(dst, halfpel_h+stride, halfpel_hv, stride, rounding, 8); - interpolate8x8_avg2(dst+8, halfpel_h+stride+8, halfpel_hv+8, stride, rounding, 8); - interpolate8x8_avg2(dst+8*stride, halfpel_h+stride+8*stride, halfpel_hv+8*stride, stride, rounding, 8); - interpolate8x8_avg2(dst+8*stride+8, halfpel_h+stride+8*stride+8, halfpel_hv+8*stride+8, stride, rounding, 8); - break; - - case 15: - interpolate16x16_lowpass_h(halfpel_h, src, stride, rounding); - interpolate8x8_avg2(halfpel_v, src+1, halfpel_h, stride, rounding, 9); - interpolate8x8_avg2(halfpel_v+8, src+1 + 8, halfpel_h+8, stride, rounding, 9); - interpolate8x8_avg2(halfpel_v+8*stride, src+1 + 8*stride, halfpel_h+8*stride, stride, rounding, 9); - interpolate8x8_avg2(halfpel_v+8*stride+8, src+1+8*stride+8, halfpel_h+8*stride+8, stride, rounding, 9); - - interpolate16x16_lowpass_v(halfpel_hv, halfpel_v, stride, rounding); - interpolate8x8_avg2(dst, halfpel_hv, halfpel_v+stride, stride, rounding, 8); - interpolate8x8_avg2(dst+8, halfpel_hv+8, halfpel_v+stride+8, stride, rounding, 8); - interpolate8x8_avg2(dst+8*stride, halfpel_hv+8*stride, halfpel_v+stride+8*stride, stride, rounding, 8); - interpolate8x8_avg2(dst+8*stride+8, halfpel_hv+8*stride+8, halfpel_v+stride+8*stride+8, stride, rounding, 8); - break; - } -} +; #endif diff -bBru src-ori/image/postprocessing.c src/image/postprocessing.c --- src-ori/image/postprocessing.c Wed Jan 12 10:32:43 2005 +++ src/image/postprocessing.c Wed Jan 12 12:40:28 2005 @@ -23,15 +23,9 @@ * ****************************************************************************/ -#include -#include -#include - -#include "../portab.h" -#include "../global.h" -#include "image.h" -#include "../utils/emms.h" -#include "postprocessing.h" +#include +#include "../global_all_dll.h" + /* Some useful (and fast) macros Note that the MIN/MAX macros assume signed shift - if your compiler @@ -285,7 +282,7 @@ int i, j; int patt[4] = { -1,0,1,0 }; - emms(); + (((struct global_all_dll *)Dll::Tls())->emms)(); srand(123457); diff -bBru src-ori/image/qpel.c src/image/qpel.c --- src-ori/image/qpel.c Wed Jan 12 11:25:40 2005 +++ src/image/qpel.c Wed Jan 12 12:40:42 2005 @@ -25,8 +25,307 @@ #ifndef XVID_AUTO_INCLUDE -#include "../portab.h" -#include "qpel.h" +#include +#include "../global_all_dll.h" + new_interpolate16x16_quarterpel(uint8_t * const cur, + uint8_t * const refn, + uint8_t * const refh, + uint8_t * const refv, + uint8_t * const refhv, + const uint32_t x, const uint32_t y, + const int32_t dx, const int dy, + const uint32_t stride, + const uint32_t rounding) + { + + const uint8_t *src; + uint8_t *dst; + uint8_t *tmp; + int32_t quads; + const XVID_QP_FUNCS *Ops; + + int32_t x_int, y_int; + + const int32_t xRef = (int)x*4 + dx; + const int32_t yRef = (int)y*4 + dy; + + Ops = (((struct global_all_dll *)Dll::Tls())->xvid_QP_Funcs); /* TODO: pass as argument */ + quads = (dx&3) | ((dy&3)<<2); + + x_int = xRef/4; + if (xRef < 0 && xRef % 4) + x_int--; + + y_int = yRef/4; + if (yRef < 0 && yRef % 4) + y_int--; + + dst = cur + y * stride + x; + src = refn + y_int * (int)stride + x_int; + + tmp = refh; /* we need at least a 16 x stride scratch block */ + + switch(quads) { + case 0: + (((struct global_all_dll *)Dll::Tls())->transfer8x8_copy)( dst, src, stride); + (((struct global_all_dll *)Dll::Tls())->transfer8x8_copy)( dst+8, src+8, stride); + (((struct global_all_dll *)Dll::Tls())->transfer8x8_copy)( dst+8*stride, src+8*stride, stride); + (((struct global_all_dll *)Dll::Tls())->transfer8x8_copy)( dst+8*stride+8, src+8*stride+8, stride); + break; + case 1: + Ops->H_Pass_Avrg(dst, src, 16, stride, rounding); + break; + case 2: + Ops->H_Pass(dst, src, 16, stride, rounding); + break; + case 3: + Ops->H_Pass_Avrg_Up(dst, src, 16, stride, rounding); + break; + case 4: + Ops->V_Pass_Avrg(dst, src, 16, stride, rounding); + break; + case 5: + Ops->H_Pass_Avrg(tmp, src, 17, stride, rounding); + Ops->V_Pass_Avrg(dst, tmp, 16, stride, rounding); + break; + case 6: + Ops->H_Pass(tmp, src, 17, stride, rounding); + Ops->V_Pass_Avrg(dst, tmp, 16, stride, rounding); + break; + case 7: + Ops->H_Pass_Avrg_Up(tmp, src, 17, stride, rounding); + Ops->V_Pass_Avrg(dst, tmp, 16, stride, rounding); + break; + case 8: + Ops->V_Pass(dst, src, 16, stride, rounding); + break; + case 9: + Ops->H_Pass_Avrg(tmp, src, 17, stride, rounding); + Ops->V_Pass(dst, tmp, 16, stride, rounding); + break; + case 10: + Ops->H_Pass(tmp, src, 17, stride, rounding); + Ops->V_Pass(dst, tmp, 16, stride, rounding); + break; + case 11: + Ops->H_Pass_Avrg_Up(tmp, src, 17, stride, rounding); + Ops->V_Pass(dst, tmp, 16, stride, rounding); + break; + case 12: + Ops->V_Pass_Avrg_Up(dst, src, 16, stride, rounding); + break; + case 13: + Ops->H_Pass_Avrg(tmp, src, 17, stride, rounding); + Ops->V_Pass_Avrg_Up(dst, tmp, 16, stride, rounding); + break; + case 14: + Ops->H_Pass(tmp, src, 17, stride, rounding); + Ops->V_Pass_Avrg_Up( dst, tmp, 16, stride, rounding); + break; + case 15: + Ops->H_Pass_Avrg_Up(tmp, src, 17, stride, rounding); + Ops->V_Pass_Avrg_Up(dst, tmp, 16, stride, rounding); + break; + } +} + new_interpolate16x8_quarterpel(uint8_t * const cur, + uint8_t * const refn, + uint8_t * const refh, + uint8_t * const refv, + uint8_t * const refhv, + const uint32_t x, const uint32_t y, + const int32_t dx, const int dy, + const uint32_t stride, + const uint32_t rounding) + { + + const uint8_t *src; + uint8_t *dst; + uint8_t *tmp; + int32_t quads; + const XVID_QP_FUNCS *Ops; + + int32_t x_int, y_int; + + const int32_t xRef = (int)x*4 + dx; + const int32_t yRef = (int)y*4 + dy; + + Ops = (((struct global_all_dll *)Dll::Tls())->xvid_QP_Funcs); /* TODO: pass as argument */ + quads = (dx&3) | ((dy&3)<<2); + + x_int = xRef/4; + if (xRef < 0 && xRef % 4) + x_int--; + + y_int = yRef/4; + if (yRef < 0 && yRef % 4) + y_int--; + + dst = cur + y * stride + x; + src = refn + y_int * (int)stride + x_int; + + tmp = refh; /* we need at least a 16 x stride scratch block */ + + switch(quads) { + case 0: + (((struct global_all_dll *)Dll::Tls())->transfer8x8_copy)( dst, src, stride); + (((struct global_all_dll *)Dll::Tls())->transfer8x8_copy)( dst+8, src+8, stride); + break; + case 1: + Ops->H_Pass_Avrg(dst, src, 8, stride, rounding); + break; + case 2: + Ops->H_Pass(dst, src, 8, stride, rounding); + break; + case 3: + Ops->H_Pass_Avrg_Up(dst, src, 8, stride, rounding); + break; + case 4: + Ops->V_Pass_Avrg_8(dst, src, 16, stride, rounding); + break; + case 5: + Ops->H_Pass_Avrg(tmp, src, 9, stride, rounding); + Ops->V_Pass_Avrg_8(dst, tmp, 16, stride, rounding); + break; + case 6: + Ops->H_Pass(tmp, src, 9, stride, rounding); + Ops->V_Pass_Avrg_8(dst, tmp, 16, stride, rounding); + break; + case 7: + Ops->H_Pass_Avrg_Up(tmp, src, 9, stride, rounding); + Ops->V_Pass_Avrg_8(dst, tmp, 16, stride, rounding); + break; + case 8: + Ops->V_Pass_8(dst, src, 16, stride, rounding); + break; + case 9: + Ops->H_Pass_Avrg(tmp, src, 9, stride, rounding); + Ops->V_Pass_8(dst, tmp, 16, stride, rounding); + break; + case 10: + Ops->H_Pass(tmp, src, 9, stride, rounding); + Ops->V_Pass_8(dst, tmp, 16, stride, rounding); + break; + case 11: + Ops->H_Pass_Avrg_Up(tmp, src, 9, stride, rounding); + Ops->V_Pass_8(dst, tmp, 16, stride, rounding); + break; + case 12: + Ops->V_Pass_Avrg_Up_8(dst, src, 16, stride, rounding); + break; + case 13: + Ops->H_Pass_Avrg(tmp, src, 9, stride, rounding); + Ops->V_Pass_Avrg_Up_8(dst, tmp, 16, stride, rounding); + break; + case 14: + Ops->H_Pass(tmp, src, 9, stride, rounding); + Ops->V_Pass_Avrg_Up_8( dst, tmp, 16, stride, rounding); + break; + case 15: + Ops->H_Pass_Avrg_Up(tmp, src, 9, stride, rounding); + Ops->V_Pass_Avrg_Up_8(dst, tmp, 16, stride, rounding); + break; + } +} + new_interpolate8x8_quarterpel(uint8_t * const cur, + uint8_t * const refn, + uint8_t * const refh, + uint8_t * const refv, + uint8_t * const refhv, + const uint32_t x, const uint32_t y, + const int32_t dx, const int dy, + const uint32_t stride, + const uint32_t rounding) + { + + const uint8_t *src; + uint8_t *dst; + uint8_t *tmp; + int32_t quads; + const XVID_QP_FUNCS *Ops; + + int32_t x_int, y_int; + + const int32_t xRef = (int)x*4 + dx; + const int32_t yRef = (int)y*4 + dy; + + Ops = (((struct global_all_dll *)Dll::Tls())->xvid_QP_Funcs); /* TODO: pass as argument */ + quads = (dx&3) | ((dy&3)<<2); + + x_int = xRef/4; + if (xRef < 0 && xRef % 4) + x_int--; + + y_int = yRef/4; + if (yRef < 0 && yRef % 4) + y_int--; + + dst = cur + y * stride + x; + src = refn + y_int * (int)stride + x_int; + + tmp = refh; /* we need at least a 16 x stride scratch block */ + + switch(quads) { + case 0: + (((struct global_all_dll *)Dll::Tls())->transfer8x8_copy)( dst, src, stride); + break; + case 1: + Ops->H_Pass_Avrg_8(dst, src, 8, stride, rounding); + break; + case 2: + Ops->H_Pass_8(dst, src, 8, stride, rounding); + break; + case 3: + Ops->H_Pass_Avrg_Up_8(dst, src, 8, stride, rounding); + break; + case 4: + Ops->V_Pass_Avrg_8(dst, src, 8, stride, rounding); + break; + case 5: + Ops->H_Pass_Avrg_8(tmp, src, 9, stride, rounding); + Ops->V_Pass_Avrg_8(dst, tmp, 8, stride, rounding); + break; + case 6: + Ops->H_Pass_8(tmp, src, 9, stride, rounding); + Ops->V_Pass_Avrg_8(dst, tmp, 8, stride, rounding); + break; + case 7: + Ops->H_Pass_Avrg_Up_8(tmp, src, 9, stride, rounding); + Ops->V_Pass_Avrg_8(dst, tmp, 8, stride, rounding); + break; + case 8: + Ops->V_Pass_8(dst, src, 8, stride, rounding); + break; + case 9: + Ops->H_Pass_Avrg_8(tmp, src, 9, stride, rounding); + Ops->V_Pass_8(dst, tmp, 8, stride, rounding); + break; + case 10: + Ops->H_Pass_8(tmp, src, 9, stride, rounding); + Ops->V_Pass_8(dst, tmp, 8, stride, rounding); + break; + case 11: + Ops->H_Pass_Avrg_Up_8(tmp, src, 9, stride, rounding); + Ops->V_Pass_8(dst, tmp, 8, stride, rounding); + break; + case 12: + Ops->V_Pass_Avrg_Up_8(dst, src, 8, stride, rounding); + break; + case 13: + Ops->H_Pass_Avrg_8(tmp, src, 9, stride, rounding); + Ops->V_Pass_Avrg_Up_8(dst, tmp, 8, stride, rounding); + break; + case 14: + Ops->H_Pass_8(tmp, src, 9, stride, rounding); + Ops->V_Pass_Avrg_Up_8( dst, tmp, 8, stride, rounding); + break; + case 15: + Ops->H_Pass_Avrg_Up_8(tmp, src, 9, stride, rounding); + Ops->V_Pass_Avrg_Up_8(dst, tmp, 8, stride, rounding); + break; + } +} + /* Quarterpel FIR definition ****************************************************************************/ @@ -81,7 +380,7 @@ #define FUNC_HA_UP H_Pass_Avrg_Up_16_C #define FUNC_VA_UP V_Pass_Avrg_Up_16_C -#include __FILE__ /* self-include ourself */ +#include __FILE__ /* note: B-frame always uses Rnd=0... */ #define STORE(d,s) (d) = ( (s)+(d)+1 ) >> 1 @@ -92,7 +391,7 @@ #define FUNC_HA_UP H_Pass_Avrg_Up_16_Add_C #define FUNC_VA_UP V_Pass_Avrg_Up_16_Add_C -#include __FILE__ /* self-include ourself */ +#include __FILE__ #undef SIZE #undef TABLE @@ -110,7 +409,7 @@ #define FUNC_HA_UP H_Pass_Avrg_Up_8_C #define FUNC_VA_UP V_Pass_Avrg_Up_8_C -#include __FILE__ /* self-include ourself */ +#include __FILE__ /* note: B-frame always uses Rnd=0... */ #define STORE(d,s) (d) = ( (s)+(d)+1 ) >> 1 @@ -121,7 +420,7 @@ #define FUNC_HA_UP H_Pass_Avrg_Up_8_Add_C #define FUNC_VA_UP V_Pass_Avrg_Up_8_Add_C -#include __FILE__ /* self-include ourself */ +#include __FILE__ #undef SIZE #undef TABLE @@ -132,8 +431,6 @@ * TODO: embed in enc/dec structure? ****************************************************************************/ -XVID_QP_FUNCS *xvid_QP_Funcs = 0; -XVID_QP_FUNCS *xvid_QP_Add_Funcs = 0; /* plain C impl. declaration * TODO: should be declared elsewhere? @@ -209,7 +506,6 @@ ****************************************************************************/ #ifdef ARCH_IS_IA32 -uint16_t xvid_Expand_mmx[256][4]; /* 8b -> 64b expansion table */ #endif /* Alternate way of filtering (cf. USE_TABLES flag in qpel_mmx.asm) @@ -220,23 +516,6 @@ * Using Symmetries (and bswap) could reduce further * the memory to 7 tables (->14K). */ -int16_t xvid_FIR_1_0_0_0[256][4]; -int16_t xvid_FIR_3_1_0_0[256][4]; -int16_t xvid_FIR_6_3_1_0[256][4]; -int16_t xvid_FIR_14_3_2_1[256][4]; -int16_t xvid_FIR_20_6_3_1[256][4]; -int16_t xvid_FIR_20_20_6_3[256][4]; -int16_t xvid_FIR_23_19_6_3[256][4]; -int16_t xvid_FIR_7_20_20_6[256][4]; -int16_t xvid_FIR_6_20_20_6[256][4]; -int16_t xvid_FIR_6_20_20_7[256][4]; -int16_t xvid_FIR_3_6_20_20[256][4]; -int16_t xvid_FIR_3_6_19_23[256][4]; -int16_t xvid_FIR_1_3_6_20[256][4]; -int16_t xvid_FIR_1_2_3_14[256][4]; -int16_t xvid_FIR_0_1_3_6[256][4]; -int16_t xvid_FIR_0_0_1_3[256][4]; -int16_t xvid_FIR_0_0_0_1[256][4]; static void Init_FIR_Table(int16_t Tab[][4], int A, int B, int C, int D) @@ -257,32 +536,32 @@ int i; for(i=0; i<256; ++i) { - xvid_Expand_mmx[i][0] = i; - xvid_Expand_mmx[i][1] = i; - xvid_Expand_mmx[i][2] = i; - xvid_Expand_mmx[i][3] = i; + (((struct global_all_dll *)Dll::Tls())->xvid_Expand_mmx)[i][0] = i; + (((struct global_all_dll *)Dll::Tls())->xvid_Expand_mmx)[i][1] = i; + (((struct global_all_dll *)Dll::Tls())->xvid_Expand_mmx)[i][2] = i; + (((struct global_all_dll *)Dll::Tls())->xvid_Expand_mmx)[i][3] = i; } #endif /* Alternate way of filtering (cf. USE_TABLES flag in qpel_mmx.asm) */ - Init_FIR_Table(xvid_FIR_1_0_0_0, -1, 0, 0, 0); - Init_FIR_Table(xvid_FIR_3_1_0_0, 3, -1, 0, 0); - Init_FIR_Table(xvid_FIR_6_3_1_0, -6, 3, -1, 0); - Init_FIR_Table(xvid_FIR_14_3_2_1, 14, -3, 2, -1); - Init_FIR_Table(xvid_FIR_20_6_3_1, 20, -6, 3, -1); - Init_FIR_Table(xvid_FIR_20_20_6_3, 20, 20, -6, 3); - Init_FIR_Table(xvid_FIR_23_19_6_3, 23, 19, -6, 3); - Init_FIR_Table(xvid_FIR_7_20_20_6, -7, 20, 20, -6); - Init_FIR_Table(xvid_FIR_6_20_20_6, -6, 20, 20, -6); - Init_FIR_Table(xvid_FIR_6_20_20_7, -6, 20, 20, -7); - Init_FIR_Table(xvid_FIR_3_6_20_20, 3, -6, 20, 20); - Init_FIR_Table(xvid_FIR_3_6_19_23, 3, -6, 19, 23); - Init_FIR_Table(xvid_FIR_1_3_6_20, -1, 3, -6, 20); - Init_FIR_Table(xvid_FIR_1_2_3_14, -1, 2, -3, 14); - Init_FIR_Table(xvid_FIR_0_1_3_6, 0, -1, 3, -6); - Init_FIR_Table(xvid_FIR_0_0_1_3, 0, 0, -1, 3); - Init_FIR_Table(xvid_FIR_0_0_0_1, 0, 0, 0, -1); + Init_FIR_Table((((struct global_all_dll *)Dll::Tls())->xvid_FIR_1_0_0_0), -1, 0, 0, 0); + Init_FIR_Table((((struct global_all_dll *)Dll::Tls())->xvid_FIR_3_1_0_0), 3, -1, 0, 0); + Init_FIR_Table((((struct global_all_dll *)Dll::Tls())->xvid_FIR_6_3_1_0), -6, 3, -1, 0); + Init_FIR_Table((((struct global_all_dll *)Dll::Tls())->xvid_FIR_14_3_2_1), 14, -3, 2, -1); + Init_FIR_Table((((struct global_all_dll *)Dll::Tls())->xvid_FIR_20_6_3_1), 20, -6, 3, -1); + Init_FIR_Table((((struct global_all_dll *)Dll::Tls())->xvid_FIR_20_20_6_3), 20, 20, -6, 3); + Init_FIR_Table((((struct global_all_dll *)Dll::Tls())->xvid_FIR_23_19_6_3), 23, 19, -6, 3); + Init_FIR_Table((((struct global_all_dll *)Dll::Tls())->xvid_FIR_7_20_20_6), -7, 20, 20, -6); + Init_FIR_Table((((struct global_all_dll *)Dll::Tls())->xvid_FIR_6_20_20_6), -6, 20, 20, -6); + Init_FIR_Table((((struct global_all_dll *)Dll::Tls())->xvid_FIR_6_20_20_7), -6, 20, 20, -7); + Init_FIR_Table((((struct global_all_dll *)Dll::Tls())->xvid_FIR_3_6_20_20), 3, -6, 20, 20); + Init_FIR_Table((((struct global_all_dll *)Dll::Tls())->xvid_FIR_3_6_19_23), 3, -6, 19, 23); + Init_FIR_Table((((struct global_all_dll *)Dll::Tls())->xvid_FIR_1_3_6_20), -1, 3, -6, 20); + Init_FIR_Table((((struct global_all_dll *)Dll::Tls())->xvid_FIR_1_2_3_14), -1, 2, -3, 14); + Init_FIR_Table((((struct global_all_dll *)Dll::Tls())->xvid_FIR_0_1_3_6), 0, -1, 3, -6); + Init_FIR_Table((((struct global_all_dll *)Dll::Tls())->xvid_FIR_0_0_1_3), 0, 0, -1, 3); + Init_FIR_Table((((struct global_all_dll *)Dll::Tls())->xvid_FIR_0_0_0_1), 0, 0, 0, -1); } diff -bBru src-ori/image/qpel.h src/image/qpel.h --- src-ori/image/qpel.h Wed Jan 12 11:26:28 2005 +++ src/image/qpel.h Wed Jan 12 12:37:55 2005 @@ -74,8 +93,8 @@ extern const XVID_QP_FUNCS xvid_QP_Add_Funcs_mmx; #endif -extern const XVID_QP_FUNCS *xvid_QP_Funcs; /* <- main pointer for enc/dec structure */ -extern const XVID_QP_FUNCS *xvid_QP_Add_Funcs; /* <- main pointer for enc/dec structure */ + + /***************************************************************************** * macros @@ -97,8 +116,7 @@ ****************************************************************************/ -static void __inline -new_interpolate16x16_quarterpel(uint8_t * const cur, +extern new_interpolate16x16_quarterpel(uint8_t * const cur, uint8_t * const refn, uint8_t * const refh, uint8_t * const refv, @@ -107,100 +125,9 @@ const int32_t dx, const int dy, const uint32_t stride, const uint32_t rounding) -{ - const uint8_t *src; - uint8_t *dst; - uint8_t *tmp; - int32_t quads; - const XVID_QP_FUNCS *Ops; - - int32_t x_int, y_int; - - const int32_t xRef = (int)x*4 + dx; - const int32_t yRef = (int)y*4 + dy; - - Ops = xvid_QP_Funcs; /* TODO: pass as argument */ - quads = (dx&3) | ((dy&3)<<2); - - x_int = xRef/4; - if (xRef < 0 && xRef % 4) - x_int--; - - y_int = yRef/4; - if (yRef < 0 && yRef % 4) - y_int--; - - dst = cur + y * stride + x; - src = refn + y_int * (int)stride + x_int; - - tmp = refh; /* we need at least a 16 x stride scratch block */ - - switch(quads) { - case 0: - transfer8x8_copy( dst, src, stride); - transfer8x8_copy( dst+8, src+8, stride); - transfer8x8_copy( dst+8*stride, src+8*stride, stride); - transfer8x8_copy( dst+8*stride+8, src+8*stride+8, stride); - break; - case 1: - Ops->H_Pass_Avrg(dst, src, 16, stride, rounding); - break; - case 2: - Ops->H_Pass(dst, src, 16, stride, rounding); - break; - case 3: - Ops->H_Pass_Avrg_Up(dst, src, 16, stride, rounding); - break; - case 4: - Ops->V_Pass_Avrg(dst, src, 16, stride, rounding); - break; - case 5: - Ops->H_Pass_Avrg(tmp, src, 17, stride, rounding); - Ops->V_Pass_Avrg(dst, tmp, 16, stride, rounding); - break; - case 6: - Ops->H_Pass(tmp, src, 17, stride, rounding); - Ops->V_Pass_Avrg(dst, tmp, 16, stride, rounding); - break; - case 7: - Ops->H_Pass_Avrg_Up(tmp, src, 17, stride, rounding); - Ops->V_Pass_Avrg(dst, tmp, 16, stride, rounding); - break; - case 8: - Ops->V_Pass(dst, src, 16, stride, rounding); - break; - case 9: - Ops->H_Pass_Avrg(tmp, src, 17, stride, rounding); - Ops->V_Pass(dst, tmp, 16, stride, rounding); - break; - case 10: - Ops->H_Pass(tmp, src, 17, stride, rounding); - Ops->V_Pass(dst, tmp, 16, stride, rounding); - break; - case 11: - Ops->H_Pass_Avrg_Up(tmp, src, 17, stride, rounding); - Ops->V_Pass(dst, tmp, 16, stride, rounding); - break; - case 12: - Ops->V_Pass_Avrg_Up(dst, src, 16, stride, rounding); - break; - case 13: - Ops->H_Pass_Avrg(tmp, src, 17, stride, rounding); - Ops->V_Pass_Avrg_Up(dst, tmp, 16, stride, rounding); - break; - case 14: - Ops->H_Pass(tmp, src, 17, stride, rounding); - Ops->V_Pass_Avrg_Up( dst, tmp, 16, stride, rounding); - break; - case 15: - Ops->H_Pass_Avrg_Up(tmp, src, 17, stride, rounding); - Ops->V_Pass_Avrg_Up(dst, tmp, 16, stride, rounding); - break; - } -} +; -static void __inline -new_interpolate16x8_quarterpel(uint8_t * const cur, +extern new_interpolate16x8_quarterpel(uint8_t * const cur, uint8_t * const refn, uint8_t * const refh, uint8_t * const refv, @@ -209,98 +136,9 @@ const int32_t dx, const int dy, const uint32_t stride, const uint32_t rounding) -{ - const uint8_t *src; - uint8_t *dst; - uint8_t *tmp; - int32_t quads; - const XVID_QP_FUNCS *Ops; - - int32_t x_int, y_int; - - const int32_t xRef = (int)x*4 + dx; - const int32_t yRef = (int)y*4 + dy; - - Ops = xvid_QP_Funcs; /* TODO: pass as argument */ - quads = (dx&3) | ((dy&3)<<2); - - x_int = xRef/4; - if (xRef < 0 && xRef % 4) - x_int--; - - y_int = yRef/4; - if (yRef < 0 && yRef % 4) - y_int--; - - dst = cur + y * stride + x; - src = refn + y_int * (int)stride + x_int; - - tmp = refh; /* we need at least a 16 x stride scratch block */ - - switch(quads) { - case 0: - transfer8x8_copy( dst, src, stride); - transfer8x8_copy( dst+8, src+8, stride); - break; - case 1: - Ops->H_Pass_Avrg(dst, src, 8, stride, rounding); - break; - case 2: - Ops->H_Pass(dst, src, 8, stride, rounding); - break; - case 3: - Ops->H_Pass_Avrg_Up(dst, src, 8, stride, rounding); - break; - case 4: - Ops->V_Pass_Avrg_8(dst, src, 16, stride, rounding); - break; - case 5: - Ops->H_Pass_Avrg(tmp, src, 9, stride, rounding); - Ops->V_Pass_Avrg_8(dst, tmp, 16, stride, rounding); - break; - case 6: - Ops->H_Pass(tmp, src, 9, stride, rounding); - Ops->V_Pass_Avrg_8(dst, tmp, 16, stride, rounding); - break; - case 7: - Ops->H_Pass_Avrg_Up(tmp, src, 9, stride, rounding); - Ops->V_Pass_Avrg_8(dst, tmp, 16, stride, rounding); - break; - case 8: - Ops->V_Pass_8(dst, src, 16, stride, rounding); - break; - case 9: - Ops->H_Pass_Avrg(tmp, src, 9, stride, rounding); - Ops->V_Pass_8(dst, tmp, 16, stride, rounding); - break; - case 10: - Ops->H_Pass(tmp, src, 9, stride, rounding); - Ops->V_Pass_8(dst, tmp, 16, stride, rounding); - break; - case 11: - Ops->H_Pass_Avrg_Up(tmp, src, 9, stride, rounding); - Ops->V_Pass_8(dst, tmp, 16, stride, rounding); - break; - case 12: - Ops->V_Pass_Avrg_Up_8(dst, src, 16, stride, rounding); - break; - case 13: - Ops->H_Pass_Avrg(tmp, src, 9, stride, rounding); - Ops->V_Pass_Avrg_Up_8(dst, tmp, 16, stride, rounding); - break; - case 14: - Ops->H_Pass(tmp, src, 9, stride, rounding); - Ops->V_Pass_Avrg_Up_8( dst, tmp, 16, stride, rounding); - break; - case 15: - Ops->H_Pass_Avrg_Up(tmp, src, 9, stride, rounding); - Ops->V_Pass_Avrg_Up_8(dst, tmp, 16, stride, rounding); - break; - } -} +; -static void __inline -new_interpolate8x8_quarterpel(uint8_t * const cur, +extern new_interpolate8x8_quarterpel(uint8_t * const cur, uint8_t * const refn, uint8_t * const refh, uint8_t * const refv, @@ -309,93 +147,6 @@ const int32_t dx, const int dy, const uint32_t stride, const uint32_t rounding) -{ - const uint8_t *src; - uint8_t *dst; - uint8_t *tmp; - int32_t quads; - const XVID_QP_FUNCS *Ops; - - int32_t x_int, y_int; - - const int32_t xRef = (int)x*4 + dx; - const int32_t yRef = (int)y*4 + dy; - - Ops = xvid_QP_Funcs; /* TODO: pass as argument */ - quads = (dx&3) | ((dy&3)<<2); - - x_int = xRef/4; - if (xRef < 0 && xRef % 4) - x_int--; - - y_int = yRef/4; - if (yRef < 0 && yRef % 4) - y_int--; - - dst = cur + y * stride + x; - src = refn + y_int * (int)stride + x_int; - - tmp = refh; /* we need at least a 16 x stride scratch block */ - - switch(quads) { - case 0: - transfer8x8_copy( dst, src, stride); - break; - case 1: - Ops->H_Pass_Avrg_8(dst, src, 8, stride, rounding); - break; - case 2: - Ops->H_Pass_8(dst, src, 8, stride, rounding); - break; - case 3: - Ops->H_Pass_Avrg_Up_8(dst, src, 8, stride, rounding); - break; - case 4: - Ops->V_Pass_Avrg_8(dst, src, 8, stride, rounding); - break; - case 5: - Ops->H_Pass_Avrg_8(tmp, src, 9, stride, rounding); - Ops->V_Pass_Avrg_8(dst, tmp, 8, stride, rounding); - break; - case 6: - Ops->H_Pass_8(tmp, src, 9, stride, rounding); - Ops->V_Pass_Avrg_8(dst, tmp, 8, stride, rounding); - break; - case 7: - Ops->H_Pass_Avrg_Up_8(tmp, src, 9, stride, rounding); - Ops->V_Pass_Avrg_8(dst, tmp, 8, stride, rounding); - break; - case 8: - Ops->V_Pass_8(dst, src, 8, stride, rounding); - break; - case 9: - Ops->H_Pass_Avrg_8(tmp, src, 9, stride, rounding); - Ops->V_Pass_8(dst, tmp, 8, stride, rounding); - break; - case 10: - Ops->H_Pass_8(tmp, src, 9, stride, rounding); - Ops->V_Pass_8(dst, tmp, 8, stride, rounding); - break; - case 11: - Ops->H_Pass_Avrg_Up_8(tmp, src, 9, stride, rounding); - Ops->V_Pass_8(dst, tmp, 8, stride, rounding); - break; - case 12: - Ops->V_Pass_Avrg_Up_8(dst, src, 8, stride, rounding); - break; - case 13: - Ops->H_Pass_Avrg_8(tmp, src, 9, stride, rounding); - Ops->V_Pass_Avrg_Up_8(dst, tmp, 8, stride, rounding); - break; - case 14: - Ops->H_Pass_8(tmp, src, 9, stride, rounding); - Ops->V_Pass_Avrg_Up_8( dst, tmp, 8, stride, rounding); - break; - case 15: - Ops->H_Pass_Avrg_Up_8(tmp, src, 9, stride, rounding); - Ops->V_Pass_Avrg_Up_8(dst, tmp, 8, stride, rounding); - break; - } -} +; #endif /* _XVID_QPEL_H_ */ diff -bBru src-ori/image/reduced.c src/image/reduced.c --- src-ori/image/reduced.c Wed Jan 12 10:32:43 2005 +++ src/image/reduced.c Wed Jan 12 12:40:33 2005 @@ -23,17 +23,9 @@ * ****************************************************************************/ -#include "../portab.h" -#include "../global.h" -#include "reduced.h" - -/* function pointers */ -COPY_UPSAMPLED_8X8_16TO8 * copy_upsampled_8x8_16to8; -ADD_UPSAMPLED_8X8_16TO8 * add_upsampled_8x8_16to8; -VFILTER_31 * vfilter_31; -HFILTER_31 * hfilter_31; -FILTER_18X18_TO_8X8 * filter_18x18_to_8x8; -FILTER_DIFF_18X18_TO_8X8 * filter_diff_18x18_to_8x8; +#include +#include "../global_all_dll.h" + /*---------------------------------------------------------------------------- * Upsampling (1/3/3/1) filter diff -bBru src-ori/image/reduced.h src/image/reduced.h --- src-ori/image/reduced.h Wed Jan 12 10:32:43 2005 +++ src/image/reduced.h Wed Jan 12 12:37:54 2005 @@ -41,40 +41,40 @@ typedef void FILTER_DIFF_18X18_TO_8X8(int16_t *Dst, const uint8_t *Src, const int BpS); -extern COPY_UPSAMPLED_8X8_16TO8 * copy_upsampled_8x8_16to8; + extern COPY_UPSAMPLED_8X8_16TO8 xvid_Copy_Upsampled_8x8_16To8_C; #ifdef ARCH_IS_IA32 extern COPY_UPSAMPLED_8X8_16TO8 xvid_Copy_Upsampled_8x8_16To8_mmx; extern COPY_UPSAMPLED_8X8_16TO8 xvid_Copy_Upsampled_8x8_16To8_xmm; #endif -extern ADD_UPSAMPLED_8X8_16TO8 * add_upsampled_8x8_16to8; + extern ADD_UPSAMPLED_8X8_16TO8 xvid_Add_Upsampled_8x8_16To8_C; #ifdef ARCH_IS_IA32 extern ADD_UPSAMPLED_8X8_16TO8 xvid_Add_Upsampled_8x8_16To8_mmx; extern ADD_UPSAMPLED_8X8_16TO8 xvid_Add_Upsampled_8x8_16To8_xmm; #endif -extern VFILTER_31 * vfilter_31; + extern VFILTER_31 xvid_VFilter_31_C; #ifdef ARCH_IS_IA32 extern VFILTER_31 xvid_VFilter_31_x86; #endif -extern HFILTER_31 * hfilter_31; + extern HFILTER_31 xvid_HFilter_31_C; #ifdef ARCH_IS_IA32 extern HFILTER_31 xvid_HFilter_31_x86; extern HFILTER_31 xvid_HFilter_31_mmx; #endif -extern FILTER_18X18_TO_8X8 * filter_18x18_to_8x8; + extern FILTER_18X18_TO_8X8 xvid_Filter_18x18_To_8x8_C; #ifdef ARCH_IS_IA32 extern FILTER_18X18_TO_8X8 xvid_Filter_18x18_To_8x8_mmx; #endif -extern FILTER_DIFF_18X18_TO_8X8 * filter_diff_18x18_to_8x8; + extern FILTER_DIFF_18X18_TO_8X8 xvid_Filter_Diff_18x18_To_8x8_C; #ifdef ARCH_IS_IA32 extern FILTER_DIFF_18X18_TO_8X8 xvid_Filter_Diff_18x18_To_8x8_mmx; Only in src-ori/image: x86_asm diff -bBru src-ori/motion/estimation.h src/motion/estimation.h --- src-ori/motion/estimation.h Wed Jan 12 10:32:43 2005 +++ src/motion/estimation.h Wed Jan 12 12:38:13 2005 @@ -71,59 +71,110 @@ typedef struct { + /* data modified by CheckCandidates */ + int32_t iMinSAD[5]; /* smallest SADs found so far */ + VECTOR currentMV[5]; /* best vectors found so far */ + VECTOR currentQMV[5]; /* as above, but used during qpel search */ + int temp[4]; /* temporary space */ + unsigned int dir; /* 'direction', set when better vector is found */ + int chromaX, chromaY, chromaSAD; /* info to make ChromaSAD faster */ + VECTOR currentQMV2; /* extra vector for SubpelRefine_fast */ + int32_t iMinSAD2; /* extra SAD value for SubpelRefine_fast */ + + /* general fields */ + int max_dx, min_dx, max_dy, min_dy; /* maximum range */ + uint32_t rounding; /* rounding type in use */ + VECTOR predMV; /* vector which predicts current vector */ + const uint8_t * RefP[6]; /* reference pictures - N, V, H, HV, cU, cV */ + const uint8_t * Cur; /* current picture */ + const uint8_t *CurU, *CurV; /* current picture - chroma planes */ + + uint8_t * RefQ; /* temporary space for interpolations */ + uint32_t lambda16; /* how much vector bits weight */ + uint32_t lambda8; /* as above - for inter4v mode */ + uint32_t iEdgedWidth; /* picture's stride */ + uint32_t iFcode; /* current fcode */ + + int qpel; /* if we're coding in qpel mode */ + int qpel_precision; /* if X and Y are in qpel precision (refinement probably) */ + int chroma; /* should we include chroma SAD? */ + int rrv; /* are we using reduced resolution? */ + + /* fields for interpolate and direct modes */ + const uint8_t * b_RefP[6]; /* backward reference pictures - N, V, H, HV, cU, cV */ + VECTOR bpredMV; /* backward prediction - used interpolate mode only */ + uint32_t bFcode; /* backward fcode - used as above */ + + /* fields for direct mode */ + VECTOR directmvF[4]; /* scaled reference vectors */ + VECTOR directmvB[4]; /* as above */ + const VECTOR * referencemv; /* pointer to not-scaled reference vectors */ + + /* BITS/R-D stuff */ + int16_t * dctSpace; /* temporary space for dct */ + uint32_t iQuant; /* current quant */ + uint32_t quant_type; /* current quant type */ + unsigned int cbp[2]; /* CBP of the best vector found so far + cbp for inter4v search */ + const uint16_t * scan_table; /* current scan table */ + const uint16_t * mpeg_quant_matrices; /* current MPEG quantization matrices */ + + } SearchData; + typedef void(CheckFunc)(const int x, const int y, SearchData * const Data, + const unsigned int Direction); + CheckFunc CheckCandidate16no4v; /* shared between p-vop and b-vop search */ uint8_t * diff -bBru src-ori/motion/estimation_bvop.c src/motion/estimation_bvop.c --- src-ori/motion/estimation_bvop.c Wed Jan 12 10:32:43 2005 +++ src/motion/estimation_bvop.c Wed Jan 12 12:44:30 2005 @@ -26,18 +26,9 @@ ****************************************************************************/ -#include -#include -#include -#include /* memcpy */ - -#include "../encoder.h" -#include "../global.h" -#include "../image/interpolate8x8.h" -#include "estimation.h" -#include "motion.h" -#include "sad.h" -#include "motion_inlines.h" +#include +#include "../global_all_dll.h" + static int32_t ChromaSAD2(const int fx, const int fy, const int bx, const int by, @@ -49,9 +40,9 @@ const INTERPOLATE8X8_PTR interpolate8x8_halfpel[] = { NULL, - interpolate8x8_halfpel_v, - interpolate8x8_halfpel_h, - interpolate8x8_halfpel_hv + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_halfpel_v), + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_halfpel_h), + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_halfpel_hv) }; int offset = (fx>>1) + (fy>>1)*stride; @@ -80,8 +71,8 @@ b_refv = (uint8_t*)data->b_RefP[5] + offset; } - sad = sad8bi(data->CurU, b_refu, f_refu, stride); - sad += sad8bi(data->CurV, b_refv, f_refv, stride); + sad = (((struct global_all_dll *)Dll::Tls())->sad8bi)(data->CurU, b_refu, f_refu, stride); + sad += (((struct global_all_dll *)Dll::Tls())->sad8bi)(data->CurV, b_refv, f_refv, stride); return sad; } @@ -128,7 +119,7 @@ t = d_mv_bits(xf, yf, data->predMV, data->iFcode, data->qpel^data->qpel_precision, 0) + d_mv_bits(xb, yb, data->bpredMV, data->iFcode, data->qpel^data->qpel_precision, 0); - sad = sad16bi(data->Cur, ReferenceF, ReferenceB, data->iEdgedWidth); + sad = (((struct global_all_dll *)Dll::Tls())->sad16bi)(data->Cur, ReferenceF, ReferenceB, data->iEdgedWidth); sad += (data->lambda16 * t * sad)>>10; if (data->chroma && sad < *data->iMinSAD) @@ -178,7 +169,7 @@ t = d_mv_bits(xf, yf, data->predMV, data->iFcode, data->qpel^data->qpel_precision, 0) + d_mv_bits(xb, yb, data->bpredMV, data->iFcode, data->qpel^data->qpel_precision, 0); - sad = sad16bi(data->Cur, ReferenceF, ReferenceB, data->iEdgedWidth); + sad = (((struct global_all_dll *)Dll::Tls())->sad16bi)(data->Cur, ReferenceF, ReferenceB, data->iEdgedWidth); sad += (data->lambda16 * t * sad)>>10; if (data->chroma && sad < *data->iMinSAD) @@ -247,7 +238,7 @@ ReferenceF = xvid_me_interpolate8x8qpel(mvs.x, mvs.y, k, 0, data); ReferenceB = xvid_me_interpolate8x8qpel(b_mvs.x, b_mvs.y, k, 1, data); - sad += sad8bi(data->Cur + 8*(k&1) + 8*(k>>1)*(data->iEdgedWidth), + sad += (((struct global_all_dll *)Dll::Tls())->sad8bi)(data->Cur + 8*(k&1) + 8*(k>>1)*(data->iEdgedWidth), ReferenceF, ReferenceB, data->iEdgedWidth); if (sad > *(data->iMinSAD)) return; } @@ -304,7 +295,7 @@ ReferenceB = GetReferenceB(b_mvs.x, b_mvs.y, 1, data); } - sad = sad16bi(data->Cur, ReferenceF, ReferenceB, data->iEdgedWidth); + sad = (((struct global_all_dll *)Dll::Tls())->sad16bi)(data->Cur, ReferenceF, ReferenceB, data->iEdgedWidth); sad += (data->lambda16 * d_mv_bits(x, y, zeroMV, 1, 0, 0) * sad)>>10; if (data->chroma && sad < *data->iMinSAD) @@ -345,7 +336,7 @@ t = d_mv_bits(x, y, data->predMV, data->iFcode, data->qpel^data->qpel_precision, data->rrv); - sad = sad16(data->Cur, Reference, data->iEdgedWidth, 256*4096); + sad = (((struct global_all_dll *)Dll::Tls())->sad16)(data->Cur, Reference, data->iEdgedWidth, 256*4096); sad += (data->lambda16 * t * sad)>>10; if (data->chroma && sad < *data->iMinSAD) @@ -377,7 +368,7 @@ t = d_mv_bits(x, y, data->predMV, data->iFcode, data->qpel^data->qpel_precision, data->rrv); - sad = sad16(data->Cur, Reference, data->iEdgedWidth, 256*4096); + sad = (((struct global_all_dll *)Dll::Tls())->sad16)(data->Cur, Reference, data->iEdgedWidth, 256*4096); sad += (data->lambda16 * t * sad)>>10; if (data->chroma && sad < *data->iMinSAD) @@ -573,14 +564,14 @@ b_dy = (b_dy >> 3) + roundtab_76[b_dy & 0xf]; b_dx = (b_dx >> 3) + roundtab_76[b_dx & 0xf]; - sum = sad8bi(pCur->u + 8 * x + 8 * y * stride, + sum = (((struct global_all_dll *)Dll::Tls())->sad8bi)(pCur->u + 8 * x + 8 * y * stride, f_Ref->u + ((int)y*8 + dy/2) * (int)stride + (int)x*8 + dx/2, b_Ref->u + ((int)y*8 + b_dy/2) * (int)stride + (int)x*8 + b_dx/2, stride); if (sum >= MAX_CHROMA_SAD_FOR_SKIP * (int)Data->iQuant) return; /* no skip */ - sum += sad8bi(pCur->v + 8*x + 8 * y * stride, + sum += (((struct global_all_dll *)Dll::Tls())->sad8bi)(pCur->v + 8*x + 8 * y * stride, f_Ref->v + ((int)y*8 + dy/2) * (int)stride + (int)x*8 + dx/2, b_Ref->v + ((int)y*8 + b_dy/2) * (int)stride + (int)x*8 + b_dx/2, stride); diff -bBru src-ori/motion/estimation_common.c src/motion/estimation_common.c --- src-ori/motion/estimation_common.c Wed Jan 12 10:32:43 2005 +++ src/motion/estimation_common.c Wed Jan 12 12:44:46 2005 @@ -25,13 +25,8 @@ * ****************************************************************************/ -#include "../encoder.h" -#include "../global.h" -#include "../image/interpolate8x8.h" -#include "estimation.h" -#include "motion.h" -#include "sad.h" -#include "motion_inlines.h" +#include +#include "../global_all_dll.h" /***************************************************************************** @@ -91,21 +86,21 @@ switch (((dx & 1) << 1) | (dy & 1)) { case 0: - sad = sad8(data->CurU, data->RefP[4] + offset, stride); - sad += sad8(data->CurV, data->RefP[5] + offset, stride); + sad = (((struct global_all_dll *)Dll::Tls())->sad8)(data->CurU, data->RefP[4] + offset, stride); + sad += (((struct global_all_dll *)Dll::Tls())->sad8)(data->CurV, data->RefP[5] + offset, stride); break; case 1: next = stride; case 2: - sad = sad8bi(data->CurU, data->RefP[4] + offset, data->RefP[4] + offset + next, stride); - sad += sad8bi(data->CurV, data->RefP[5] + offset, data->RefP[5] + offset + next, stride); + sad = (((struct global_all_dll *)Dll::Tls())->sad8bi)(data->CurU, data->RefP[4] + offset, data->RefP[4] + offset + next, stride); + sad += (((struct global_all_dll *)Dll::Tls())->sad8bi)(data->CurV, data->RefP[5] + offset, data->RefP[5] + offset + next, stride); break; default: - interpolate8x8_halfpel_hv(data->RefQ, data->RefP[4] + offset, stride, data->rounding); - sad = sad8(data->CurU, data->RefQ, stride); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_halfpel_hv)(data->RefQ, data->RefP[4] + offset, stride, data->rounding); + sad = (((struct global_all_dll *)Dll::Tls())->sad8)(data->CurU, data->RefQ, stride); - interpolate8x8_halfpel_hv(data->RefQ, data->RefP[5] + offset, stride, data->rounding); - sad += sad8(data->CurV, data->RefQ, stride); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_halfpel_hv)(data->RefQ, data->RefP[5] + offset, stride, data->rounding); + sad += (((struct global_all_dll *)Dll::Tls())->sad8)(data->CurV, data->RefQ, stride); break; } data->chromaSAD = sad; /* backup, part 2 */ @@ -134,19 +129,19 @@ ref2 += 8 * (block&1) + 8 * (block>>1) * iEdgedWidth; ref3 += 8 * (block&1) + 8 * (block>>1) * iEdgedWidth; ref4 += 8 * (block&1) + 8 * (block>>1) * iEdgedWidth; - interpolate8x8_avg4(Reference, ref1, ref2, ref3, ref4, iEdgedWidth, rounding); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_avg4)(Reference, ref1, ref2, ref3, ref4, iEdgedWidth, rounding); break; case 1: /* x halfpel, y qpel - top or bottom during qpel refinement */ ref2 = GetReferenceB(halfpel_x, y - halfpel_y, dir, data); ref2 += 8 * (block&1) + 8 * (block>>1) * iEdgedWidth; - interpolate8x8_avg2(Reference, ref1, ref2, iEdgedWidth, rounding, 8); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_avg2)(Reference, ref1, ref2, iEdgedWidth, rounding, 