/* transfrm.c,  前向/后向變換*/
#include <stdio.h>#include <math.h>#include "global.h"static short PACKED_0[4] = { 0, 0, 0, 0 };
static short PACKED_128[4] = { 128, 128, 128, 128 };
static unsigned char PACKED_128B[8] = { 128, 128, 128, 128, 128, 128, 128, 128 };

static void add_pred(unsigned char *pred, unsigned char *cur,  int lx, short *blk);static void sub_pred(unsigned char *pred, unsigned char *cur,  int lx, short *blk);void transform(pred,cur,mbi,blocks)unsigned char *pred[], *cur[];struct mbinfo *mbi;short blocks[][64];{  int i, j, i1, j1, k, n, cc, offs, lx;  k = 0;  for (j=0; j<height2; j+=16)    for (i=0; i<width; i+=16)    {      for (n=0; n<block_count; n++)      {        cc = (n<4) ? 0 : (n&1)+1; /* color component index */        if (cc==0)        {          /* luminance */          if ((pict_struct==FRAME_PICTURE) && mbi[k].dct_type)          {            /* field DCT */            offs = i + ((n&1)<<3) + width*(j+((n&2)>>1));            lx = width<<1;          }          else          {            /* frame DCT */            offs = i + ((n&1)<<3) + width2*(j+((n&2)<<2));            lx = width2;          }          if (pict_struct==BOTTOM_FIELD)            offs += width;        }        else        {          /* chrominance */          /* scale coordinates */          i1 = (chroma_format==CHROMA444) ? i : i>>1;          j1 = (chroma_format!=CHROMA420) ? j : j>>1;          if ((pict_struct==FRAME_PICTURE) && mbi[k].dct_type              && (chroma_format!=CHROMA420))          {            /* field DCT */            offs = i1 + (n&8) + chrom_width*(j1+((n&2)>>1));            lx = chrom_width<<1;          }          else          {            /* frame DCT */            offs = i1 + (n&8) + chrom_width2*(j1+((n&2)<<2));            lx = chrom_width2;          }          if (pict_struct==BOTTOM_FIELD)            offs += chrom_width;        }        sub_pred(pred[cc]+offs,cur[cc]+offs,lx,blocks[k*block_count+n]);        fdct(blocks[k*block_count+n]);      }      k++;    }}/* 反向變換預測誤差和加預測*/void itransform(pred,cur,mbi,blocks)unsigned char *pred[],*cur[];struct mbinfo *mbi;short blocks[][64];{  int i, j, i1, j1, k, n, cc, offs, lx;

  k = 0;  for (j=0; j<height2; j+=16)    for (i=0; i<width; i+=16)    {      for (n=0; n<block_count; n++)      {        cc = (n<4) ? 0 : (n&1)+1; /* color component index */        if (cc==0)        {          /* luminance */          if ((pict_struct==FRAME_PICTURE) && mbi[k].dct_type)          {            /* field DCT */            offs = i + ((n&1)<<3) + width*(j+((n&2)>>1));            lx = width<<1;          }          else          {            /* frame DCT */            offs = i + ((n&1)<<3) + width2*(j+((n&2)<<2));            lx = width2;          }          if (pict_struct==BOTTOM_FIELD)            offs += width;        }        else        {          /* chrominance */          /* scale coordinates */          i1 = (chroma_format==CHROMA444) ? i : i>>1;          j1 = (chroma_format!=CHROMA420) ? j : j>>1;          if ((pict_struct==FRAME_PICTURE) && mbi[k].dct_type              && (chroma_format!=CHROMA420))          {            /* field DCT */            offs = i1 + (n&8) + chrom_width*(j1+((n&2)>>1));            lx = chrom_width<<1;          }          else          {            /* frame DCT */            offs = i1 + (n&8) + chrom_width2*(j1+((n&2)<<2));            lx = chrom_width2;          }          if (pict_struct==BOTTOM_FIELD)            offs += chrom_width;        }        idct(blocks[k*block_count+n]);        add_pred(pred[cc]+offs,cur[cc]+offs,lx,blocks[k*block_count+n]);      }      k++;    }}/* 加預測和預測誤差*/static void add_pred(pred,cur,lx,blk)unsigned char *pred, *cur;int lx;short *blk;{	int i, j;

	if(cpu_MMX)
	{
		// I do loop unrolling on the inner loop and the outer loop !
		_asm
		{
			mov esi, blk				;// esi = blk
			mov ebx, pred				;// ebx = pred
			mov edi, cur				;// edi = cur
			mov edx, lx					;// edx = lx
			lea ecx, [2*edx]			;// ecx = 2 * lx
			movq mm0, [esi]				;// mm0 = 4 words into esi[0..7]
			movq mm1, [esi+8]			;// mm1 = 4 words into esi[8..15]
			movq mm2, [ebx]				;// mm2 = 4 words into ebx[0..7]
			movq mm3, [esi+16]			;// mm3 = 4 words into esi[16..23]
			movq mm4, [esi+24]			;// mm4 = 4 words into esi[24..31]
			movq mm5, [ebx+edx]			;// mm5 = 4 words into (ebx + edx)[0..7]
			movq mm6, mm2
			movq mm7, mm5
			punpcklbw mm2, PACKED_0		;// unpack the lower 4 bytes into mm2
			punpckhbw mm6, PACKED_0		;// unpack the upper 4 bytes into mm6
			punpcklbw mm5, PACKED_0		;// unpack the lower 4 bytes into mm5
			punpckhbw mm7, PACKED_0		;// unpack the upper 4 bytes into mm7
			paddw mm0, mm2				;// mm0 += mm2
			paddw mm1, mm6				;// mm1 += mm6
			paddw mm3, mm