8); break; case 2: /* x qpel, y halfpel - left or right during qpel refinement */ ref2 = GetReferenceB(x - halfpel_x, halfpel_y, dir, data); ref2 += 8 * (block&1) + 8 * (block>>1) * iEdgedWidth; - interpolate8x8_avg2(Reference, ref1, ref2, iEdgedWidth, rounding, 8); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_avg2)(Reference, ref1, ref2, iEdgedWidth, rounding, 8); break; default: /* pure halfpel position */ @@ -177,26 +172,26 @@ ref2 = GetReferenceB(halfpel_x, y - halfpel_y, dir, data); ref3 = GetReferenceB(x - halfpel_x, halfpel_y, dir, data); ref4 = GetReferenceB(x - halfpel_x, y - halfpel_y, dir, data); - interpolate8x8_avg4(Reference, ref1, ref2, ref3, ref4, iEdgedWidth, rounding); - interpolate8x8_avg4(Reference+8, ref1+8, ref2+8, ref3+8, ref4+8, iEdgedWidth, rounding); - interpolate8x8_avg4(Reference+8*iEdgedWidth, ref1+8*iEdgedWidth, ref2+8*iEdgedWidth, ref3+8*iEdgedWidth, ref4+8*iEdgedWidth, iEdgedWidth, rounding); - interpolate8x8_avg4(Reference+8*iEdgedWidth+8, ref1+8*iEdgedWidth+8, ref2+8*iEdgedWidth+8, ref3+8*iEdgedWidth+8, ref4+8*iEdgedWidth+8, iEdgedWidth, rounding); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_avg4)(Reference, ref1, ref2, ref3, ref4, iEdgedWidth, rounding); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_avg4)(Reference+8, ref1+8, ref2+8, ref3+8, ref4+8, iEdgedWidth, rounding); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_avg4)(Reference+8*iEdgedWidth, ref1+8*iEdgedWidth, ref2+8*iEdgedWidth, ref3+8*iEdgedWidth, ref4+8*iEdgedWidth, iEdgedWidth, rounding); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_avg4)(Reference+8*iEdgedWidth+8, ref1+8*iEdgedWidth+8, ref2+8*iEdgedWidth+8, ref3+8*iEdgedWidth+8, ref4+8*iEdgedWidth+8, iEdgedWidth, rounding); break; case 1: /* x halfpel, y qpel - top or bottom during qpel refinement */ ref2 = GetReferenceB(halfpel_x, y - halfpel_y, dir, data); - interpolate8x8_avg2(Reference, ref1, ref2, iEdgedWidth, rounding, 8); - interpolate8x8_avg2(Reference+8, ref1+8, ref2+8, iEdgedWidth, rounding, 8); - interpolate8x8_avg2(Reference+8*iEdgedWidth, ref1+8*iEdgedWidth, ref2+8*iEdgedWidth, iEdgedWidth, rounding, 8); - interpolate8x8_avg2(Reference+8*iEdgedWidth+8, ref1+8*iEdgedWidth+8, ref2+8*iEdgedWidth+8, iEdgedWidth, rounding, 8); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_avg2)(Reference, ref1, ref2, iEdgedWidth, rounding, 8); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_avg2)(Reference+8, ref1+8, ref2+8, iEdgedWidth, rounding, 8); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_avg2)(Reference+8*iEdgedWidth, ref1+8*iEdgedWidth, ref2+8*iEdgedWidth, iEdgedWidth, rounding, 8); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_avg2)(Reference+8*iEdgedWidth+8, ref1+8*iEdgedWidth+8, ref2+8*iEdgedWidth+8, iEdgedWidth, rounding, 8); break; case 2: /* x qpel, y halfpel - left or right during qpel refinement */ ref2 = GetReferenceB(x - halfpel_x, halfpel_y, dir, data); - interpolate8x8_avg2(Reference, ref1, ref2, iEdgedWidth, rounding, 8); - interpolate8x8_avg2(Reference+8, ref1+8, ref2+8, iEdgedWidth, rounding, 8); - interpolate8x8_avg2(Reference+8*iEdgedWidth, ref1+8*iEdgedWidth, ref2+8*iEdgedWidth, iEdgedWidth, rounding, 8); - interpolate8x8_avg2(Reference+8*iEdgedWidth+8, ref1+8*iEdgedWidth+8, ref2+8*iEdgedWidth+8, iEdgedWidth, rounding, 8); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_avg2)(Reference, ref1, ref2, iEdgedWidth, rounding, 8); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_avg2)(Reference+8, ref1+8, ref2+8, iEdgedWidth, rounding, 8); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_avg2)(Reference+8*iEdgedWidth, ref1+8*iEdgedWidth, ref2+8*iEdgedWidth, iEdgedWidth, rounding, 8); + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_avg2)(Reference+8*iEdgedWidth+8, ref1+8*iEdgedWidth+8, ref2+8*iEdgedWidth+8, iEdgedWidth, rounding, 8); break; diff -bBru src-ori/motion/estimation_gmc.c src/motion/estimation_gmc.c --- src-ori/motion/estimation_gmc.c Wed Jan 12 10:32:43 2005 +++ src/motion/estimation_gmc.c Wed Jan 12 12:45:23 2005 @@ -23,20 +23,9 @@ * ****************************************************************************/ -#include -#include -#include -#include -#include - -#include "../encoder.h" -#include "../prediction/mbprediction.h" -#include "estimation.h" -#include "motion.h" -#include "sad.h" -#include "gmc.h" -#include "../utils/emms.h" -#include "motion_inlines.h" +#include +#include "../global_all_dll.h" + static void CheckCandidate16I(const int x, const int y, SearchData * const data, const unsigned int Direction) @@ -49,7 +38,7 @@ Reference = GetReference(x, y, data); - sad = sad16(data->Cur, Reference, data->iEdgedWidth, 256*4096); + sad = (((struct global_all_dll *)Dll::Tls())->sad16)(data->Cur, Reference, data->iEdgedWidth, 256*4096); if (sad < data->iMinSAD[0]) { data->iMinSAD[0] = sad; @@ -137,7 +126,7 @@ Data.iFcode = current->fcode; - if (sadInit) (*sadInit) (); + if ((((struct global_all_dll *)Dll::Tls())->sadInit)) (*(((struct global_all_dll *)Dll::Tls())->sadInit)) (); for (y = 0; y < pParam->mb_height; y ++) { for (x = 0; x < pParam->mb_width; x ++) { @@ -210,8 +199,8 @@ && (abs(mv.y - (pMB+MBw)->mvs[0].y) < deltay) ) { const int iEdgedWidth = pParam->edged_width; const uint8_t *const pCur = current->image.y + 16*(my*iEdgedWidth + mx); - if ( (sad16 ( pCur, pCur+1 , iEdgedWidth, 65536) >= gradx ) - && (sad16 ( pCur, pCur+iEdgedWidth, iEdgedWidth, 65536) >= grady ) ) + if ( ((((struct global_all_dll *)Dll::Tls())->sad16) ( pCur, pCur+1 , iEdgedWidth, 65536) >= gradx ) + && ((((struct global_all_dll *)Dll::Tls())->sad16) ( pCur, pCur+iEdgedWidth, iEdgedWidth, 65536) >= grady ) ) { pMB->mcsel = 1; num++; } @@ -219,7 +208,7 @@ /* only use "structured" blocks */ } } - emms(); + (((struct global_all_dll *)Dll::Tls())->emms)(); /* further filtering would be possible, but during iteration, remaining outliers usually are removed, too */ @@ -330,8 +319,8 @@ const int mbnum = mx + my * MBw; MACROBLOCK *const pMB = &pMBs[mbnum]; const uint8_t *const pCur = current->image.y + 16*(my*iEdgedWidth + mx); - if ( (sad16 ( pCur, pCur+1 , iEdgedWidth, 65536) >= gradx ) - && (sad16 ( pCur, pCur+iEdgedWidth, iEdgedWidth, 65536) >= grady ) ) + if ( ((((struct global_all_dll *)Dll::Tls())->sad16) ( pCur, pCur+1 , iEdgedWidth, 65536) >= gradx ) + && ((((struct global_all_dll *)Dll::Tls())->sad16) ( pCur, pCur+iEdgedWidth, iEdgedWidth, 65536) >= grady ) ) { pMB->mcsel = 1; gmc.duv[0].x += pMB->mvs[0].x; gmc.duv[0].y += pMB->mvs[0].y; @@ -369,25 +358,6 @@ return gmc; } -int -GlobalMotionEstRefine( - WARPPOINTS *const startwp, - MACROBLOCK * const pMBs, - const MBParam * const pParam, - const FRAMEINFO * const current, - const FRAMEINFO * const reference, - const IMAGE * const pCurr, - const IMAGE * const pRef, - const IMAGE * const pRefH, - const IMAGE * const pRefV, - const IMAGE * const pRefHV) -{ - uint8_t* GMCblock = (uint8_t*)malloc(16*pParam->edged_width); - WARPPOINTS bestwp=*startwp; - WARPPOINTS centerwp,currwp; - int gmcminSAD=0; - int gmcSAD=0; - int direction; #if 0 int mx,my; #endif @@ -416,6 +384,27 @@ } } #endif +int +GlobalMotionEstRefine( + WARPPOINTS *const startwp, + MACROBLOCK * const pMBs, + const MBParam * const pParam, + const FRAMEINFO * const current, + const FRAMEINFO * const reference, + const IMAGE * const pCurr, + const IMAGE * const pRef, + const IMAGE * const pRefH, + const IMAGE * const pRefV, + const IMAGE * const pRefHV) +{ + uint8_t* GMCblock = (uint8_t*)malloc(16*pParam->edged_width); + WARPPOINTS bestwp=*startwp; + WARPPOINTS centerwp,currwp; + int gmcminSAD=0; + int gmcSAD=0; + int direction; + + gmcminSAD = globalSAD(&bestwp, pParam, pMBs, current, pRef, pCurr, GMCblock); @@ -610,7 +599,7 @@ mx, my, pParam->m_rounding_type); - iSAD = sad16 ( pCurr->y + 16*(my*iEdgedWidth + mx), + iSAD = (((struct global_all_dll *)Dll::Tls())->sad16) ( pCurr->y + 16*(my*iEdgedWidth + mx), GMCblock , iEdgedWidth, 65536); iSAD -= pMBs[mbnum].sad16; diff -bBru src-ori/motion/estimation_pvop.c src/motion/estimation_pvop.c --- src-ori/motion/estimation_pvop.c Wed Jan 12 10:32:43 2005 +++ src/motion/estimation_pvop.c Wed Jan 12 12:45:10 2005 @@ -25,20 +25,9 @@ * ****************************************************************************/ -#include -#include -#include -#include /* memcpy */ - -#include "../encoder.h" -#include "../prediction/mbprediction.h" -#include "../global.h" -#include "../utils/timer.h" -#include "../image/interpolate8x8.h" -#include "estimation.h" -#include "motion.h" -#include "sad.h" -#include "motion_inlines.h" +#include +#include "../global_all_dll.h" + static const int xvid_me_lambda_vec8[32] = { 0 ,(int)(1.00235 * NEIGH_TEND_8X8 + 0.5), @@ -70,7 +59,7 @@ Reference = GetReference(x, y, data); - sad = sad16v(data->Cur, Reference, data->iEdgedWidth, data->temp); + sad = (((struct global_all_dll *)Dll::Tls())->sad16v)(data->Cur, Reference, data->iEdgedWidth, data->temp); t = d_mv_bits(x, y, data->predMV, data->iFcode, data->qpel, 0); sad += (data->lambda16 * t * sad)>>10; @@ -110,7 +99,7 @@ Reference = xvid_me_interpolate16x16qpel(x, y, 0, data); - sad = sad16v(data->Cur, Reference, data->iEdgedWidth, data->temp); + sad = (((struct global_all_dll *)Dll::Tls())->sad16v)(data->Cur, Reference, data->iEdgedWidth, data->temp); t = d_mv_bits(x, y, data->predMV, data->iFcode, 0, 0); sad += (data->lambda16 * t * sad)>>10; @@ -160,7 +149,7 @@ current = data->currentQMV; } - sad = sad8(data->Cur, Reference, data->iEdgedWidth); + sad = (((struct global_all_dll *)Dll::Tls())->sad8)(data->Cur, Reference, data->iEdgedWidth); t = d_mv_bits(x, y, data->predMV, data->iFcode, data->qpel^data->qpel_precision, 0); sad += (data->lambda8 * t * (sad+NEIGH_8X8_BIAS))>>10; @@ -186,7 +175,7 @@ Reference = xvid_me_interpolate8x8qpel(x, y, 0, 0, data); current = data->currentQMV; - sad = sad8(data->Cur, Reference, data->iEdgedWidth); + sad = (((struct global_all_dll *)Dll::Tls())->sad8)(data->Cur, Reference, data->iEdgedWidth); t = d_mv_bits(x, y, data->predMV, data->iFcode, data->qpel^data->qpel_precision, 0); sad += (data->lambda8 * t * (sad+NEIGH_8X8_BIAS))>>10; @@ -247,19 +236,19 @@ { int offset = (x + y*stride)*8; if(!rrv) { - uint32_t sadC = sad8(current->u + offset, + uint32_t sadC = (((struct global_all_dll *)Dll::Tls())->sad8)(current->u + offset, reference->u + offset, stride); if (sadC > iQuant * MAX_CHROMA_SAD_FOR_SKIP) return 0; - sadC += sad8(current->v + offset, + sadC += (((struct global_all_dll *)Dll::Tls())->sad8)(current->v + offset, reference->v + offset, stride); if (sadC > iQuant * MAX_CHROMA_SAD_FOR_SKIP) return 0; return 1; } else { - uint32_t sadC = sad16(current->u + 2*offset, + uint32_t sadC = (((struct global_all_dll *)Dll::Tls())->sad16)(current->u + 2*offset, reference->u + 2*offset, stride, 256*4096); if (sadC > iQuant * MAX_CHROMA_SAD_FOR_SKIP*4) return 0; - sadC += sad16(current->v + 2*offset, + sadC += (((struct global_all_dll *)Dll::Tls())->sad16)(current->v + 2*offset, reference->v + 2*offset, stride, 256*4096); if (sadC > iQuant * MAX_CHROMA_SAD_FOR_SKIP*4) return 0; return 1; @@ -415,12 +404,12 @@ /* mcsel */ if (coding_type == S_VOP) { - int32_t iSAD = sad16(Data->Cur, + int32_t iSAD = (((struct global_all_dll *)Dll::Tls())->sad16)(Data->Cur, vGMC->y + 16*y*Data->iEdgedWidth + 16*x, Data->iEdgedWidth, 65536); if (Data->chroma) { - iSAD += sad8(Data->CurU, vGMC->u + 8*y*(Data->iEdgedWidth/2) + 8*x, Data->iEdgedWidth/2); - iSAD += sad8(Data->CurV, vGMC->v + 8*y*(Data->iEdgedWidth/2) + 8*x, Data->iEdgedWidth/2); + iSAD += (((struct global_all_dll *)Dll::Tls())->sad8)(Data->CurU, vGMC->u + 8*y*(Data->iEdgedWidth/2) + 8*x, Data->iEdgedWidth/2); + iSAD += (((struct global_all_dll *)Dll::Tls())->sad8)(Data->CurV, vGMC->v + 8*y*(Data->iEdgedWidth/2) + 8*x, Data->iEdgedWidth/2); } if (iSAD <= sad) { /* mode decision GMC */ @@ -444,12 +433,12 @@ if (InterBias < sad) { int32_t deviation; if (!Data->rrv) - deviation = dev16(Data->Cur, Data->iEdgedWidth); + deviation = (((struct global_all_dll *)Dll::Tls())->dev16)(Data->Cur, Data->iEdgedWidth); else - deviation = dev16(Data->Cur, Data->iEdgedWidth) + /* dev32() */ - dev16(Data->Cur+16, Data->iEdgedWidth) + - dev16(Data->Cur + 16*Data->iEdgedWidth, Data->iEdgedWidth) + - dev16(Data->Cur+16+16*Data->iEdgedWidth, Data->iEdgedWidth); + deviation = (((struct global_all_dll *)Dll::Tls())->dev16)(Data->Cur, Data->iEdgedWidth) + /* dev32() */ + (((struct global_all_dll *)Dll::Tls())->dev16)(Data->Cur+16, Data->iEdgedWidth) + + (((struct global_all_dll *)Dll::Tls())->dev16)(Data->Cur + 16*Data->iEdgedWidth, Data->iEdgedWidth) + + (((struct global_all_dll *)Dll::Tls())->dev16)(Data->Cur+16+16*Data->iEdgedWidth, Data->iEdgedWidth); if (deviation < (sad - InterBias)) mode = MODE_INTRA; } @@ -918,7 +908,7 @@ } Data.RefQ = pRefV->u; /* a good place, also used in MC (for similar purpose) */ - if (sadInit) (*sadInit) (); + if ((((struct global_all_dll *)Dll::Tls())->sadInit)) (*(((struct global_all_dll *)Dll::Tls())->sadInit)) (); for (y = 0; y < mb_height; y++) { for (x = 0; x < mb_width; x++) { @@ -926,7 +916,7 @@ MACROBLOCK *prevMB = &reference->mbs[x + y * pParam->mb_width]; if (!Data.rrv) pMB->sad16 = - sad16v(pCurrent->y + (x + y * iEdgedWidth) * 16, + (((struct global_all_dll *)Dll::Tls())->sad16v)(pCurrent->y + (x + y * iEdgedWidth) * 16, pRef->y + (x + y * iEdgedWidth) * 16, pParam->edged_width, pMB->sad8 ); @@ -936,9 +926,9 @@ pParam->edged_width, pMB->sad8 ); if (Data.chroma) { - Data.chromaSAD = sad8(pCurrent->u + x*8 + y*(iEdgedWidth/2)*8, + Data.chromaSAD = (((struct global_all_dll *)Dll::Tls())->sad8)(pCurrent->u + x*8 + y*(iEdgedWidth/2)*8, pRef->u + x*8 + y*(iEdgedWidth/2)*8, iEdgedWidth/2) - + sad8(pCurrent->v + (x + y*(iEdgedWidth/2))*8, + + (((struct global_all_dll *)Dll::Tls())->sad8)(pCurrent->v + (x + y*(iEdgedWidth/2))*8, pRef->v + (x + y*(iEdgedWidth/2))*8, iEdgedWidth/2); pMB->sad16 += Data.chromaSAD; } diff -bBru src-ori/motion/estimation_rd_based.c src/motion/estimation_rd_based.c --- src-ori/motion/estimation_rd_based.c Wed Jan 12 10:32:43 2005 +++ src/motion/estimation_rd_based.c Wed Jan 12 12:43:52 2005 @@ -26,24 +26,9 @@ /* RD mode decision and search */ -#include -#include -#include -#include /* memcpy */ - -#include "../encoder.h" -#include "../bitstream/mbcoding.h" -#include "../prediction/mbprediction.h" -#include "../global.h" -#include "../image/interpolate8x8.h" -#include "estimation.h" -#include "motion.h" -#include "sad.h" -#include "../bitstream/zigzag.h" -#include "../quant/quant.h" -#include "../bitstream/vlc_codes.h" -#include "../dct/fdct.h" -#include "motion_inlines.h" +#include +#include "../global_all_dll.h" + /* rd = BITS_MULT*bits + LAMBDA*distortion */ #define LAMBDA ( (int)(BITS_MULT*1.0) ) @@ -62,19 +47,19 @@ int bits; int distortion = 0; - fdct(data); + (((struct global_all_dll *)Dll::Tls())->fdct)(data); - if (quant_type) sum = quant_h263_inter(coeff, data, quant, mpeg_quant_matrices); - else sum = quant_mpeg_inter(coeff, data, quant, mpeg_quant_matrices); + if (quant_type) sum = (((struct global_all_dll *)Dll::Tls())->quant_h263_inter)(coeff, data, quant, mpeg_quant_matrices); + else sum = (((struct global_all_dll *)Dll::Tls())->quant_mpeg_inter)(coeff, data, quant, mpeg_quant_matrices); if (sum > 0) { *cbp |= 1 << (5 - block); bits = BITS_MULT * CodeCoeffInter_CalcBits(coeff, scan_table); - if (quant_type) dequant_h263_inter(dqcoeff, coeff, quant, mpeg_quant_matrices); - else dequant_mpeg_inter(dqcoeff, coeff, quant, mpeg_quant_matrices); + if (quant_type) (((struct global_all_dll *)Dll::Tls())->dequant_h263_inter)(dqcoeff, coeff, quant, mpeg_quant_matrices); + else (((struct global_all_dll *)Dll::Tls())->dequant_mpeg_inter)(dqcoeff, coeff, quant, mpeg_quant_matrices); - distortion = sse8_16bit(data, dqcoeff, 8*sizeof(int16_t)); + distortion = (((struct global_all_dll *)Dll::Tls())->sse8_16bit)(data, dqcoeff, 8*sizeof(int16_t)); } else { const static int16_t zero_block[64] = { @@ -88,7 +73,7 @@ 0, 0, 0, 0, 0, 0, 0, 0, }; bits = 0; - distortion = sse8_16bit(data, zero_block, 8*sizeof(int16_t)); + distortion = (((struct global_all_dll *)Dll::Tls())->sse8_16bit)(data, zero_block, 8*sizeof(int16_t)); } @@ -117,14 +102,14 @@ unsigned int distortion = 0; const uint32_t iDcScaler = get_dc_scaler(quant, block < 4); - fdct(coeff); + (((struct global_all_dll *)Dll::Tls())->fdct)(coeff); if (quant_type) { - quant_h263_intra(qcoeff, coeff, quant, iDcScaler, mpeg_quant_matrices); - dequant_h263_intra(dqcoeff, qcoeff, quant, iDcScaler, mpeg_quant_matrices); + (((struct global_all_dll *)Dll::Tls())->quant_h263_intra)(qcoeff, coeff, quant, iDcScaler, mpeg_quant_matrices); + (((struct global_all_dll *)Dll::Tls())->dequant_h263_intra)(dqcoeff, qcoeff, quant, iDcScaler, mpeg_quant_matrices); } else { - quant_mpeg_intra(qcoeff, coeff, quant, iDcScaler, mpeg_quant_matrices); - dequant_mpeg_intra(dqcoeff, qcoeff, quant, iDcScaler, mpeg_quant_matrices); + (((struct global_all_dll *)Dll::Tls())->quant_mpeg_intra)(qcoeff, coeff, quant, iDcScaler, mpeg_quant_matrices); + (((struct global_all_dll *)Dll::Tls())->dequant_mpeg_intra)(dqcoeff, qcoeff, quant, iDcScaler, mpeg_quant_matrices); } predict_acdc(pMB-(x+mb_width*y), x, y, mb_width, block, qcoeff, @@ -167,7 +152,7 @@ bits[1] += coded = CodeCoeffIntra_CalcBits(qcoeff, scan_tables[direction]); if (coded > 0) cbp[1] |= 1 << (5 - block); - distortion = sse8_16bit(coeff, dqcoeff, 8*sizeof(int16_t)); + distortion = (((struct global_all_dll *)Dll::Tls())->sse8_16bit)(coeff, dqcoeff, 8*sizeof(int16_t)); return (LAMBDA*distortion)/(quant*quant); } @@ -200,7 +185,7 @@ for(i = 0; i < 4; i++) { int s = 8*((i&1) + (i>>1)*data->iEdgedWidth); - transfer_8to16subro(in, data->Cur + s, ptr + s, data->iEdgedWidth); + (((struct global_all_dll *)Dll::Tls())->transfer_8to16subro)(in, data->Cur + s, ptr + s, data->iEdgedWidth); rd += data->temp[i] = Block_CalcBits(coeff, in, data->dctSpace + 128, data->iQuant, data->quant_type, &cbp, i, data->scan_table, data->mpeg_quant_matrices); } @@ -225,16 +210,16 @@ /* chroma U */ ptr = interpolate8x8_switch2(data->RefQ, data->RefP[4], 0, 0, xc, yc, data->iEdgedWidth/2, data->rounding); - transfer_8to16subro(in, data->CurU, ptr, data->iEdgedWidth/2); + (((struct global_all_dll *)Dll::Tls())->transfer_8to16subro)(in, data->CurU, ptr, data->iEdgedWidth/2); rd += Block_CalcBits(coeff, in, data->dctSpace + 128, data->iQuant, data->quant_type, &cbp, 4, data->scan_table, data->mpeg_quant_matrices); if (rd >= data->iMinSAD[0]) return; /* chroma V */ ptr = interpolate8x8_switch2(data->RefQ, data->RefP[5], 0, 0, xc, yc, data->iEdgedWidth/2, data->rounding); - transfer_8to16subro(in, data->CurV, ptr, data->iEdgedWidth/2); + (((struct global_all_dll *)Dll::Tls())->transfer_8to16subro)(in, data->CurV, ptr, data->iEdgedWidth/2); rd += Block_CalcBits(coeff, in, data->dctSpace + 128, data->iQuant, data->quant_type, &cbp, 5, data->scan_table, data->mpeg_quant_matrices); - rd += BITS_MULT*mcbpc_inter_tab[(MODE_INTER & 7) | ((cbp & 3) << 3)].len; + rd += BITS_MULT*(((struct global_all_dll *)Dll::Tls())->mcbpc_inter_tab)[(MODE_INTER & 7) | ((cbp & 3) << 3)].len; if (rd < data->iMinSAD[0]) { data->iMinSAD[0] = rd; @@ -265,7 +250,7 @@ current = data->currentQMV; } - transfer_8to16subro(in, data->Cur, ptr, data->iEdgedWidth); + (((struct global_all_dll *)Dll::Tls())->transfer_8to16subro)(in, data->Cur, ptr, data->iEdgedWidth); rd = Block_CalcBits(coeff, in, data->dctSpace + 128, data->iQuant, data->quant_type, &cbp, 5, data->scan_table, data->mpeg_quant_matrices); rd += BITS_MULT*d_mv_bits(x, y, data->predMV, data->iFcode, data->qpel^data->qpel_precision, 0); @@ -467,17 +452,17 @@ /* chroma U */ ptr = interpolate8x8_switch2(Data->RefQ + 64, Data->RefP[4], 0, 0, sumx, sumy, Data->iEdgedWidth/2, Data->rounding); - transfer_8to16subro(in, Data->CurU, ptr, Data->iEdgedWidth/2); + (((struct global_all_dll *)Dll::Tls())->transfer_8to16subro)(in, Data->CurU, ptr, Data->iEdgedWidth/2); bits += Block_CalcBits(coeff, in, Data->dctSpace + 128, Data->iQuant, Data->quant_type, &cbp, 4, Data->scan_table, Data->mpeg_quant_matrices); if (bits >= *Data->iMinSAD) return bits; /* chroma V */ ptr = interpolate8x8_switch2(Data->RefQ + 64, Data->RefP[5], 0, 0, sumx, sumy, Data->iEdgedWidth/2, Data->rounding); - transfer_8to16subro(in, Data->CurV, ptr, Data->iEdgedWidth/2); + (((struct global_all_dll *)Dll::Tls())->transfer_8to16subro)(in, Data->CurV, ptr, Data->iEdgedWidth/2); bits += Block_CalcBits(coeff, in, Data->dctSpace + 128, Data->iQuant, Data->quant_type, &cbp, 5, Data->scan_table, Data->mpeg_quant_matrices); - bits += BITS_MULT*mcbpc_inter_tab[(MODE_INTER4V & 7) | ((cbp & 3) << 3)].len; + bits += BITS_MULT*(((struct global_all_dll *)Dll::Tls())->mcbpc_inter_tab)[(MODE_INTER4V & 7) | ((cbp & 3) << 3)].len; *Data->cbp = cbp; return bits; @@ -497,7 +482,7 @@ for(i = 0; i < 4; i++) { int s = 8*((i&1) + (i>>1)*Data->iEdgedWidth); - transfer_8to16copy(in, Data->Cur + s, Data->iEdgedWidth); + (((struct global_all_dll *)Dll::Tls())->transfer_8to16copy)(in, Data->Cur + s, Data->iEdgedWidth); distortion = Block_CalcBitsIntra(pMB, x, y, mb_width, i, in, coeff, dqcoeff, @@ -513,7 +498,7 @@ bits2 += BITS_MULT*xvid_cbpy_tab[cbp[1]>>2].len; /*chroma U */ - transfer_8to16copy(in, Data->CurU, Data->iEdgedWidth/2); + (((struct global_all_dll *)Dll::Tls())->transfer_8to16copy)(in, Data->CurU, Data->iEdgedWidth/2); distortion = Block_CalcBitsIntra(pMB, x, y, mb_width, 4, in, coeff, dqcoeff, predictors[4], iQuant, Data->quant_type, bits, cbp, Data->mpeg_quant_matrices); bits1 += distortion + BITS_MULT * bits[0]; @@ -523,15 +508,15 @@ return bits1; /* chroma V */ - transfer_8to16copy(in, Data->CurV, Data->iEdgedWidth/2); + (((struct global_all_dll *)Dll::Tls())->transfer_8to16copy)(in, Data->CurV, Data->iEdgedWidth/2); distortion = Block_CalcBitsIntra(pMB, x, y, mb_width, 5, in, coeff, dqcoeff, predictors[5], iQuant, Data->quant_type, bits, cbp, Data->mpeg_quant_matrices); bits1 += distortion + BITS_MULT * bits[0]; bits2 += distortion + BITS_MULT * bits[1]; - bits1 += BITS_MULT*mcbpc_inter_tab[(MODE_INTRA & 7) | ((cbp[0] & 3) << 3)].len; - bits2 += BITS_MULT*mcbpc_inter_tab[(MODE_INTRA & 7) | ((cbp[1] & 3) << 3)].len; + bits1 += BITS_MULT*(((struct global_all_dll *)Dll::Tls())->mcbpc_inter_tab)[(MODE_INTRA & 7) | ((cbp[0] & 3) << 3)].len; + bits2 += BITS_MULT*(((struct global_all_dll *)Dll::Tls())->mcbpc_inter_tab)[(MODE_INTRA & 7) | ((cbp[1] & 3) << 3)].len; *Data->cbp = bits1 <= bits2 ? cbp[0] : cbp[1]; @@ -548,7 +533,7 @@ for(i = 0; i < 4; i++) { int s = 8*((i&1) + (i>>1)*Data->iEdgedWidth); - transfer_8to16subro(in, Data->Cur + s, vGMC->y + s + 16*(x+y*Data->iEdgedWidth), Data->iEdgedWidth); + (((struct global_all_dll *)Dll::Tls())->transfer_8to16subro)(in, Data->Cur + s, vGMC->y + s + 16*(x+y*Data->iEdgedWidth), Data->iEdgedWidth); bits += Block_CalcBits(coeff, in, Data->dctSpace + 128, Data->iQuant, Data->quant_type, &cbp, i, Data->scan_table, Data->mpeg_quant_matrices); if (bits >= Data->iMinSAD[0]) return bits; } @@ -556,16 +541,16 @@ bits += BITS_MULT*xvid_cbpy_tab[15-(cbp>>2)].len; /*chroma U */ - transfer_8to16subro(in, Data->CurU, vGMC->u + 8*(x+y*(Data->iEdgedWidth/2)), Data->iEdgedWidth/2); + (((struct global_all_dll *)Dll::Tls())->transfer_8to16subro)(in, Data->CurU, vGMC->u + 8*(x+y*(Data->iEdgedWidth/2)), Data->iEdgedWidth/2); bits += Block_CalcBits(coeff, in, Data->dctSpace + 128, Data->iQuant, Data->quant_type, &cbp, 4, Data->scan_table, Data->mpeg_quant_matrices); if (bits >= Data->iMinSAD[0]) return bits; /* chroma V */ - transfer_8to16subro(in, Data->CurV , vGMC->v + 8*(x+y*(Data->iEdgedWidth/2)), Data->iEdgedWidth/2); + (((struct global_all_dll *)Dll::Tls())->transfer_8to16subro)(in, Data->CurV , vGMC->v + 8*(x+y*(Data->iEdgedWidth/2)), Data->iEdgedWidth/2); bits += Block_CalcBits(coeff, in, Data->dctSpace + 128, Data->iQuant, Data->quant_type, &cbp, 5, Data->scan_table, Data->mpeg_quant_matrices); - bits += BITS_MULT*mcbpc_inter_tab[(MODE_INTER & 7) | ((cbp & 3) << 3)].len; + bits += BITS_MULT*(((struct global_all_dll *)Dll::Tls())->mcbpc_inter_tab)[(MODE_INTER & 7) | ((cbp & 3) << 3)].len; *Data->cbp = cbp; @@ -734,12 +719,12 @@ /* mcsel */ if (coding_type == S_VOP) { - int32_t iSAD = sad16(Data->Cur, + int32_t iSAD = (((struct global_all_dll *)Dll::Tls())->sad16)(Data->Cur, vGMC->y + 16*y*Data->iEdgedWidth + 16*x, Data->iEdgedWidth, 65536); if (Data->chroma) { - iSAD += sad8(Data->CurU, vGMC->u + 8*y*(Data->iEdgedWidth/2) + 8*x, Data->iEdgedWidth/2); - iSAD += sad8(Data->CurV, vGMC->v + 8*y*(Data->iEdgedWidth/2) + 8*x, Data->iEdgedWidth/2); + iSAD += (((struct global_all_dll *)Dll::Tls())->sad8)(Data->CurU, vGMC->u + 8*y*(Data->iEdgedWidth/2) + 8*x, Data->iEdgedWidth/2); + iSAD += (((struct global_all_dll *)Dll::Tls())->sad8)(Data->CurV, vGMC->v + 8*y*(Data->iEdgedWidth/2) + 8*x, Data->iEdgedWidth/2); } if (iSAD <= sad) { /* mode decision GMC */ @@ -782,11 +767,11 @@ *Data->iMinSAD = min_rd = gmc_rd; mode = MODE_INTER; cbp = *Data->cbp; - sad = sad16(Data->Cur, + sad = (((struct global_all_dll *)Dll::Tls())->sad16)(Data->Cur, vGMC->y + 16*y*Data->iEdgedWidth + 16*x, Data->iEdgedWidth, 65536); if (Data->chroma) { - sad += sad8(Data->CurU, vGMC->u + 8*y*(Data->iEdgedWidth/2) + 8*x, Data->iEdgedWidth/2); - sad += sad8(Data->CurV, vGMC->v + 8*y*(Data->iEdgedWidth/2) + 8*x, Data->iEdgedWidth/2); + sad += (((struct global_all_dll *)Dll::Tls())->sad8)(Data->CurU, vGMC->u + 8*y*(Data->iEdgedWidth/2) + 8*x, Data->iEdgedWidth/2); + sad += (((struct global_all_dll *)Dll::Tls())->sad8)(Data->CurV, vGMC->v + 8*y*(Data->iEdgedWidth/2) + 8*x, Data->iEdgedWidth/2); } } } @@ -837,12 +822,12 @@ if (InterBias < sad) { int32_t deviation; if (!Data->rrv) - deviation = dev16(Data->Cur, Data->iEdgedWidth); + deviation = (((struct global_all_dll *)Dll::Tls())->dev16)(Data->Cur, Data->iEdgedWidth); else - deviation = dev16(Data->Cur, Data->iEdgedWidth) + /* dev32() */ - dev16(Data->Cur+16, Data->iEdgedWidth) + - dev16(Data->Cur + 16*Data->iEdgedWidth, Data->iEdgedWidth) + - dev16(Data->Cur+16+16*Data->iEdgedWidth, Data->iEdgedWidth); + deviation = (((struct global_all_dll *)Dll::Tls())->dev16)(Data->Cur, Data->iEdgedWidth) + /* dev32() */ + (((struct global_all_dll *)Dll::Tls())->dev16)(Data->Cur+16, Data->iEdgedWidth) + + (((struct global_all_dll *)Dll::Tls())->dev16)(Data->Cur + 16*Data->iEdgedWidth, Data->iEdgedWidth) + + (((struct global_all_dll *)Dll::Tls())->dev16)(Data->Cur+16+16*Data->iEdgedWidth, Data->iEdgedWidth); if (deviation < (sad - InterBias)) mode = MODE_INTRA; } diff -bBru src-ori/motion/gmc.c src/motion/gmc.c --- src-ori/motion/gmc.c Wed Jan 12 10:32:43 2005 +++ src/motion/gmc.c Wed Jan 12 12:43:24 2005 @@ -23,12 +23,9 @@ * ****************************************************************************/ -#include "../portab.h" -#include "../global.h" -#include "../encoder.h" -#include "gmc.h" +#include +#include "../global_all_dll.h" -#include /* ************************************************************ * Pts = 2 or 3 Only in src-ori/motion: ia64_asm diff -bBru src-ori/motion/motion_comp.c src/motion/motion_comp.c --- src-ori/motion/motion_comp.c Wed Jan 12 11:40:18 2005 +++ src/motion/motion_comp.c Wed Jan 12 12:44:05 2005 @@ -24,15 +24,9 @@ * ****************************************************************************/ -#include +#include +#include "../global_all_dll.h" -#include "../encoder.h" -#include "../utils/mbfunctions.h" -#include "../image/interpolate8x8.h" -#include "../image/qpel.h" -#include "../image/reduced.h" -#include "../utils/timer.h" -#include "motion.h" #ifndef RSHIFT #define RSHIFT(a,b) ((a) > 0 ? ((a) + (1<<((b)-1)))>>(b) : ((a) + (1<<((b)-1))-1)>>(b)) @@ -133,13 +127,13 @@ } else ptr = get_ref(ref, refh, refv, refhv, x, y, 1, dx, dy, stride); - transfer_8to16sub(dct_codes, cur + y * stride + x, + (((struct global_all_dll *)Dll::Tls())->transfer_8to16sub)(dct_codes, cur + y * stride + x, ptr, stride); - transfer_8to16sub(dct_codes+64, cur + y * stride + x + 8, + (((struct global_all_dll *)Dll::Tls())->transfer_8to16sub)(dct_codes+64, cur + y * stride + x + 8, ptr + 8, stride); - transfer_8to16sub(dct_codes+128, cur + y * stride + x + 8*stride, + (((struct global_all_dll *)Dll::Tls())->transfer_8to16sub)(dct_codes+128, cur + y * stride + x + 8*stride, ptr + 8*stride, stride); - transfer_8to16sub(dct_codes+192, cur + y * stride + x + 8*stride+8, + (((struct global_all_dll *)Dll::Tls())->transfer_8to16sub)(dct_codes+192, cur + y * stride + x + 8*stride+8, ptr + 8*stride + 8, stride); } else { /* reduced_resolution */ @@ -148,17 +142,17 @@ ptr = get_ref(ref, refh, refv, refhv, x, y, 1, dx, dy, stride); - filter_18x18_to_8x8(dct_codes, cur+y*stride + x, stride); - filter_diff_18x18_to_8x8(dct_codes, ptr, stride); + (((struct global_all_dll *)Dll::Tls())->filter_18x18_to_8x8)(dct_codes, cur+y*stride + x, stride); + (((struct global_all_dll *)Dll::Tls())->filter_diff_18x18_to_8x8)(dct_codes, ptr, stride); - filter_18x18_to_8x8(dct_codes+64, cur+y*stride + x + 16, stride); - filter_diff_18x18_to_8x8(dct_codes+64, ptr + 16, stride); + (((struct global_all_dll *)Dll::Tls())->filter_18x18_to_8x8)(dct_codes+64, cur+y*stride + x + 16, stride); + (((struct global_all_dll *)Dll::Tls())->filter_diff_18x18_to_8x8)(dct_codes+64, ptr + 16, stride); - filter_18x18_to_8x8(dct_codes+128, cur+(y+16)*stride + x, stride); - filter_diff_18x18_to_8x8(dct_codes+128, ptr + 16*stride, stride); + (((struct global_all_dll *)Dll::Tls())->filter_18x18_to_8x8)(dct_codes+128, cur+(y+16)*stride + x, stride); + (((struct global_all_dll *)Dll::Tls())->filter_diff_18x18_to_8x8)(dct_codes+128, ptr + 16*stride, stride); - filter_18x18_to_8x8(dct_codes+192, cur+(y+16)*stride + x + 16, stride); - filter_diff_18x18_to_8x8(dct_codes+192, ptr + 16*stride + 16, stride); + (((struct global_all_dll *)Dll::Tls())->filter_18x18_to_8x8)(dct_codes+192, cur+(y+16)*stride + x + 16, stride); + (((struct global_all_dll *)Dll::Tls())->filter_diff_18x18_to_8x8)(dct_codes+192, ptr + 16*stride + 16, stride); transfer32x32_copy(cur + y*stride + x, ptr, stride); } @@ -200,7 +194,7 @@ } else ptr = ref + ((int)y + dy/4)*(int)stride + (int)x + dx/4; /* fullpixel position */ } else ptr = get_ref(ref, refh, refv, refhv, x, y, 1, dx, dy, stride); - transfer_8to16sub(dct_codes, cur + y * stride + x, ptr, stride); + (((struct global_all_dll *)Dll::Tls())->transfer_8to16sub)(dct_codes, cur + y * stride + x, ptr, stride); } else { /* reduced_resolution */ @@ -208,8 +202,8 @@ ptr = get_ref(ref, refh, refv, refhv, x, y, 1, dx, dy, stride); - filter_18x18_to_8x8(dct_codes, cur+y*stride + x, stride); - filter_diff_18x18_to_8x8(dct_codes, ptr, stride); + (((struct global_all_dll *)Dll::Tls())->filter_18x18_to_8x8)(dct_codes, cur+y*stride + x, stride); + (((struct global_all_dll *)Dll::Tls())->filter_diff_18x18_to_8x8)(dct_codes, ptr, stride); transfer16x16_copy(cur + y*stride + x, ptr, stride); } @@ -252,11 +246,11 @@ { /* uv-block-based compensation */ if (!rrv) { - transfer_8to16sub(coeff, Cur->u + 8 * j * stride + 8 * i, + (((struct global_all_dll *)Dll::Tls())->transfer_8to16sub)(coeff, Cur->u + 8 * j * stride + 8 * i, interpolate8x8_switch2(temp, Ref->u, 8 * i, 8 * j, dx, dy, stride, rounding), stride); - transfer_8to16sub(coeff + 64, Cur->v + 8 * j * stride + 8 * i, + (((struct global_all_dll *)Dll::Tls())->transfer_8to16sub)(coeff + 64, Cur->v + 8 * j * stride + 8 * i, interpolate8x8_switch2(temp, Ref->v, 8 * i, 8 * j, dx, dy, stride, rounding), stride); @@ -266,14 +260,14 @@ current = Cur->u + 16*j*stride + 16*i; reference = temp - 16*j*stride - 16*i; interpolate18x18_switch(reference, Ref->u, 16*i, 16*j, dx, dy, stride, rounding); - filter_18x18_to_8x8(coeff, current, stride); - filter_diff_18x18_to_8x8(coeff, temp, stride); + (((struct global_all_dll *)Dll::Tls())->filter_18x18_to_8x8)(coeff, current, stride); + (((struct global_all_dll *)Dll::Tls())->filter_diff_18x18_to_8x8)(coeff, temp, stride); transfer16x16_copy(current, temp, stride); current = Cur->v + 16*j*stride + 16*i; interpolate18x18_switch(reference, Ref->v, 16*i, 16*j, dx, dy, stride, rounding); - filter_18x18_to_8x8(coeff + 64, current, stride); - filter_diff_18x18_to_8x8(coeff + 64, temp, stride); + (((struct global_all_dll *)Dll::Tls())->filter_18x18_to_8x8)(coeff + 64, current, stride); + (((struct global_all_dll *)Dll::Tls())->filter_diff_18x18_to_8x8)(coeff + 64, temp, stride); transfer16x16_copy(current, temp, stride); } } @@ -308,10 +302,10 @@ ref->y + 16 * (i + j * edged_width), edged_width); - transfer8x8_copy(cur->u + 8 * (i + j * edged_width/2), + (((struct global_all_dll *)Dll::Tls())->transfer8x8_copy)(cur->u + 8 * (i + j * edged_width/2), ref->u + 8 * (i + j * edged_width/2), edged_width / 2); - transfer8x8_copy(cur->v + 8 * (i + j * edged_width/2), + (((struct global_all_dll *)Dll::Tls())->transfer8x8_copy)(cur->v + 8 * (i + j * edged_width/2), ref->v + 8 * (i + j * edged_width/2), edged_width / 2); return; @@ -326,21 +320,21 @@ /* call normal routine once, easier than "if (mcsel)"ing all the time */ - transfer_8to16sub(&dct_codes[0*64], cur->y + 16*j*edged_width + 16*i, + (((struct global_all_dll *)Dll::Tls())->transfer_8to16sub)(&dct_codes[0*64], cur->y + 16*j*edged_width + 16*i, refGMC->y + 16*j*edged_width + 16*i, edged_width); - transfer_8to16sub(&dct_codes[1*64], cur->y + 16*j*edged_width + 16*i+8, + (((struct global_all_dll *)Dll::Tls())->transfer_8to16sub)(&dct_codes[1*64], cur->y + 16*j*edged_width + 16*i+8, refGMC->y + 16*j*edged_width + 16*i+8, edged_width); - transfer_8to16sub(&dct_codes[2*64], cur->y + (16*j+8)*edged_width + 16*i, + (((struct global_all_dll *)Dll::Tls())->transfer_8to16sub)(&dct_codes[2*64], cur->y + (16*j+8)*edged_width + 16*i, refGMC->y + (16*j+8)*edged_width + 16*i, edged_width); - transfer_8to16sub(&dct_codes[3*64], cur->y + (16*j+8)*edged_width + 16*i+8, + (((struct global_all_dll *)Dll::Tls())->transfer_8to16sub)(&dct_codes[3*64], cur->y + (16*j+8)*edged_width + 16*i+8, refGMC->y + (16*j+8)*edged_width + 16*i+8, edged_width); /* lumi is needed earlier for mode decision, but chroma should be done block-based, but it isn't, yet. */ - transfer_8to16sub(&dct_codes[4 * 64], cur->u + 8 *j*edged_width/2 + 8*i, + (((struct global_all_dll *)Dll::Tls())->transfer_8to16sub)(&dct_codes[4 * 64], cur->u + 8 *j*edged_width/2 + 8*i, refGMC->u + 8 *j*edged_width/2 + 8*i, edged_width/2); - transfer_8to16sub(&dct_codes[5 * 64], cur->v + 8*j* edged_width/2 + 8*i, + (((struct global_all_dll *)Dll::Tls())->transfer_8to16sub)(&dct_codes[5 * 64], cur->v + 8*j* edged_width/2 + 8*i, refGMC->v + 8*j* edged_width/2 + 8*i, edged_width/2); return; @@ -485,7 +479,7 @@ i, j, 16, b_dx, b_dy, edged_width); } for (k = 0; k < 4; k++) - transfer_8to16sub2(&dct_codes[k * 64], + (((struct global_all_dll *)Dll::Tls())->transfer_8to16sub2)(&dct_codes[k * 64], cur->y + (i * 16+(k&1)*8) + (j * 16+((k>>1)*8)) * edged_width, ptr1 + (k&1)*8 + (k>>1)*8*edged_width, ptr2 + (k&1)*8 + (k>>1)*8*edged_width, edged_width); @@ -535,7 +529,7 @@ ptr2 = get_ref(b_ref->y, b_refh->y, b_refv->y, b_refhv->y, 2*i + (k&1), 2*j + (k>>1), 8, b_dx, b_dy, edged_width); } - transfer_8to16sub2(&dct_codes[k * 64], + (((struct global_all_dll *)Dll::Tls())->transfer_8to16sub2)(&dct_codes[k * 64], cur->y + (i * 16+(k&1)*8) + (j * 16+((k>>1)*8)) * edged_width, ptr1, ptr2, edged_width); @@ -550,7 +544,7 @@ } /* v block-based chroma interpolation for direct and interpolate modes */ - transfer_8to16sub2(&dct_codes[4 * 64], + (((struct global_all_dll *)Dll::Tls())->transfer_8to16sub2)(&dct_codes[4 * 64], cur->u + (j * 8) * edged_width / 2 + (i * 8), interpolate8x8_switch2(tmp, b_ref->u, 8 * i, 8 * j, b_dx, b_dy, edged_width / 2, 0), @@ -558,7 +552,7 @@ dx, dy, edged_width / 2, 0), edged_width / 2); - transfer_8to16sub2(&dct_codes[5 * 64], + (((struct global_all_dll *)Dll::Tls())->transfer_8to16sub2)(&dct_codes[5 * 64], cur->v + (j * 8) * edged_width / 2 + (i * 8), interpolate8x8_switch2(tmp, b_ref->v, 8 * i, 8 * j, b_dx, b_dy, edged_width / 2, 0), diff -bBru src-ori/motion/motion_inlines.h src/motion/motion_inlines.h --- src-ori/motion/motion_inlines.h Wed Jan 12 10:32:43 2005 +++ src/motion/motion_inlines.h Wed Jan 12 12:38:15 2005 @@ -34,8 +34,7 @@ * Calculate the min/max range * relative to the _MACROBLOCK_ position */ -static void __inline -get_range(int32_t * const min_dx, +static void __inline get_range(int32_t * const min_dx, int32_t * const max_dx, int32_t * const min_dy, int32_t * const max_dy, @@ -47,7 +46,8 @@ const uint32_t fcode, const int precision, /* 2 for qpel, 1 for halfpel */ const int rrv) -{ + { + int k; const int search_range = 16 << fcode; int high = search_range - 1; @@ -82,9 +82,9 @@ 7, 7, 7, 6, 4, 3, 2, 1 }; -static __inline uint32_t -d_mv_bits(int x, int y, const VECTOR pred, const uint32_t iFcode, const int qpel, const int rrv) -{ +static __inline uint32_t d_mv_bits(int x, int y, const VECTOR pred, const uint32_t iFcode, const int qpel, const int rrv) + { + unsigned int bits; x <<= qpel; @@ -106,17 +106,17 @@ return bits; } -static __inline const uint8_t * -GetReference(const int x, const int y, const SearchData * const data) -{ +static __inline const uint8_t * GetReference(const int x, const int y, const SearchData * const data) + { + const int picture = ((x&1)<<1) | (y&1); const int offset = (x>>1) + (y>>1)*data->iEdgedWidth; return data->RefP[picture] + offset; } -static __inline const uint8_t * -GetReferenceB(const int x, const int y, const uint32_t dir, const SearchData * const data) -{ +static __inline const uint8_t * GetReferenceB(const int x, const int y, const uint32_t dir, const SearchData * const data) + { + /* dir : 0 = forward, 1 = backward */ const uint8_t *const *const direction = ( dir == 0 ? data->RefP : data->b_RefP ); const int picture = ((x&1)<<1) | (y&1); @@ -124,9 +124,9 @@ return direction[picture] + offset; } -static __inline void -ZeroMacroblockP(MACROBLOCK *pMB, const int32_t sad) -{ +static __inline void ZeroMacroblockP(MACROBLOCK *pMB, const int32_t sad) + { + pMB->mode = MODE_INTER; pMB->mvs[0] = pMB->mvs[1] = pMB->mvs[2] = pMB->mvs[3] = zeroMV; pMB->qmvs[0] = pMB->qmvs[1] = pMB->qmvs[2] = pMB->qmvs[3] = zeroMV; @@ -136,9 +136,9 @@ } /* check if given vector is equal to any vector checked before */ -static __inline int -vector_repeats(const VECTOR * const pmv, const unsigned int i) -{ +static __inline int vector_repeats(const VECTOR * const pmv, const unsigned int i) + { + unsigned int j; for (j = 0; j < i; j++) if (MVequal(pmv[i], pmv[j])) return 1; /* same vector has been checked already */ @@ -147,9 +147,9 @@ /* make a binary mask that prevents diamonds/squares from checking a vector which has been checked as a prediction */ -static __inline int -make_mask(const VECTOR * const pmv, const unsigned int i, const unsigned int current) -{ +static __inline int make_mask(const VECTOR * const pmv, const unsigned int i, const unsigned int current) + { + unsigned int mask = 255, j; for (j = 0; j < i; j++) { if (pmv[current].x == pmv[j].x) { Only in src-ori/motion/ppc_asm: README diff -bBru src-ori/motion/ppc_asm/sad_altivec.c src/motion/ppc_asm/sad_altivec.c --- src-ori/motion/ppc_asm/sad_altivec.c Wed Jan 12 10:32:43 2005 +++ src/motion/ppc_asm/sad_altivec.c Wed Jan 12 12:45:14 2005 @@ -1,7 +1,12 @@ -/* - Copyright (C) 2002 Benjamin Herrenschmidt + + + + + +/* + Copyright (C) 2002 Benjamin Herrenschmidt 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 @@ -32,7 +32,8 @@ register vector unsigned int zerovec asm("%v31"); #endif -#include +#include +#include "../../global_all_dll.h" #undef DEBUG Only in src-ori/motion/ppc_asm: sad_altivec.s diff -bBru src-ori/motion/sad.c src/motion/sad.c --- src-ori/motion/sad.c Wed Jan 12 10:32:43 2005 +++ src/motion/sad.c Wed Jan 12 12:43:30 2005 @@ -23,21 +23,11 @@ * ****************************************************************************/ -#include "../portab.h" -#include "../global.h" -#include "sad.h" - -#include - -sad16FuncPtr sad16; -sad8FuncPtr sad8; -sad16biFuncPtr sad16bi; -sad8biFuncPtr sad8bi; /* not really sad16, but no difference in prototype */ -dev16FuncPtr dev16; -sad16vFuncPtr sad16v; -sse8Func_16bitPtr sse8_16bit; +#include +#include "../global_all_dll.h" + + -sadInitFuncPtr sadInit; uint32_t @@ -215,10 +205,10 @@ const uint32_t stride, int32_t *sad) { - sad[0] = sad8(cur, ref, stride); - sad[1] = sad8(cur + 8, ref + 8, stride); - sad[2] = sad8(cur + 8*stride, ref + 8*stride, stride); - sad[3] = sad8(cur + 8*stride + 8, ref + 8*stride + 8, stride); + sad[0] = (((struct global_all_dll *)Dll::Tls())->sad8)(cur, ref, stride); + sad[1] = (((struct global_all_dll *)Dll::Tls())->sad8)(cur + 8, ref + 8, stride); + sad[2] = (((struct global_all_dll *)Dll::Tls())->sad8)(cur + 8*stride, ref + 8*stride, stride); + sad[3] = (((struct global_all_dll *)Dll::Tls())->sad8)(cur + 8*stride + 8, ref + 8*stride + 8, stride); return sad[0]+sad[1]+sad[2]+sad[3]; } @@ -228,10 +218,10 @@ const uint32_t stride, int32_t *sad) { - sad[0] = sad16(cur, ref, stride, 256*4096); - sad[1] = sad16(cur + 16, ref + 16, stride, 256*4096); - sad[2] = sad16(cur + 16*stride, ref + 16*stride, stride, 256*4096); - sad[3] = sad16(cur + 16*stride + 16, ref + 16*stride + 16, stride, 256*4096); + sad[0] = (((struct global_all_dll *)Dll::Tls())->sad16)(cur, ref, stride, 256*4096); + sad[1] = (((struct global_all_dll *)Dll::Tls())->sad16)(cur + 16, ref + 16, stride, 256*4096); + sad[2] = (((struct global_all_dll *)Dll::Tls())->sad16)(cur + 16*stride, ref + 16*stride, stride, 256*4096); + sad[3] = (((struct global_all_dll *)Dll::Tls())->sad16)(cur + 16*stride + 16, ref + 16*stride + 16, stride, 256*4096); return sad[0]+sad[1]+sad[2]+sad[3]; } diff -bBru src-ori/motion/sad.h src/motion/sad.h --- src-ori/motion/sad.h Wed Jan 12 10:32:43 2005 +++ src/motion/sad.h Wed Jan 12 12:38:14 2005 @@ -31,7 +31,7 @@ typedef void (sadInitFunc) (void); typedef sadInitFunc *sadInitFuncPtr; -extern sadInitFuncPtr sadInit; + sadInitFunc sadInit_altivec; typedef uint32_t(sad16Func) (const uint8_t * const cur, @@ -36,10 +36,13 @@ typedef uint32_t(sad16Func) (const uint8_t * const cur, const uint8_t * const ref, + const uint32_t stride, + const uint32_t best_sad); + typedef sad16Func *sad16FuncPtr; -extern sad16FuncPtr sad16; + sad16Func sad16_c; #ifdef ARCH_IS_IA32 @@ -61,9 +64,11 @@ typedef uint32_t(sad8Func) (const uint8_t * const cur, const uint8_t * const ref, + const uint32_t stride); + typedef sad8Func *sad8FuncPtr; -extern sad8FuncPtr sad8; + sad8Func sad8_c; #ifdef ARCH_IS_IA32 @@ -82,10 +87,13 @@ typedef uint32_t(sad16biFunc) (const uint8_t * const cur, const uint8_t * const ref1, + const uint8_t * const ref2, + const uint32_t stride); + typedef sad16biFunc *sad16biFuncPtr; -extern sad16biFuncPtr sad16bi; + sad16biFunc sad16bi_c; #ifdef ARCH_IS_IA32 @@ -101,10 +109,13 @@ typedef uint32_t(sad8biFunc) (const uint8_t * const cur, const uint8_t * const ref1, + const uint8_t * const ref2, + const uint32_t stride); + typedef sad8biFunc *sad8biFuncPtr; -extern sad8biFuncPtr sad8bi; + sad8biFunc sad8bi_c; #ifdef ARCH_IS_IA32 @@ -117,8 +128,9 @@ typedef uint32_t(dev16Func) (const uint8_t * const cur, const uint32_t stride); + typedef dev16Func *dev16FuncPtr; -extern dev16FuncPtr dev16; + dev16Func dev16_c; typedef uint32_t (sad16vFunc)( const uint8_t * const cur, @@ -123,9 +135,11 @@ typedef uint32_t (sad16vFunc)( const uint8_t * const cur, const uint8_t * const ref, + const uint32_t stride, int32_t *sad8); + typedef sad16vFunc *sad16vFuncPtr; -extern sad16vFuncPtr sad16v; + sad16vFunc sad16v_c; sad16vFunc sad32v_c; @@ -149,9 +163,11 @@ typedef uint32_t (sse8Func_16bit)(const int16_t * cur, const int16_t * ref, + const uint32_t stride); + typedef sse8Func_16bit *sse8Func_16bitPtr; -extern sse8Func_16bitPtr sse8_16bit; + sse8Func_16bit sse8_16bit_c; #ifdef ARCH_IS_IA32 diff -bBru src-ori/motion/vop_type_decision.c src/motion/vop_type_decision.c --- src-ori/motion/vop_type_decision.c Wed Jan 12 10:32:43 2005 +++ src/motion/vop_type_decision.c Wed Jan 12 12:44:34 2005 @@ -23,14 +23,8 @@ * ****************************************************************************/ -#include "../encoder.h" -#include "../prediction/mbprediction.h" -#include "estimation.h" -#include "motion.h" -#include "sad.h" -#include "gmc.h" -#include "../utils/emms.h" -#include "motion_inlines.h" +#include +#include "../global_all_dll.h" #define INTRA_THRESH 2000 @@ -186,7 +180,7 @@ InterThresh -= P_SENS_BIAS * bCount; if (InterThresh < INTER_THRESH_MIN) InterThresh = INTER_THRESH_MIN; - if (sadInit) (*sadInit) (); + if ((((struct global_all_dll *)Dll::Tls())->sadInit)) (*(((struct global_all_dll *)Dll::Tls())->sadInit)) (); for (y = 1; y < pParam->mb_height-1; y += 2) { for (x = 1; x < pParam->mb_width-1; x += 2) { @@ -206,7 +200,7 @@ for (i = 0; i < 4; i++) { int dev; MACROBLOCK *pMB = &pMBs[x+(i&1) + (y+(i>>1)) * pParam->mb_width]; - dev = dev16(pCurrent->y + (x + (i&1) + (y + (i>>1)) * pParam->edged_width) * 16, + dev = (((struct global_all_dll *)Dll::Tls())->dev16)(pCurrent->y + (x + (i&1) + (y + (i>>1)) * pParam->edged_width) * 16, pParam->edged_width); complexity += MAX(dev, 300); @@ -229,6 +223,7 @@ if (sSAD > IntraThresh2) return I_VOP; if (sSAD > InterThresh) return P_VOP; - emms(); + (((struct global_all_dll *)Dll::Tls())->emms)(); return B_VOP; } + Only in src-ori/motion: x86_asm diff -bBru src-ori/plugins/plugin_2pass1.c src/plugins/plugin_2pass1.c --- src-ori/plugins/plugin_2pass1.c Wed Jan 12 10:32:43 2005 +++ src/plugins/plugin_2pass1.c Wed Jan 12 12:45:54 2005 @@ -26,12 +26,9 @@ * *****************************************************************************/ -#include -#include /* errno var (or function with recent libc) */ -#include /* strerror() */ +#include +#include "../global_all_dll.h" -#include "../xvid.h" -#include "../image/image.h" /* This preprocessor constant controls wheteher or not, first pass is done @@ -42,8 +39,7 @@ /* context struct */ -typedef struct -{ +typedef struct { FILE * stat_file; double fq_error; diff -bBru src-ori/plugins/plugin_2pass2.c src/plugins/plugin_2pass2.c --- src-ori/plugins/plugin_2pass2.c Wed Jan 12 10:32:43 2005 +++ src/plugins/plugin_2pass2.c Wed Jan 12 12:46:25 2005 @@ -39,12 +39,9 @@ to fight most common problems without user's knowladge */ #define SMART_OVERFLOW_SETTING -#include -#include -#include +#include +#include "../global_all_dll.h" -#include "../xvid.h" -#include "../image/image.h" /***************************************************************************** * Some default settings @@ -76,8 +73,7 @@ ****************************************************************************/ /* Statistics */ -typedef struct { - int type; /* first pass type */ +typedef struct { int type; /* first pass type */ int quant; /* first pass quant */ int blks[3]; /* k,m,y blks */ int length; /* first pass length */ @@ -90,9 +86,9 @@ double weight; } twopass_stat_t; + /* Context struct */ -typedef struct -{ +typedef struct { xvid_plugin_2pass2_t param; /*---------------------------------- @@ -277,6 +274,7 @@ /*---------------------------------------------------------------------------- *--------------------------------------------------------------------------*/ +#define _INIT(a, b) if((a) <= 0) (a) = (b) static int rc_2pass2_create(xvid_plg_create_t * create, rc_2pass2_t **handle) { @@ -291,7 +289,6 @@ rc->param = *param; /* Initialize all defaults */ -#define _INIT(a, b) if((a) <= 0) (a) = (b) /* Let's set our defaults if needed */ _INIT(rc->param.keyframe_boost, DEFAULT_KEYFRAME_BOOST); _INIT(rc->param.overflow_control_strength, DEFAULT_OVERFLOW_CONTROL_STRENGTH); @@ -452,6 +450,10 @@ /*---------------------------------------------------------------------------- *--------------------------------------------------------------------------*/ +#if 0 + /* Leave this one alone, as it impacts badly on quality */ + overflow *= framesize_factor; +#endif static int rc_2pass2_before(rc_2pass2_t * rc, xvid_plg_data_t * data) { @@ -581,10 +583,6 @@ /* Treat only the overflow part concerned by this frame type and size */ overflow *= frametype_factor; -#if 0 - /* Leave this one alone, as it impacts badly on quality */ - overflow *= framesize_factor; -#endif /* Apply the overflow strength imposed by the user */ overflow *= (rc->param.overflow_control_strength/100.0f); diff -bBru src-ori/plugins/plugin_dump.c src/plugins/plugin_dump.c --- src-ori/plugins/plugin_dump.c Wed Jan 12 10:32:43 2005 +++ src/plugins/plugin_dump.c Wed Jan 12 12:45:50 2005 @@ -23,10 +23,9 @@ * ****************************************************************************/ -#include +#include +#include "../global_all_dll.h" -#include "../xvid.h" -#include "../image/image.h" int xvid_plugin_dump(void * handle, int opt, void * param1, void * param2) diff -bBru src-ori/plugins/plugin_fixed.c src/plugins/plugin_fixed.c --- src-ori/plugins/plugin_fixed.c Wed Jan 12 10:32:43 2005 +++ src/plugins/plugin_fixed.c Wed Jan 12 12:46:33 2005 @@ -24,12 +24,11 @@ ****************************************************************************/ -#include "../xvid.h" -#include "../image/image.h" +#include +#include "../global_all_dll.h" -typedef struct -{ +typedef struct { int32_t quant_increment; int32_t quant_base; diff -bBru src-ori/plugins/plugin_lumimasking.c src/plugins/plugin_lumimasking.c --- src-ori/plugins/plugin_lumimasking.c Wed Jan 12 10:32:43 2005 +++ src/plugins/plugin_lumimasking.c Wed Jan 12 12:45:50 2005 @@ -24,18 +24,15 @@ * ****************************************************************************/ -#include +#include +#include "../global_all_dll.h" -#include "../xvid.h" -#include "../portab.h" -#include "../utils/emms.h" /***************************************************************************** * Private data type ****************************************************************************/ -typedef struct -{ +typedef struct { float *quant; float *val; } lumi_data_t; diff -bBru src-ori/plugins/plugin_psnr.c src/plugins/plugin_psnr.c --- src-ori/plugins/plugin_psnr.c Wed Jan 12 10:32:43 2005 +++ src/plugins/plugin_psnr.c Wed Jan 12 12:45:51 2005 @@ -23,10 +23,9 @@ * ****************************************************************************/ -#include +#include +#include "../global_all_dll.h" -#include "../xvid.h" -#include "../image/image.h" int xvid_plugin_psnr(void * handle, int opt, void * param1, void * param2) diff -bBru src-ori/plugins/plugin_single.c src/plugins/plugin_single.c --- src-ori/plugins/plugin_single.c Wed Jan 12 10:32:43 2005 +++ src/plugins/plugin_single.c Wed Jan 12 12:45:46 2005 @@ -25,10 +25,9 @@ ****************************************************************************/ -#include +#include +#include "../global_all_dll.h" -#include "../xvid.h" -#include "../image/image.h" #define DEFAULT_INITIAL_QUANTIZER 4 @@ -37,8 +36,7 @@ #define DEFAULT_AVERAGING_PERIOD 100 #define DEFAULT_BUFFER 100 -typedef struct -{ +typedef struct { int reaction_delay_factor; int averaging_period; int buffer; diff -bBru src-ori/portab.h src/portab.h --- src-ori/portab.h Wed Jan 12 10:32:43 2005 +++ src/portab.h Wed Jan 12 12:38:11 2005 @@ -34,7 +34,7 @@ /* Buffer size for msvc implementation because it outputs to DebugOutput */ #if defined(_DEBUG) -extern unsigned int xvid_debug; + #define DPRINTF_BUF_SZ 1024 #endif @@ -131,8 +131,8 @@ * DPRINTF function when not compiling in _DEBUG mode */ # ifdef _DEBUG -static __inline void DPRINTF(int level, char *fmt, ...) -{ +static __inline void DPRINTF(int level, char *fmt, ...) { + if (xvid_debug & level) { va_list args; char buf[DPRINTF_BUF_SZ]; @@ -163,8 +163,8 @@ # if defined(ARCH_IS_IA32) # define BSWAP(a) __asm mov eax,a __asm bswap eax __asm mov a, eax -static __inline int64_t read_counter(void) -{ +static __inline int64_t read_counter(void) { + int64_t ts; uint32_t ts1, ts2; __asm { @@ -185,8 +185,8 @@ (((a) >> 8) & 0xff00) | (((a) >> 24) & 0xff)) # include -static __inline int64_t read_counter(void) -{ +static __inline int64_t read_counter(void) { + return (int64_t)clock(); } @@ -221,8 +221,8 @@ # include # include -static __inline void DPRINTF(int level, char *format, ...) -{ +static __inline void DPRINTF(int level, char *format, ...) { + va_list args; va_start(args, format); if(xvid_debug & level) { @@ -246,8 +246,8 @@ # if defined(ARCH_IS_IA32) # define BSWAP(a) __asm__ ( "bswapl %0\n" : "=r" (a) : "0" (a) ); -static __inline int64_t read_counter(void) -{ +static __inline int64_t read_counter(void) { + int64_t ts; uint32_t ts1, ts2; __asm__ __volatile__("rdtsc\n\t":"=a"(ts1), "=d"(ts2)); @@ -262,22 +262,22 @@ # define BSWAP(a) __asm__ __volatile__ \ ( "lwbrx %0,0,%1; eieio" : "=r" (a) : "r" (&(a)), "m" (a)); -static __inline unsigned long get_tbl(void) -{ +static __inline unsigned long get_tbl(void) { + unsigned long tbl; asm volatile ("mftb %0":"=r" (tbl)); return tbl; } -static __inline unsigned long get_tbu(void) -{ +static __inline unsigned long get_tbu(void) { + unsigned long tbl; asm volatile ("mftbu %0":"=r" (tbl)); return tbl; } -static __inline int64_t read_counter(void) -{ +static __inline int64_t read_counter(void) { + unsigned long tb, tu; do { tu = get_tbu(); @@ -294,8 +294,8 @@ ("mux1 %1 = %0, @rev" ";;" \ "shr.u %1 = %1, 32" : "=r" (a) : "r" (a)); -static __inline int64_t read_counter(void) -{ +static __inline int64_t read_counter(void) { + unsigned long result; __asm__ __volatile__("mov %0=ar.itc" : "=r"(result) :: "memory"); return result; @@ -310,8 +310,8 @@ (((a) >> 8) & 0xff00) | (((a) >> 24) & 0xff)) # include -static __inline int64_t read_counter(void) -{ +static __inline int64_t read_counter(void) { + return (int64_t)clock(); } @@ -336,8 +336,8 @@ # include # ifdef _DEBUG -static __inline void DPRINTF(int level, char *fmt, ...) -{ +static __inline void DPRINTF(int level, char *fmt, ...) { + if (xvid_debug & level) { va_list args; char buf[DPRINTF_BUF_SZ]; @@ -362,8 +362,8 @@ # define BSWAP(a) __asm mov eax,a __asm bswap eax __asm mov a, eax -static __inline int64_t read_counter(void) -{ +static __inline int64_t read_counter(void) { + uint64_t ts; uint32_t ts1, ts2; __asm { @@ -415,8 +415,8 @@ # include # include -static __inline void DPRINTF(int level, char *format, ...) -{ +static __inline void DPRINTF(int level, char *format, ...) { + va_list args; va_start(args, format); if(xvid_debug & level) { @@ -434,8 +434,8 @@ (((a) >> 8) & 0xff00) | (((a) >> 24) & 0xff)) # include -static __inline int64_t read_counter(void) -{ +static __inline int64_t read_counter(void) { + return (int64_t)clock(); } diff -bBru src-ori/prediction/mbprediction.c src/prediction/mbprediction.c --- src-ori/prediction/mbprediction.c Wed Jan 12 10:58:23 2005 +++ src/prediction/mbprediction.c Wed Jan 12 12:45:43 2005 @@ -24,15 +24,9 @@ * ****************************************************************************/ -#include +#include +#include "../global_all_dll.h" -#include "../global.h" -#include "../encoder.h" -#include "mbprediction.h" -#include "../utils/mbfunctions.h" -#include "../bitstream/cbp.h" -#include "../bitstream/mbcoding.h" -#include "../bitstream/zigzag.h" static int __inline @@ -262,10 +256,10 @@ /***************************************************************************** ****************************************************************************/ -/* encoder: subtract predictors from qcoeff[] and calculate S1/S2 -returns sum of coeefficients *saved* if prediction is enabled +/* encoder: subtract predictors from qcoeff[] and calculate S1/S2 +returns sum of coeefficients *saved* if prediction is enabled S1 = sum of all (qcoeff - prediction) S2 = sum of all qcoeff */ @@ -443,7 +437,7 @@ apply_acdc(pMB, j, &qcoeff[j * 64], predictors[j]); } - pMB->cbp = calc_cbp(qcoeff); + pMB->cbp = (((struct global_all_dll *)Dll::Tls())->calc_cbp)(qcoeff); } } Only in src-ori/quant: ia64_asm diff -bBru src-ori/quant/quant.h src/quant/quant.h --- src-ori/quant/quant.h Wed Jan 12 10:32:43 2005 +++ src/quant/quant.h Wed Jan 12 12:37:58 2005 @@ -41,10 +45,10 @@ typedef quant_intraFunc *quant_intraFuncPtr; /* Global function pointers */ -extern quant_intraFuncPtr quant_h263_intra; -extern quant_intraFuncPtr quant_mpeg_intra; -extern quant_intraFuncPtr dequant_h263_intra; -extern quant_intraFuncPtr dequant_mpeg_intra; + + + + /***************************************************************************** * Known implementation of Intra (de)Quant functions @@ -97,10 +104,10 @@ typedef quant_interFunc *quant_interFuncPtr; /* Global function pointers */ -extern quant_interFuncPtr quant_h263_inter; -extern quant_interFuncPtr quant_mpeg_inter; -extern quant_interFuncPtr dequant_h263_inter; -extern quant_interFuncPtr dequant_mpeg_inter; + + + + /***************************************************************************** * Known implementation of Inter (de)Quant functions diff -bBru src-ori/quant/quant_h263.c src/quant/quant_h263.c --- src-ori/quant/quant_h263.c Wed Jan 12 10:32:43 2005 +++ src/quant/quant_h263.c Wed Jan 12 12:41:19 2005 @@ -23,20 +23,14 @@ * ****************************************************************************/ -#include "../global.h" -#include "quant.h" +#include +#include "../global_all_dll.h" /***************************************************************************** * Global function pointers ****************************************************************************/ -/* Quant */ -quant_intraFuncPtr quant_h263_intra; -quant_interFuncPtr quant_h263_inter; - -/* DeQuant */ -quant_intraFuncPtr dequant_h263_intra; -quant_interFuncPtr dequant_h263_inter; + /***************************************************************************** * Local data diff -bBru src-ori/quant/quant_matrix.c src/quant/quant_matrix.c --- src-ori/quant/quant_matrix.c Wed Jan 12 10:32:43 2005 +++ src/quant/quant_matrix.c Wed Jan 12 12:41:17 2005 @@ -24,7 +24,8 @@ * ****************************************************************************/ -#include "quant_matrix.h" +#include +#include "../global_all_dll.h" #define FIX(X) (((X)==1) ? 0xFFFF : ((1UL << 16) / (X) + 1)) #define FIXL(X) ((1UL << 16) / (X) - 1) diff -bBru src-ori/quant/quant_mpeg.c src/quant/quant_mpeg.c --- src-ori/quant/quant_mpeg.c Wed Jan 12 10:32:43 2005 +++ src/quant/quant_mpeg.c Wed Jan 12 12:41:23 2005 @@ -23,21 +23,14 @@ * ****************************************************************************/ -#include "../global.h" -#include "quant.h" -#include "quant_matrix.h" +#include +#include "../global_all_dll.h" /***************************************************************************** * Global function pointers ****************************************************************************/ -/* Quant */ -quant_intraFuncPtr quant_mpeg_intra; -quant_interFuncPtr quant_mpeg_inter; - -/* DeQuant */ -quant_intraFuncPtr dequant_mpeg_intra; -quant_interFuncPtr dequant_mpeg_inter; + /***************************************************************************** * Local data Only in src-ori/quant: x86_asm diff -bBru src-ori/utils/emms.c src/utils/emms.c --- src-ori/utils/emms.c Wed Jan 12 10:32:43 2005 +++ src/utils/emms.c Wed Jan 12 12:41:27 2005 @@ -23,14 +23,13 @@ * ****************************************************************************/ -#include "emms.h" -#include "../portab.h" +#include +#include "../global_all_dll.h" /***************************************************************************** * Library data, declared here ****************************************************************************/ -emmsFuncPtr emms; /***************************************************************************** diff -bBru src-ori/utils/emms.h src/utils/emms.h --- src-ori/utils/emms.h Wed Jan 12 10:32:43 2005 +++ src/utils/emms.h Wed Jan 12 12:37:59 2005 @@ -27,7 +27,7 @@ #define _EMMS_H_ /***************************************************************************** - * emms API + * (((struct global_all_dll *)Dll::Tls())->emms) API ****************************************************************************/ typedef void (emmsFunc) (); @@ -35,7 +35,7 @@ typedef emmsFunc *emmsFuncPtr; /* Our global function pointer - defined in emms.c */ -extern emmsFuncPtr emms; + /* Implemented functions */ Only in src-ori/utils: ia64_asm diff -bBru src-ori/utils/mbfunctions.h src/utils/mbfunctions.h --- src-ori/utils/mbfunctions.h Wed Jan 12 10:32:43 2005 +++ src/utils/mbfunctions.h Wed Jan 12 12:38:00 2005 @@ -61,8 +61,9 @@ typedef uint32_t (MBFIELDTEST) (int16_t data[6 * 64]); /* function pointer for field test */ typedef MBFIELDTEST *MBFIELDTEST_PTR; + /* global field test pointer for xvid.c */ -extern MBFIELDTEST_PTR MBFieldTest; + /* field test implementations */ MBFIELDTEST MBFieldTest_c; diff -bBru src-ori/utils/mbtransquant.c src/utils/mbtransquant.c --- src-ori/utils/mbtransquant.c Wed Jan 12 10:32:43 2005 +++ src/utils/mbtransquant.c Wed Jan 12 12:41:55 2005 @@ -25,27 +25,12 @@ * ****************************************************************************/ -#include -#include -#include - -#include "../portab.h" -#include "mbfunctions.h" - -#include "../global.h" -#include "mem_transfer.h" -#include "timer.h" -#include "../bitstream/mbcoding.h" -#include "../bitstream/zigzag.h" -#include "../dct/fdct.h" -#include "../dct/idct.h" -#include "../quant/quant.h" -#include "../encoder.h" +#include +#include "../global_all_dll.h" + + -#include "../image/reduced.h" -#include "../quant/quant_matrix.h" -MBFIELDTEST_PTR MBFieldTest; /* * Skip blocks having a coefficient sum below this value. This value will be @@ -62,7 +47,7 @@ static __inline uint32_t MBDecideFieldDCT(int16_t data[6 * 64]) { - uint32_t field = MBFieldTest(data); + uint32_t field = (((struct global_all_dll *)Dll::Tls())->MBFieldTest)(data); if (field) MBFrameToField(data); @@ -91,12 +76,12 @@ /* Perform DCT */ start_timer(); - fdct(&data[0 * 64]); - fdct(&data[1 * 64]); - fdct(&data[2 * 64]); - fdct(&data[3 * 64]); - fdct(&data[4 * 64]); - fdct(&data[5 * 64]); + (((struct global_all_dll *)Dll::Tls())->fdct)(&data[0 * 64]); + (((struct global_all_dll *)Dll::Tls())->fdct)(&data[1 * 64]); + (((struct global_all_dll *)Dll::Tls())->fdct)(&data[2 * 64]); + (((struct global_all_dll *)Dll::Tls())->fdct)(&data[3 * 64]); + (((struct global_all_dll *)Dll::Tls())->fdct)(&data[4 * 64]); + (((struct global_all_dll *)Dll::Tls())->fdct)(&data[5 * 64]); stop_dct_timer(); } @@ -106,12 +91,12 @@ const uint8_t cbp) { start_timer(); - if(cbp & (1 << (5 - 0))) idct(&data[0 * 64]); - if(cbp & (1 << (5 - 1))) idct(&data[1 * 64]); - if(cbp & (1 << (5 - 2))) idct(&data[2 * 64]); - if(cbp & (1 << (5 - 3))) idct(&data[3 * 64]); - if(cbp & (1 << (5 - 4))) idct(&data[4 * 64]); - if(cbp & (1 << (5 - 5))) idct(&data[5 * 64]); + if(cbp & (1 << (5 - 0))) (((struct global_all_dll *)Dll::Tls())->idct)(&data[0 * 64]); + if(cbp & (1 << (5 - 1))) (((struct global_all_dll *)Dll::Tls())->idct)(&data[1 * 64]); + if(cbp & (1 << (5 - 2))) (((struct global_all_dll *)Dll::Tls())->idct)(&data[2 * 64]); + if(cbp & (1 << (5 - 3))) (((struct global_all_dll *)Dll::Tls())->idct)(&data[3 * 64]); + if(cbp & (1 << (5 - 4))) (((struct global_all_dll *)Dll::Tls())->idct)(&data[4 * 64]); + if(cbp & (1 << (5 - 5))) (((struct global_all_dll *)Dll::Tls())->idct)(&data[5 * 64]); stop_idct_timer(); } @@ -128,8 +113,8 @@ quant_intraFuncPtr const quant[2] = { - quant_h263_intra, - quant_mpeg_intra + (((struct global_all_dll *)Dll::Tls())->quant_h263_intra), + (((struct global_all_dll *)Dll::Tls())->quant_mpeg_intra) }; mpeg = !!(pParam->vol_flags & XVID_VOL_MPEGQUANT); @@ -159,8 +144,8 @@ quant_intraFuncPtr const dequant[2] = { - dequant_h263_intra, - dequant_mpeg_intra + (((struct global_all_dll *)Dll::Tls())->dequant_h263_intra), + (((struct global_all_dll *)Dll::Tls())->dequant_mpeg_intra) }; mpeg = !!(pParam->vol_flags & XVID_VOL_MPEGQUANT); @@ -204,8 +189,8 @@ quant_interFuncPtr const quant[2] = { - quant_h263_inter, - quant_mpeg_inter + (((struct global_all_dll *)Dll::Tls())->quant_h263_inter), + (((struct global_all_dll *)Dll::Tls())->quant_mpeg_inter) }; mpeg = !!(pParam->vol_flags & XVID_VOL_MPEGQUANT); @@ -279,8 +264,8 @@ quant_interFuncPtr const dequant[2] = { - dequant_h263_inter, - dequant_mpeg_inter + (((struct global_all_dll *)Dll::Tls())->dequant_h263_inter), + (((struct global_all_dll *)Dll::Tls())->dequant_mpeg_inter) }; mpeg = !!(pParam->vol_flags & XVID_VOL_MPEGQUANT); @@ -316,8 +301,8 @@ const IMAGE * const pCurrent = &frame->image; transfer_operation_8to16_t * const functions[2] = { - (transfer_operation_8to16_t *)transfer_8to16copy, - (transfer_operation_8to16_t *)filter_18x18_to_8x8 + (transfer_operation_8to16_t *)(((struct global_all_dll *)Dll::Tls())->transfer_8to16copy), + (transfer_operation_8to16_t *)(((struct global_all_dll *)Dll::Tls())->filter_18x18_to_8x8) }; transfer_operation_8to16_t *transfer_op = NULL; @@ -366,10 +351,10 @@ /* Array of function pointers, indexed by [vop_reduced<<1+add] */ transfer_operation_16to8_t * const functions[4] = { - (transfer_operation_16to8_t*)transfer_16to8copy, - (transfer_operation_16to8_t*)transfer_16to8add, - (transfer_operation_16to8_t*)copy_upsampled_8x8_16to8, - (transfer_operation_16to8_t*)add_upsampled_8x8_16to8 + (transfer_operation_16to8_t*)(((struct global_all_dll *)Dll::Tls())->transfer_16to8copy), + (transfer_operation_16to8_t*)(((struct global_all_dll *)Dll::Tls())->transfer_16to8add), + (transfer_operation_16to8_t*)(((struct global_all_dll *)Dll::Tls())->copy_upsampled_8x8_16to8), + (transfer_operation_16to8_t*)(((struct global_all_dll *)Dll::Tls())->add_upsampled_8x8_16to8) }; transfer_operation_16to8_t *transfer_op = NULL; @@ -743,7 +728,7 @@ 12,13,13,13,13,13,13,13,13,14,16,16,16,16,17,17,17,17,18,18,18,18,18,18,18,18,19,19,19,19,19,19}; -static const uint8_t * const B16_17_Code_Len[24] = { /* levels [1..24] */ +static const uint8_t * const B16_17_Code_Len[24] = { Code_Len20,Code_Len19,Code_Len18,Code_Len17, Code_Len16,Code_Len15,Code_Len14,Code_Len13, Code_Len12,Code_Len11,Code_Len10,Code_Len9, @@ -752,7 +737,7 @@ Code_Len2, Code_Len1, Code_Len1, Code_Len1, }; -static const uint8_t * const B16_17_Code_Len_Last[6] = { /* levels [1..6] */ +static const uint8_t * const B16_17_Code_Len_Last[6] = { Code_Len24,Code_Len23,Code_Len22,Code_Len21, Code_Len3, Code_Len1, }; @@ -784,6 +769,7 @@ } /* this routine has been strippen of all debug code */ + typedef struct { int16_t Run, Level; } NODE; static int dct_quantize_trellis_c(int16_t *const Out, const int16_t *const In, @@ -799,7 +785,6 @@ * *very* slightly (~0.01dB), whereas speed drops to crawling level :) * Well, actually, taking 1 more coeff past Non_Zero into account sometimes * helps. */ - typedef struct { int16_t Run, Level; } NODE; NODE Nodes[65], Last; uint32_t Run_Costs0[64+1]; @@ -1080,6 +1065,7 @@ } + typedef struct { int16_t Run, Level; } NODE; static int dct_quantize_trellis_h263_c(int16_t *const Out, const int16_t *const In, int Q, const uint16_t * const Zigzag, int Non_Zero) { @@ -1090,7 +1076,6 @@ * slightly (~0.01dB), whereas speed drops to crawling level :) * Well, actually, taking 1 more coeff past Non_Zero into account sometimes helps. */ - typedef struct { int16_t Run, Level; } NODE; NODE Nodes[65], Last; uint32_t Run_Costs0[64+1]; diff -bBru src-ori/utils/mem_align.c src/utils/mem_align.c --- src-ori/utils/mem_align.c Wed Jan 12 10:32:43 2005 +++ src/utils/mem_align.c Wed Jan 12 12:41:28 2005 @@ -23,9 +23,8 @@ * ****************************************************************************/ -#include -#include -#include "mem_align.h" +#include +#include "../global_all_dll.h" /***************************************************************************** * xvid_malloc diff -bBru src-ori/utils/mem_transfer.c src/utils/mem_transfer.c --- src-ori/utils/mem_transfer.c Wed Jan 12 10:32:43 2005 +++ src/utils/mem_transfer.c Wed Jan 12 12:41:32 2005 @@ -23,20 +23,34 @@ * ****************************************************************************/ -#include "../global.h" -#include "mem_transfer.h" +#include +#include "../global_all_dll.h" + void transfer16x16_copy(uint8_t * const dst, + const uint8_t * const src, + const uint32_t stride) + { + + (((struct global_all_dll *)Dll::Tls())->transfer8x8_copy)(dst, src, stride); + (((struct global_all_dll *)Dll::Tls())->transfer8x8_copy)(dst + 8, src + 8, stride); + (((struct global_all_dll *)Dll::Tls())->transfer8x8_copy)(dst + 8*stride, src + 8*stride, stride); + (((struct global_all_dll *)Dll::Tls())->transfer8x8_copy)(dst + 8*stride + 8, src + 8*stride + 8, stride); +} + void transfer32x32_copy(uint8_t * const dst, + const uint8_t * const src, + const uint32_t stride) + { + + transfer16x16_copy(dst, src, stride); + transfer16x16_copy(dst + 16, src + 16, stride); + transfer16x16_copy(dst + 16*stride, src + 16*stride, stride); + transfer16x16_copy(dst + 16*stride + 16, src + 16*stride + 16, stride); +} + /* Function pointers - Initialized in the xvid.c module */ -TRANSFER_8TO16COPY_PTR transfer_8to16copy; -TRANSFER_16TO8COPY_PTR transfer_16to8copy; -TRANSFER_8TO16SUB_PTR transfer_8to16sub; -TRANSFER_8TO16SUBRO_PTR transfer_8to16subro; -TRANSFER_8TO16SUB2_PTR transfer_8to16sub2; -TRANSFER_16TO8ADD_PTR transfer_16to8add; -TRANSFER8X8_COPY_PTR transfer8x8_copy; /***************************************************************************** diff -bBru src-ori/utils/mem_transfer.h src/utils/mem_transfer.h --- src-ori/utils/mem_transfer.h Wed Jan 12 10:32:43 2005 +++ src/utils/mem_transfer.h Wed Jan 12 12:38:03 2005 @@ -37,7 +39,7 @@ typedef TRANSFER_8TO16COPY *TRANSFER_8TO16COPY_PTR; /* Our global function pointer - Initialized in xvid.c */ -extern TRANSFER_8TO16COPY_PTR transfer_8to16copy; + /* Implemented functions */ extern TRANSFER_8TO16COPY transfer_8to16copy_c; @@ -61,7 +65,7 @@ typedef TRANSFER_16TO8COPY *TRANSFER_16TO8COPY_PTR; /* Our global function pointer - Initialized in xvid.c */ -extern TRANSFER_16TO8COPY_PTR transfer_16to8copy; + /* Implemented functions */ extern TRANSFER_16TO8COPY transfer_16to8copy_c; @@ -87,7 +94,7 @@ typedef TRANSFER_8TO16SUB *TRANSFER_8TO16SUB_PTR; /* Our global function pointer - Initialized in xvid.c */ -extern TRANSFER_8TO16SUB_PTR transfer_8to16sub; + /* Implemented functions */ extern TRANSFER_8TO16SUB transfer_8to16sub_c; @@ -113,7 +123,7 @@ typedef TRANSFER_8TO16SUBRO *TRANSFER_8TO16SUBRO_PTR; /* Our global function pointer - Initialized in xvid.c */ -extern TRANSFER_8TO16SUBRO_PTR transfer_8to16subro; + /* Implemented functions */ extern TRANSFER_8TO16SUBRO transfer_8to16subro_c; @@ -136,7 +150,7 @@ typedef TRANSFER_8TO16SUB2 *TRANSFER_8TO16SUB2_PTR; /* Our global function pointer - Initialized in xvid.c */ -extern TRANSFER_8TO16SUB2_PTR transfer_8to16sub2; + /* Implemented functions */ extern TRANSFER_8TO16SUB2 transfer_8to16sub2_c; @@ -162,7 +178,7 @@ typedef TRANSFER_16TO8ADD *TRANSFER_16TO8ADD_PTR; /* Our global function pointer - Initialized in xvid.c */ -extern TRANSFER_16TO8ADD_PTR transfer_16to8add; + /* Implemented functions */ extern TRANSFER_16TO8ADD transfer_16to8add_c; @@ -187,7 +205,7 @@ typedef TRANSFER8X8_COPY *TRANSFER8X8_COPY_PTR; /* Our global function pointer - Initialized in xvid.c */ -extern TRANSFER8X8_COPY_PTR transfer8x8_copy; + /* Implemented functions */ extern TRANSFER8X8_COPY transfer8x8_copy_c; @@ -201,27 +219,15 @@ extern TRANSFER8X8_COPY transfer8x8_copy_ia64; #endif -static __inline void -transfer16x16_copy(uint8_t * const dst, +extern void transfer16x16_copy(uint8_t * const dst, const uint8_t * const src, const uint32_t stride) -{ - transfer8x8_copy(dst, src, stride); - transfer8x8_copy(dst + 8, src + 8, stride); - transfer8x8_copy(dst + 8*stride, src + 8*stride, stride); - transfer8x8_copy(dst + 8*stride + 8, src + 8*stride + 8, stride); -} +; -static __inline void -transfer32x32_copy(uint8_t * const dst, +extern void transfer32x32_copy(uint8_t * const dst, const uint8_t * const src, const uint32_t stride) -{ - transfer16x16_copy(dst, src, stride); - transfer16x16_copy(dst + 16, src + 16, stride); - transfer16x16_copy(dst + 16*stride, src + 16*stride, stride); - transfer16x16_copy(dst + 16*stride + 16, src + 16*stride + 16, stride); -} +; #endif diff -bBru src-ori/utils/timer.c src/utils/timer.c --- src-ori/utils/timer.c Wed Jan 12 10:32:43 2005 +++ src/utils/timer.c Wed Jan 12 12:41:27 2005 @@ -23,9 +23,81 @@ * ****************************************************************************/ -#include -#include -#include "timer.h" +#include +#include "../global_all_dll.h" + void start_timer() + { + +} + void start_global_timer() + { + +} + void stop_dct_timer() + { + +} + void stop_idct_timer() + { + +} + void stop_motion_timer() + { + +} + void stop_comp_timer() + { + +} + void stop_edges_timer() + { + +} + void stop_inter_timer() + { + +} + void stop_quant_timer() + { + +} + void stop_iquant_timer() + { + +} + void stop_conv_timer() + { + +} + void stop_transfer_timer() + { + +} + void init_timer() + { + +} + void write_timer() + { + +} + void stop_coding_timer() + { + +} + void stop_interlacing_timer() + { + +} + void stop_prediction_timer() + { + +} + void stop_global_timer() + { + +} + #if defined(_PROFILING_) @@ -51,7 +123,6 @@ struct ts tim; -double frequency = 0.0; /* determine cpu frequency @@ -81,7 +152,7 @@ void init_timer() { - frequency = get_freq(); + (((struct global_all_dll *)Dll::Tls())->frequency) = get_freq(); count_frames = 0; @@ -270,21 +341,21 @@ "Coding:\nTotal time: %f ms (%3f percent of total encoding time)\n\n\n" "Interlacing:\nTotal time: %f ms (%3f percent of total encoding time)\n\n\n" "Overall encoding time: %f ms, we measured %f ms (%3f percent)\n", - (float) (tim.dct / frequency), dct_per, - (float) (tim.quant / frequency), quant_per, - (float) (tim.idct / frequency), idct_per, - (float) (tim.iquant / frequency), iquant_per, - (float) (tim.motion / frequency), mot_per, - (float) (tim.comp / frequency), comp_per, - (float) (tim.edges / frequency), edges_per, - (float) (tim.inter / frequency), inter_per, - (float) (tim.conv / frequency), conv_per, - (float) (tim.trans / frequency), trans_per, - (float) (tim.prediction / frequency), pred_per, - (float) (tim.coding / frequency), cod_per, - (float) (tim.interlacing / frequency), interlacing_per, - (float) (tim.overall / frequency), - (float) (sum_ticks / frequency), measured); + (float) (tim.dct / (((struct global_all_dll *)Dll::Tls())->frequency)), dct_per, + (float) (tim.quant / (((struct global_all_dll *)Dll::Tls())->frequency)), quant_per, + (float) (tim.idct / (((struct global_all_dll *)Dll::Tls())->frequency)), idct_per, + (float) (tim.iquant / (((struct global_all_dll *)Dll::Tls())->frequency)), iquant_per, + (float) (tim.motion / (((struct global_all_dll *)Dll::Tls())->frequency)), mot_per, + (float) (tim.comp / (((struct global_all_dll *)Dll::Tls())->frequency)), comp_per, + (float) (tim.edges / (((struct global_all_dll *)Dll::Tls())->frequency)), edges_per, + (float) (tim.inter / (((struct global_all_dll *)Dll::Tls())->frequency)), inter_per, + (float) (tim.conv / (((struct global_all_dll *)Dll::Tls())->frequency)), conv_per, + (float) (tim.trans / (((struct global_all_dll *)Dll::Tls())->frequency)), trans_per, + (float) (tim.prediction / (((struct global_all_dll *)Dll::Tls())->frequency)), pred_per, + (float) (tim.coding / (((struct global_all_dll *)Dll::Tls())->frequency)), cod_per, + (float) (tim.interlacing / (((struct global_all_dll *)Dll::Tls())->frequency)), interlacing_per, + (float) (tim.overall / (((struct global_all_dll *)Dll::Tls())->frequency)), + (float) (sum_ticks / (((struct global_all_dll *)Dll::Tls())->frequency)), measured); fclose(fp); } diff -bBru src-ori/utils/timer.h src/utils/timer.h --- src-ori/utils/timer.h Wed Jan 12 10:32:43 2005 +++ src/utils/timer.h Wed Jan 12 12:37:59 2005 @@ -53,78 +53,42 @@ #else -static __inline void -start_timer() -{ -} -static __inline void -start_global_timer() -{ -} -static __inline void -stop_dct_timer() -{ -} -static __inline void -stop_idct_timer() -{ -} -static __inline void -stop_motion_timer() -{ -} -static __inline void -stop_comp_timer() -{ -} -static __inline void -stop_edges_timer() -{ -} -static __inline void -stop_inter_timer() -{ -} -static __inline void -stop_quant_timer() -{ -} -static __inline void -stop_iquant_timer() -{ -} -static __inline void -stop_conv_timer() -{ -} -static __inline void -stop_transfer_timer() -{ -} -static __inline void -init_timer() -{ -} -static __inline void -write_timer() -{ -} -static __inline void -stop_coding_timer() -{ -} -static __inline void -stop_interlacing_timer() -{ -} -static __inline void -stop_prediction_timer() -{ -} -static __inline void -stop_global_timer() -{ -} +extern void start_timer() +; +extern void start_global_timer() +; +extern void stop_dct_timer() +; +extern void stop_idct_timer() +; +extern void stop_motion_timer() +; +extern void stop_comp_timer() +; +extern void stop_edges_timer() +; +extern void stop_inter_timer() +; +extern void stop_quant_timer() +; +extern void stop_iquant_timer() +; +extern void stop_conv_timer() +; +extern void stop_transfer_timer() +; +extern void init_timer() +; +extern void write_timer() +; +extern void stop_coding_timer() +; +extern void stop_interlacing_timer() +; +extern void stop_prediction_timer() +; +extern void stop_global_timer() +; #endif Only in src-ori/utils: x86_asm diff -bBru src-ori/xvid.c src/xvid.c --- src-ori/xvid.c Wed Jan 12 10:32:43 2005 +++ src/xvid.c Wed Jan 12 12:45:36 2005 @@ -23,48 +23,22 @@ * ****************************************************************************/ -#include -#include -#include -#include - -#include "xvid.h" -#include "decoder.h" -#include "encoder.h" -#include "bitstream/cbp.h" -#include "dct/idct.h" -#include "dct/fdct.h" -#include "image/colorspace.h" -#include "image/interpolate8x8.h" -#include "image/reduced.h" -#include "utils/mem_transfer.h" -#include "utils/mbfunctions.h" -#include "quant/quant.h" -#include "motion/motion.h" -#include "motion/sad.h" -#include "utils/emms.h" -#include "utils/timer.h" -#include "bitstream/mbcoding.h" -#include "image/qpel.h" -#include "image/postprocessing.h" +#include +#include "global_all_dll.h" + #if defined(_DEBUG) -unsigned int xvid_debug = 0; /* xvid debug mask */ #endif #if defined(ARCH_IS_IA32) #if defined(_MSC_VER) -# include #else -# include -# include - static jmp_buf mark; static void sigill_handler(int signal) { - longjmp(mark, 1); + longjmp((((struct global_all_dll *)Dll::Tls())->mark), 1); } #endif @@ -103,7 +77,7 @@ return -1; } - jmpret = setjmp(mark); + jmpret = setjmp((((struct global_all_dll *)Dll::Tls())->mark)); if (jmpret == 0) { func(); @@ -173,133 +147,133 @@ init_vlc_tables(); /* Fixed Point Forward/Inverse DCT transformations */ - fdct = fdct_int32; - idct = idct_int32; + (((struct global_all_dll *)Dll::Tls())->fdct) = fdct_int32; + (((struct global_all_dll *)Dll::Tls())->idct) = idct_int32; /* Only needed on PPC Altivec archs */ - sadInit = 0; + (((struct global_all_dll *)Dll::Tls())->sadInit) = 0; /* Restore FPU context : emms_c is a nop functions */ - emms = emms_c; + (((struct global_all_dll *)Dll::Tls())->emms) = emms_c; /* Qpel stuff */ - xvid_QP_Funcs = &xvid_QP_Funcs_C; - xvid_QP_Add_Funcs = &xvid_QP_Add_Funcs_C; + (((struct global_all_dll *)Dll::Tls())->xvid_QP_Funcs) = &xvid_QP_Funcs_C; + (((struct global_all_dll *)Dll::Tls())->xvid_QP_Add_Funcs) = &xvid_QP_Add_Funcs_C; xvid_Init_QP(); /* Quantization functions */ - quant_h263_intra = quant_h263_intra_c; - quant_h263_inter = quant_h263_inter_c; - dequant_h263_intra = dequant_h263_intra_c; - dequant_h263_inter = dequant_h263_inter_c; - - quant_mpeg_intra = quant_mpeg_intra_c; - quant_mpeg_inter = quant_mpeg_inter_c; - dequant_mpeg_intra = dequant_mpeg_intra_c; - dequant_mpeg_inter = dequant_mpeg_inter_c; + (((struct global_all_dll *)Dll::Tls())->quant_h263_intra) = quant_h263_intra_c; + (((struct global_all_dll *)Dll::Tls())->quant_h263_inter) = quant_h263_inter_c; + (((struct global_all_dll *)Dll::Tls())->dequant_h263_intra) = dequant_h263_intra_c; + (((struct global_all_dll *)Dll::Tls())->dequant_h263_inter) = dequant_h263_inter_c; + + (((struct global_all_dll *)Dll::Tls())->quant_mpeg_intra) = quant_mpeg_intra_c; + (((struct global_all_dll *)Dll::Tls())->quant_mpeg_inter) = quant_mpeg_inter_c; + (((struct global_all_dll *)Dll::Tls())->dequant_mpeg_intra) = dequant_mpeg_intra_c; + (((struct global_all_dll *)Dll::Tls())->dequant_mpeg_inter) = dequant_mpeg_inter_c; /* Block transfer related functions */ - transfer_8to16copy = transfer_8to16copy_c; - transfer_16to8copy = transfer_16to8copy_c; - transfer_8to16sub = transfer_8to16sub_c; - transfer_8to16subro = transfer_8to16subro_c; - transfer_8to16sub2 = transfer_8to16sub2_c; - transfer_16to8add = transfer_16to8add_c; - transfer8x8_copy = transfer8x8_copy_c; + (((struct global_all_dll *)Dll::Tls())->transfer_8to16copy) = transfer_8to16copy_c; + (((struct global_all_dll *)Dll::Tls())->transfer_16to8copy) = transfer_16to8copy_c; + (((struct global_all_dll *)Dll::Tls())->transfer_8to16sub) = transfer_8to16sub_c; + (((struct global_all_dll *)Dll::Tls())->transfer_8to16subro) = transfer_8to16subro_c; + (((struct global_all_dll *)Dll::Tls())->transfer_8to16sub2) = transfer_8to16sub2_c; + (((struct global_all_dll *)Dll::Tls())->transfer_16to8add) = transfer_16to8add_c; + (((struct global_all_dll *)Dll::Tls())->transfer8x8_copy) = transfer8x8_copy_c; /* Interlacing functions */ - MBFieldTest = MBFieldTest_c; + (((struct global_all_dll *)Dll::Tls())->MBFieldTest) = MBFieldTest_c; /* Image interpolation related functions */ - interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_c; - interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_c; - interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_c; - - interpolate16x16_lowpass_h = interpolate16x16_lowpass_h_c; - interpolate16x16_lowpass_v = interpolate16x16_lowpass_v_c; - interpolate16x16_lowpass_hv = interpolate16x16_lowpass_hv_c; - - interpolate8x8_lowpass_h = interpolate8x8_lowpass_h_c; - interpolate8x8_lowpass_v = interpolate8x8_lowpass_v_c; - interpolate8x8_lowpass_hv = interpolate8x8_lowpass_hv_c; + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_halfpel_h) = interpolate8x8_halfpel_h_c; + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_halfpel_v) = interpolate8x8_halfpel_v_c; + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_halfpel_hv) = interpolate8x8_halfpel_hv_c; + + (((struct global_all_dll *)Dll::Tls())->interpolate16x16_lowpass_h) = interpolate16x16_lowpass_h_c; + (((struct global_all_dll *)Dll::Tls())->interpolate16x16_lowpass_v) = interpolate16x16_lowpass_v_c; + (((struct global_all_dll *)Dll::Tls())->interpolate16x16_lowpass_hv) = interpolate16x16_lowpass_hv_c; + + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_lowpass_h) = interpolate8x8_lowpass_h_c; + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_lowpass_v) = interpolate8x8_lowpass_v_c; + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_lowpass_hv) = interpolate8x8_lowpass_hv_c; - interpolate8x8_6tap_lowpass_h = interpolate8x8_6tap_lowpass_h_c; - interpolate8x8_6tap_lowpass_v = interpolate8x8_6tap_lowpass_v_c; + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_6tap_lowpass_h) = interpolate8x8_6tap_lowpass_h_c; + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_6tap_lowpass_v) = interpolate8x8_6tap_lowpass_v_c; - interpolate8x8_avg2 = interpolate8x8_avg2_c; - interpolate8x8_avg4 = interpolate8x8_avg4_c; + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_avg2) = interpolate8x8_avg2_c; + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_avg4) = interpolate8x8_avg4_c; /* reduced resolution */ - copy_upsampled_8x8_16to8 = xvid_Copy_Upsampled_8x8_16To8_C; - add_upsampled_8x8_16to8 = xvid_Add_Upsampled_8x8_16To8_C; - vfilter_31 = xvid_VFilter_31_C; - hfilter_31 = xvid_HFilter_31_C; - filter_18x18_to_8x8 = xvid_Filter_18x18_To_8x8_C; - filter_diff_18x18_to_8x8 = xvid_Filter_Diff_18x18_To_8x8_C; + (((struct global_all_dll *)Dll::Tls())->copy_upsampled_8x8_16to8) = xvid_Copy_Upsampled_8x8_16To8_C; + (((struct global_all_dll *)Dll::Tls())->add_upsampled_8x8_16to8) = xvid_Add_Upsampled_8x8_16To8_C; + (((struct global_all_dll *)Dll::Tls())->vfilter_31) = xvid_VFilter_31_C; + (((struct global_all_dll *)Dll::Tls())->hfilter_31) = xvid_HFilter_31_C; + (((struct global_all_dll *)Dll::Tls())->filter_18x18_to_8x8) = xvid_Filter_18x18_To_8x8_C; + (((struct global_all_dll *)Dll::Tls())->filter_diff_18x18_to_8x8) = xvid_Filter_Diff_18x18_To_8x8_C; /* Initialize internal colorspace transformation tables */ colorspace_init(); /* All colorspace transformation functions User Format->YV12 */ - yv12_to_yv12 = yv12_to_yv12_c; - rgb555_to_yv12 = rgb555_to_yv12_c; - rgb565_to_yv12 = rgb565_to_yv12_c; - rgb444_to_yv12 = rgb444_to_yv12_c; - bgr_to_yv12 = bgr_to_yv12_c; - bgra_to_yv12 = bgra_to_yv12_c; - abgr_to_yv12 = abgr_to_yv12_c; - rgba_to_yv12 = rgba_to_yv12_c; - argb_to_yv12 = argb_to_yv12_c; - yuyv_to_yv12 = yuyv_to_yv12_c; - uyvy_to_yv12 = uyvy_to_yv12_c; - - rgb555i_to_yv12 = rgb555i_to_yv12_c; - rgb565i_to_yv12 = rgb565i_to_yv12_c; - bgri_to_yv12 = bgri_to_yv12_c; - bgrai_to_yv12 = bgrai_to_yv12_c; - abgri_to_yv12 = abgri_to_yv12_c; - rgbai_to_yv12 = rgbai_to_yv12_c; - argbi_to_yv12 = argbi_to_yv12_c; - yuyvi_to_yv12 = yuyvi_to_yv12_c; - uyvyi_to_yv12 = uyvyi_to_yv12_c; + (((struct global_all_dll *)Dll::Tls())->yv12_to_yv12) = yv12_to_yv12_c; + (((struct global_all_dll *)Dll::Tls())->rgb555_to_yv12) = rgb555_to_yv12_c; + (((struct global_all_dll *)Dll::Tls())->rgb565_to_yv12) = rgb565_to_yv12_c; + (((struct global_all_dll *)Dll::Tls())->rgb444_to_yv12) = rgb444_to_yv12_c; + (((struct global_all_dll *)Dll::Tls())->bgr_to_yv12) = bgr_to_yv12_c; + (((struct global_all_dll *)Dll::Tls())->bgra_to_yv12) = bgra_to_yv12_c; + (((struct global_all_dll *)Dll::Tls())->abgr_to_yv12) = abgr_to_yv12_c; + (((struct global_all_dll *)Dll::Tls())->rgba_to_yv12) = rgba_to_yv12_c; + (((struct global_all_dll *)Dll::Tls())->argb_to_yv12) = argb_to_yv12_c; + (((struct global_all_dll *)Dll::Tls())->yuyv_to_yv12) = yuyv_to_yv12_c; + (((struct global_all_dll *)Dll::Tls())->uyvy_to_yv12) = uyvy_to_yv12_c; + + (((struct global_all_dll *)Dll::Tls())->rgb555i_to_yv12) = rgb555i_to_yv12_c; + (((struct global_all_dll *)Dll::Tls())->rgb565i_to_yv12) = rgb565i_to_yv12_c; + (((struct global_all_dll *)Dll::Tls())->bgri_to_yv12) = bgri_to_yv12_c; + (((struct global_all_dll *)Dll::Tls())->bgrai_to_yv12) = bgrai_to_yv12_c; + (((struct global_all_dll *)Dll::Tls())->abgri_to_yv12) = abgri_to_yv12_c; + (((struct global_all_dll *)Dll::Tls())->rgbai_to_yv12) = rgbai_to_yv12_c; + (((struct global_all_dll *)Dll::Tls())->argbi_to_yv12) = argbi_to_yv12_c; + (((struct global_all_dll *)Dll::Tls())->yuyvi_to_yv12) = yuyvi_to_yv12_c; + (((struct global_all_dll *)Dll::Tls())->uyvyi_to_yv12) = uyvyi_to_yv12_c; /* All colorspace transformation functions YV12->User format */ - yv12_to_rgb555 = yv12_to_rgb555_c; - yv12_to_rgb565 = yv12_to_rgb565_c; - yv12_to_rgb444 = yv12_to_rgb444_c; - yv12_to_bgr = yv12_to_bgr_c; - yv12_to_bgra = yv12_to_bgra_c; - yv12_to_abgr = yv12_to_abgr_c; - yv12_to_rgba = yv12_to_rgba_c; - yv12_to_argb = yv12_to_argb_c; - yv12_to_yuyv = yv12_to_yuyv_c; - yv12_to_uyvy = yv12_to_uyvy_c; - - yv12_to_rgb555i = yv12_to_rgb555i_c; - yv12_to_rgb565i = yv12_to_rgb565i_c; - yv12_to_bgri = yv12_to_bgri_c; - yv12_to_bgrai = yv12_to_bgrai_c; - yv12_to_abgri = yv12_to_abgri_c; - yv12_to_rgbai = yv12_to_rgbai_c; - yv12_to_argbi = yv12_to_argbi_c; - yv12_to_yuyvi = yv12_to_yuyvi_c; - yv12_to_uyvyi = yv12_to_uyvyi_c; + (((struct global_all_dll *)Dll::Tls())->yv12_to_rgb555) = yv12_to_rgb555_c; + (((struct global_all_dll *)Dll::Tls())->yv12_to_rgb565) = yv12_to_rgb565_c; + (((struct global_all_dll *)Dll::Tls())->yv12_to_rgb444) = yv12_to_rgb444_c; + (((struct global_all_dll *)Dll::Tls())->yv12_to_bgr) = yv12_to_bgr_c; + (((struct global_all_dll *)Dll::Tls())->yv12_to_bgra) = yv12_to_bgra_c; + (((struct global_all_dll *)Dll::Tls())->yv12_to_abgr) = yv12_to_abgr_c; + (((struct global_all_dll *)Dll::Tls())->yv12_to_rgba) = yv12_to_rgba_c; + (((struct global_all_dll *)Dll::Tls())->yv12_to_argb) = yv12_to_argb_c; + (((struct global_all_dll *)Dll::Tls())->yv12_to_yuyv) = yv12_to_yuyv_c; + (((struct global_all_dll *)Dll::Tls())->yv12_to_uyvy) = yv12_to_uyvy_c; + + (((struct global_all_dll *)Dll::Tls())->yv12_to_rgb555i) = yv12_to_rgb555i_c; + (((struct global_all_dll *)Dll::Tls())->yv12_to_rgb565i) = yv12_to_rgb565i_c; + (((struct global_all_dll *)Dll::Tls())->yv12_to_bgri) = yv12_to_bgri_c; + (((struct global_all_dll *)Dll::Tls())->yv12_to_bgrai) = yv12_to_bgrai_c; + (((struct global_all_dll *)Dll::Tls())->yv12_to_abgri) = yv12_to_abgri_c; + (((struct global_all_dll *)Dll::Tls())->yv12_to_rgbai) = yv12_to_rgbai_c; + (((struct global_all_dll *)Dll::Tls())->yv12_to_argbi) = yv12_to_argbi_c; + (((struct global_all_dll *)Dll::Tls())->yv12_to_yuyvi) = yv12_to_yuyvi_c; + (((struct global_all_dll *)Dll::Tls())->yv12_to_uyvyi) = yv12_to_uyvyi_c; /* Functions used in motion estimation algorithms */ - calc_cbp = calc_cbp_c; - sad16 = sad16_c; - sad8 = sad8_c; - sad16bi = sad16bi_c; - sad8bi = sad8bi_c; - dev16 = dev16_c; - sad16v = sad16v_c; - sse8_16bit = sse8_16bit_c; + (((struct global_all_dll *)Dll::Tls())->calc_cbp) = calc_cbp_c; + (((struct global_all_dll *)Dll::Tls())->sad16) = sad16_c; + (((struct global_all_dll *)Dll::Tls())->sad8) = sad8_c; + (((struct global_all_dll *)Dll::Tls())->sad16bi) = sad16bi_c; + (((struct global_all_dll *)Dll::Tls())->sad8bi) = sad8bi_c; + (((struct global_all_dll *)Dll::Tls())->dev16) = dev16_c; + (((struct global_all_dll *)Dll::Tls())->sad16v) = sad16v_c; + (((struct global_all_dll *)Dll::Tls())->sse8_16bit) = sse8_16bit_c; #if defined(ARCH_IS_IA32) if ((cpu_flags & XVID_CPU_ASM)) { - vfilter_31 = xvid_VFilter_31_x86; - hfilter_31 = xvid_HFilter_31_x86; + (((struct global_all_dll *)Dll::Tls())->vfilter_31) = xvid_VFilter_31_x86; + (((struct global_all_dll *)Dll::Tls())->hfilter_31) = xvid_HFilter_31_x86; } if ((cpu_flags & XVID_CPU_MMX) || (cpu_flags & XVID_CPU_MMXEXT) || @@ -307,255 +281,255 @@ (cpu_flags & XVID_CPU_SSE) || (cpu_flags & XVID_CPU_SSE2)) { /* Restore FPU context : emms_c is a nop functions */ - emms = emms_mmx; + (((struct global_all_dll *)Dll::Tls())->emms) = emms_mmx; } if ((cpu_flags & XVID_CPU_MMX)) { /* Forward and Inverse Discrete Cosine Transformation functions */ - fdct = fdct_mmx_skal; - idct = idct_mmx; + (((struct global_all_dll *)Dll::Tls())->fdct) = fdct_mmx_skal; + (((struct global_all_dll *)Dll::Tls())->idct) = idct_mmx; /* Qpel stuff */ - xvid_QP_Funcs = &xvid_QP_Funcs_mmx; - xvid_QP_Add_Funcs = &xvid_QP_Add_Funcs_mmx; + (((struct global_all_dll *)Dll::Tls())->xvid_QP_Funcs) = &xvid_QP_Funcs_mmx; + (((struct global_all_dll *)Dll::Tls())->xvid_QP_Add_Funcs) = &xvid_QP_Add_Funcs_mmx; /* Quantization related functions */ - quant_h263_intra = quant_h263_intra_mmx; - quant_h263_inter = quant_h263_inter_mmx; - dequant_h263_intra = dequant_h263_intra_mmx; - dequant_h263_inter = dequant_h263_inter_mmx; - - quant_mpeg_intra = quant_mpeg_intra_mmx; - quant_mpeg_inter = quant_mpeg_inter_mmx; - dequant_mpeg_intra = dequant_mpeg_intra_mmx; - dequant_mpeg_inter = dequant_mpeg_inter_mmx; + (((struct global_all_dll *)Dll::Tls())->quant_h263_intra) = quant_h263_intra_mmx; + (((struct global_all_dll *)Dll::Tls())->quant_h263_inter) = quant_h263_inter_mmx; + (((struct global_all_dll *)Dll::Tls())->dequant_h263_intra) = dequant_h263_intra_mmx; + (((struct global_all_dll *)Dll::Tls())->dequant_h263_inter) = dequant_h263_inter_mmx; + + (((struct global_all_dll *)Dll::Tls())->quant_mpeg_intra) = quant_mpeg_intra_mmx; + (((struct global_all_dll *)Dll::Tls())->quant_mpeg_inter) = quant_mpeg_inter_mmx; + (((struct global_all_dll *)Dll::Tls())->dequant_mpeg_intra) = dequant_mpeg_intra_mmx; + (((struct global_all_dll *)Dll::Tls())->dequant_mpeg_inter) = dequant_mpeg_inter_mmx; /* Block related functions */ - transfer_8to16copy = transfer_8to16copy_mmx; - transfer_16to8copy = transfer_16to8copy_mmx; - transfer_8to16sub = transfer_8to16sub_mmx; - transfer_8to16subro = transfer_8to16subro_mmx; - transfer_8to16sub2 = transfer_8to16sub2_mmx; - transfer_16to8add = transfer_16to8add_mmx; - transfer8x8_copy = transfer8x8_copy_mmx; + (((struct global_all_dll *)Dll::Tls())->transfer_8to16copy) = transfer_8to16copy_mmx; + (((struct global_all_dll *)Dll::Tls())->transfer_16to8copy) = transfer_16to8copy_mmx; + (((struct global_all_dll *)Dll::Tls())->transfer_8to16sub) = transfer_8to16sub_mmx; + (((struct global_all_dll *)Dll::Tls())->transfer_8to16subro) = transfer_8to16subro_mmx; + (((struct global_all_dll *)Dll::Tls())->transfer_8to16sub2) = transfer_8to16sub2_mmx; + (((struct global_all_dll *)Dll::Tls())->transfer_16to8add) = transfer_16to8add_mmx; + (((struct global_all_dll *)Dll::Tls())->transfer8x8_copy) = transfer8x8_copy_mmx; /* Interlacing Functions */ - MBFieldTest = MBFieldTest_mmx; + (((struct global_all_dll *)Dll::Tls())->MBFieldTest) = MBFieldTest_mmx; /* Image Interpolation related functions */ - interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_mmx; - interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_mmx; - interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_mmx; + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_halfpel_h) = interpolate8x8_halfpel_h_mmx; + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_halfpel_v) = interpolate8x8_halfpel_v_mmx; + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_halfpel_hv) = interpolate8x8_halfpel_hv_mmx; - interpolate8x8_6tap_lowpass_h = interpolate8x8_6tap_lowpass_h_mmx; - interpolate8x8_6tap_lowpass_v = interpolate8x8_6tap_lowpass_v_mmx; + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_6tap_lowpass_h) = interpolate8x8_6tap_lowpass_h_mmx; + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_6tap_lowpass_v) = interpolate8x8_6tap_lowpass_v_mmx; - interpolate8x8_avg2 = interpolate8x8_avg2_mmx; - interpolate8x8_avg4 = interpolate8x8_avg4_mmx; + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_avg2) = interpolate8x8_avg2_mmx; + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_avg4) = interpolate8x8_avg4_mmx; /* reduced resolution */ - copy_upsampled_8x8_16to8 = xvid_Copy_Upsampled_8x8_16To8_mmx; - add_upsampled_8x8_16to8 = xvid_Add_Upsampled_8x8_16To8_mmx; - hfilter_31 = xvid_HFilter_31_mmx; - filter_18x18_to_8x8 = xvid_Filter_18x18_To_8x8_mmx; - filter_diff_18x18_to_8x8 = xvid_Filter_Diff_18x18_To_8x8_mmx; + (((struct global_all_dll *)Dll::Tls())->copy_upsampled_8x8_16to8) = xvid_Copy_Upsampled_8x8_16To8_mmx; + (((struct global_all_dll *)Dll::Tls())->add_upsampled_8x8_16to8) = xvid_Add_Upsampled_8x8_16To8_mmx; + (((struct global_all_dll *)Dll::Tls())->hfilter_31) = xvid_HFilter_31_mmx; + (((struct global_all_dll *)Dll::Tls())->filter_18x18_to_8x8) = xvid_Filter_18x18_To_8x8_mmx; + (((struct global_all_dll *)Dll::Tls())->filter_diff_18x18_to_8x8) = xvid_Filter_Diff_18x18_To_8x8_mmx; /* image input xxx_to_yv12 related functions */ - yv12_to_yv12 = yv12_to_yv12_mmx; - bgr_to_yv12 = bgr_to_yv12_mmx; - bgra_to_yv12 = bgra_to_yv12_mmx; - yuyv_to_yv12 = yuyv_to_yv12_mmx; - uyvy_to_yv12 = uyvy_to_yv12_mmx; + (((struct global_all_dll *)Dll::Tls())->yv12_to_yv12) = yv12_to_yv12_mmx; + (((struct global_all_dll *)Dll::Tls())->bgr_to_yv12) = bgr_to_yv12_mmx; + (((struct global_all_dll *)Dll::Tls())->bgra_to_yv12) = bgra_to_yv12_mmx; + (((struct global_all_dll *)Dll::Tls())->yuyv_to_yv12) = yuyv_to_yv12_mmx; + (((struct global_all_dll *)Dll::Tls())->uyvy_to_yv12) = uyvy_to_yv12_mmx; /* image output yv12_to_xxx related functions */ - yv12_to_bgr = yv12_to_bgr_mmx; - yv12_to_bgra = yv12_to_bgra_mmx; - yv12_to_yuyv = yv12_to_yuyv_mmx; - yv12_to_uyvy = yv12_to_uyvy_mmx; + (((struct global_all_dll *)Dll::Tls())->yv12_to_bgr) = yv12_to_bgr_mmx; + (((struct global_all_dll *)Dll::Tls())->yv12_to_bgra) = yv12_to_bgra_mmx; + (((struct global_all_dll *)Dll::Tls())->yv12_to_yuyv) = yv12_to_yuyv_mmx; + (((struct global_all_dll *)Dll::Tls())->yv12_to_uyvy) = yv12_to_uyvy_mmx; - yv12_to_yuyvi = yv12_to_yuyvi_mmx; - yv12_to_uyvyi = yv12_to_uyvyi_mmx; + (((struct global_all_dll *)Dll::Tls())->yv12_to_yuyvi) = yv12_to_yuyvi_mmx; + (((struct global_all_dll *)Dll::Tls())->yv12_to_uyvyi) = yv12_to_uyvyi_mmx; /* Motion estimation related functions */ - calc_cbp = calc_cbp_mmx; - sad16 = sad16_mmx; - sad8 = sad8_mmx; - sad16bi = sad16bi_mmx; - sad8bi = sad8bi_mmx; - dev16 = dev16_mmx; - sad16v = sad16v_mmx; - sse8_16bit = sse8_16bit_mmx; + (((struct global_all_dll *)Dll::Tls())->calc_cbp) = calc_cbp_mmx; + (((struct global_all_dll *)Dll::Tls())->sad16) = sad16_mmx; + (((struct global_all_dll *)Dll::Tls())->sad8) = sad8_mmx; + (((struct global_all_dll *)Dll::Tls())->sad16bi) = sad16bi_mmx; + (((struct global_all_dll *)Dll::Tls())->sad8bi) = sad8bi_mmx; + (((struct global_all_dll *)Dll::Tls())->dev16) = dev16_mmx; + (((struct global_all_dll *)Dll::Tls())->sad16v) = sad16v_mmx; + (((struct global_all_dll *)Dll::Tls())->sse8_16bit) = sse8_16bit_mmx; } /* these 3dnow functions are faster than mmx, but slower than xmm. */ if ((cpu_flags & XVID_CPU_3DNOW)) { - emms = emms_3dn; + (((struct global_all_dll *)Dll::Tls())->emms) = emms_3dn; /* ME functions */ - sad16bi = sad16bi_3dn; - sad8bi = sad8bi_3dn; + (((struct global_all_dll *)Dll::Tls())->sad16bi) = sad16bi_3dn; + (((struct global_all_dll *)Dll::Tls())->sad8bi) = sad8bi_3dn; - yuyv_to_yv12 = yuyv_to_yv12_3dn; - uyvy_to_yv12 = uyvy_to_yv12_3dn; + (((struct global_all_dll *)Dll::Tls())->yuyv_to_yv12) = yuyv_to_yv12_3dn; + (((struct global_all_dll *)Dll::Tls())->uyvy_to_yv12) = uyvy_to_yv12_3dn; } if ((cpu_flags & XVID_CPU_MMXEXT)) { /* DCT */ - fdct = fdct_xmm_skal; - idct = idct_xmm; + (((struct global_all_dll *)Dll::Tls())->fdct) = fdct_xmm_skal; + (((struct global_all_dll *)Dll::Tls())->idct) = idct_xmm; /* Interpolation */ - interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_xmm; - interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_xmm; - interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_xmm; + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_halfpel_h) = interpolate8x8_halfpel_h_xmm; + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_halfpel_v) = interpolate8x8_halfpel_v_xmm; + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_halfpel_hv) = interpolate8x8_halfpel_hv_xmm; /* reduced resolution */ - copy_upsampled_8x8_16to8 = xvid_Copy_Upsampled_8x8_16To8_xmm; - add_upsampled_8x8_16to8 = xvid_Add_Upsampled_8x8_16To8_xmm; + (((struct global_all_dll *)Dll::Tls())->copy_upsampled_8x8_16to8) = xvid_Copy_Upsampled_8x8_16To8_xmm; + (((struct global_all_dll *)Dll::Tls())->add_upsampled_8x8_16to8) = xvid_Add_Upsampled_8x8_16To8_xmm; /* Quantization */ - quant_mpeg_intra = quant_mpeg_intra_xmm; - quant_mpeg_inter = quant_mpeg_inter_xmm; + (((struct global_all_dll *)Dll::Tls())->quant_mpeg_intra) = quant_mpeg_intra_xmm; + (((struct global_all_dll *)Dll::Tls())->quant_mpeg_inter) = quant_mpeg_inter_xmm; - dequant_h263_intra = dequant_h263_intra_xmm; - dequant_h263_inter = dequant_h263_inter_xmm; + (((struct global_all_dll *)Dll::Tls())->dequant_h263_intra) = dequant_h263_intra_xmm; + (((struct global_all_dll *)Dll::Tls())->dequant_h263_inter) = dequant_h263_inter_xmm; /* Buffer transfer */ - transfer_8to16sub2 = transfer_8to16sub2_xmm; + (((struct global_all_dll *)Dll::Tls())->transfer_8to16sub2) = transfer_8to16sub2_xmm; /* Colorspace transformation */ - yv12_to_yv12 = yv12_to_yv12_xmm; - yuyv_to_yv12 = yuyv_to_yv12_xmm; - uyvy_to_yv12 = uyvy_to_yv12_xmm; + (((struct global_all_dll *)Dll::Tls())->yv12_to_yv12) = yv12_to_yv12_xmm; + (((struct global_all_dll *)Dll::Tls())->yuyv_to_yv12) = yuyv_to_yv12_xmm; + (((struct global_all_dll *)Dll::Tls())->uyvy_to_yv12) = uyvy_to_yv12_xmm; /* ME functions */ - sad16 = sad16_xmm; - sad8 = sad8_xmm; - sad16bi = sad16bi_xmm; - sad8bi = sad8bi_xmm; - dev16 = dev16_xmm; - sad16v = sad16v_xmm; + (((struct global_all_dll *)Dll::Tls())->sad16) = sad16_xmm; + (((struct global_all_dll *)Dll::Tls())->sad8) = sad8_xmm; + (((struct global_all_dll *)Dll::Tls())->sad16bi) = sad16bi_xmm; + (((struct global_all_dll *)Dll::Tls())->sad8bi) = sad8bi_xmm; + (((struct global_all_dll *)Dll::Tls())->dev16) = dev16_xmm; + (((struct global_all_dll *)Dll::Tls())->sad16v) = sad16v_xmm; } if ((cpu_flags & XVID_CPU_3DNOW)) { /* Interpolation */ - interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_3dn; - interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_3dn; - interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_3dn; + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_halfpel_h) = interpolate8x8_halfpel_h_3dn; + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_halfpel_v) = interpolate8x8_halfpel_v_3dn; + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_halfpel_hv) = interpolate8x8_halfpel_hv_3dn; } if ((cpu_flags & XVID_CPU_3DNOWEXT)) { /* Buffer transfer */ - transfer_8to16copy = transfer_8to16copy_3dne; - transfer_16to8copy = transfer_16to8copy_3dne; - transfer_8to16sub = transfer_8to16sub_3dne; - transfer_8to16subro = transfer_8to16subro_3dne; - transfer_16to8add = transfer_16to8add_3dne; - transfer8x8_copy = transfer8x8_copy_3dne; + (((struct global_all_dll *)Dll::Tls())->transfer_8to16copy) = transfer_8to16copy_3dne; + (((struct global_all_dll *)Dll::Tls())->transfer_16to8copy) = transfer_16to8copy_3dne; + (((struct global_all_dll *)Dll::Tls())->transfer_8to16sub) = transfer_8to16sub_3dne; + (((struct global_all_dll *)Dll::Tls())->transfer_8to16subro) = transfer_8to16subro_3dne; + (((struct global_all_dll *)Dll::Tls())->transfer_16to8add) = transfer_16to8add_3dne; + (((struct global_all_dll *)Dll::Tls())->transfer8x8_copy) = transfer8x8_copy_3dne; if ((cpu_flags & XVID_CPU_MMXEXT)) { /* Inverse DCT */ - idct = idct_3dne; + (((struct global_all_dll *)Dll::Tls())->idct) = idct_3dne; /* Buffer transfer */ - transfer_8to16sub2 = transfer_8to16sub2_3dne; + (((struct global_all_dll *)Dll::Tls())->transfer_8to16sub2) = transfer_8to16sub2_3dne; /* Interpolation */ - interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_3dne; - interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_3dne; - interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_3dne; + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_halfpel_h) = interpolate8x8_halfpel_h_3dne; + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_halfpel_v) = interpolate8x8_halfpel_v_3dne; + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_halfpel_hv) = interpolate8x8_halfpel_hv_3dne; /* Quantization */ - quant_h263_intra = quant_h263_intra_3dne; /* cmov only */ - quant_h263_inter = quant_h263_inter_3dne; - dequant_mpeg_intra = dequant_mpeg_intra_3dne; /* cmov only */ - dequant_mpeg_inter = dequant_mpeg_inter_3dne; - dequant_h263_intra = dequant_h263_intra_3dne; - dequant_h263_inter = dequant_h263_inter_3dne; + (((struct global_all_dll *)Dll::Tls())->quant_h263_intra) = quant_h263_intra_3dne; /* cmov only */ + (((struct global_all_dll *)Dll::Tls())->quant_h263_inter) = quant_h263_inter_3dne; + (((struct global_all_dll *)Dll::Tls())->dequant_mpeg_intra) = dequant_mpeg_intra_3dne; /* cmov only */ + (((struct global_all_dll *)Dll::Tls())->dequant_mpeg_inter) = dequant_mpeg_inter_3dne; + (((struct global_all_dll *)Dll::Tls())->dequant_h263_intra) = dequant_h263_intra_3dne; + (((struct global_all_dll *)Dll::Tls())->dequant_h263_inter) = dequant_h263_inter_3dne; /* ME functions */ - calc_cbp = calc_cbp_3dne; + (((struct global_all_dll *)Dll::Tls())->calc_cbp) = calc_cbp_3dne; - sad16 = sad16_3dne; - sad8 = sad8_3dne; - sad16bi = sad16bi_3dne; - sad8bi = sad8bi_3dne; - dev16 = dev16_3dne; + (((struct global_all_dll *)Dll::Tls())->sad16) = sad16_3dne; + (((struct global_all_dll *)Dll::Tls())->sad8) = sad8_3dne; + (((struct global_all_dll *)Dll::Tls())->sad16bi) = sad16bi_3dne; + (((struct global_all_dll *)Dll::Tls())->sad8bi) = sad8bi_3dne; + (((struct global_all_dll *)Dll::Tls())->dev16) = dev16_3dne; } } if ((cpu_flags & XVID_CPU_SSE2)) { - calc_cbp = calc_cbp_sse2; + (((struct global_all_dll *)Dll::Tls())->calc_cbp) = calc_cbp_sse2; /* Quantization */ - quant_h263_intra = quant_h263_intra_sse2; - quant_h263_inter = quant_h263_inter_sse2; - dequant_h263_intra = dequant_h263_intra_sse2; - dequant_h263_inter = dequant_h263_inter_sse2; + (((struct global_all_dll *)Dll::Tls())->quant_h263_intra) = quant_h263_intra_sse2; + (((struct global_all_dll *)Dll::Tls())->quant_h263_inter) = quant_h263_inter_sse2; + (((struct global_all_dll *)Dll::Tls())->dequant_h263_intra) = dequant_h263_intra_sse2; + (((struct global_all_dll *)Dll::Tls())->dequant_h263_inter) = dequant_h263_inter_sse2; /* SAD operators */ - sad16 = sad16_sse2; - dev16 = dev16_sse2; + (((struct global_all_dll *)Dll::Tls())->sad16) = sad16_sse2; + (((struct global_all_dll *)Dll::Tls())->dev16) = dev16_sse2; /* DCT operators * no iDCT because it's not "Walken matching" */ - fdct = fdct_sse2_skal; + (((struct global_all_dll *)Dll::Tls())->fdct) = fdct_sse2_skal; } #endif /* ARCH_IS_IA32 */ #if defined(ARCH_IS_IA64) if ((cpu_flags & XVID_CPU_ASM)) { /* use assembler routines? */ idct_ia64_init(); - fdct = fdct_ia64; - idct = idct_ia64; /*not yet working, crashes */ - interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_ia64; - interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_ia64; - interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_ia64; - sad16 = sad16_ia64; - sad16bi = sad16bi_ia64; - sad8 = sad8_ia64; - dev16 = dev16_ia64; + (((struct global_all_dll *)Dll::Tls())->fdct) = fdct_ia64; + (((struct global_all_dll *)Dll::Tls())->idct) = idct_ia64; /*not yet working, crashes */ + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_halfpel_h) = interpolate8x8_halfpel_h_ia64; + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_halfpel_v) = interpolate8x8_halfpel_v_ia64; + (((struct global_all_dll *)Dll::Tls())->interpolate8x8_halfpel_hv) = interpolate8x8_halfpel_hv_ia64; + (((struct global_all_dll *)Dll::Tls())->sad16) = sad16_ia64; + (((struct global_all_dll *)Dll::Tls())->sad16bi) = sad16bi_ia64; + (((struct global_all_dll *)Dll::Tls())->sad8) = sad8_ia64; + (((struct global_all_dll *)Dll::Tls())->dev16) = dev16_ia64; /* Halfpel8_Refine = Halfpel8_Refine_ia64; */ - quant_h263_intra = quant_h263_intra_ia64; - quant_h263_inter = quant_h263_inter_ia64; - dequant_h263_intra = dequant_h263_intra_ia64; - dequant_h263_inter = dequant_h263_inter_ia64; - transfer_8to16copy = transfer_8to16copy_ia64; - transfer_16to8copy = transfer_16to8copy_ia64; - transfer_8to16sub = transfer_8to16sub_ia64; - transfer_8to16sub2 = transfer_8to16sub2_ia64; - transfer_16to8add = transfer_16to8add_ia64; - transfer8x8_copy = transfer8x8_copy_ia64; + (((struct global_all_dll *)Dll::Tls())->quant_h263_intra) = quant_h263_intra_ia64; + (((struct global_all_dll *)Dll::Tls())->quant_h263_inter) = quant_h263_inter_ia64; + (((struct global_all_dll *)Dll::Tls())->dequant_h263_intra) = dequant_h263_intra_ia64; + (((struct global_all_dll *)Dll::Tls())->dequant_h263_inter) = dequant_h263_inter_ia64; + (((struct global_all_dll *)Dll::Tls())->transfer_8to16copy) = transfer_8to16copy_ia64; + (((struct global_all_dll *)Dll::Tls())->transfer_16to8copy) = transfer_16to8copy_ia64; + (((struct global_all_dll *)Dll::Tls())->transfer_8to16sub) = transfer_8to16sub_ia64; + (((struct global_all_dll *)Dll::Tls())->transfer_8to16sub2) = transfer_8to16sub2_ia64; + (((struct global_all_dll *)Dll::Tls())->transfer_16to8add) = transfer_16to8add_ia64; + (((struct global_all_dll *)Dll::Tls())->transfer8x8_copy) = transfer8x8_copy_ia64; } #endif #if defined(ARCH_IS_PPC) if ((cpu_flags & XVID_CPU_ASM)) { - calc_cbp = calc_cbp_ppc; + (((struct global_all_dll *)Dll::Tls())->calc_cbp) = calc_cbp_ppc; } if ((cpu_flags & XVID_CPU_ALTIVEC)) { - calc_cbp = calc_cbp_altivec; - fdct = fdct_altivec; - idct = idct_altivec; - sadInit = sadInit_altivec; - sad16 = sad16_altivec; - sad8 = sad8_altivec; - dev16 = dev16_altivec; + (((struct global_all_dll *)Dll::Tls())->calc_cbp) = calc_cbp_altivec; + (((struct global_all_dll *)Dll::Tls())->fdct) = fdct_altivec; + (((struct global_all_dll *)Dll::Tls())->idct) = idct_altivec; + (((struct global_all_dll *)Dll::Tls())->sadInit) = sadInit_altivec; + (((struct global_all_dll *)Dll::Tls())->sad16) = sad16_altivec; + (((struct global_all_dll *)Dll::Tls())->sad8) = sad8_altivec; + (((struct global_all_dll *)Dll::Tls())->dev16) = dev16_altivec; } #endif #if defined(_DEBUG) - xvid_debug = init->debug; + (((struct global_all_dll *)Dll::Tls())->xvid_debug) = init->debug; #endif return 0; @@ -582,6 +556,9 @@ } +#if 0 + const int flip1 = (convert->input.colorspace & XVID_CSP_VFLIP) ^ (convert->output.colorspace & XVID_CSP_VFLIP); +#endif static int xvid_gbl_convert(xvid_gbl_convert_t* convert) { @@ -594,9 +571,6 @@ if (XVID_VERSION_MAJOR(convert->version) != 1) /* v1.x.x */ return XVID_ERR_VERSION; -#if 0 - const int flip1 = (convert->input.colorspace & XVID_CSP_VFLIP) ^ (convert->output.colorspace & XVID_CSP_VFLIP); -#endif width = convert->width; height = convert->height; width2 = convert->width/2; @@ -618,7 +592,7 @@ } - emms(); + (((struct global_all_dll *)Dll::Tls())->emms)(); return 0; }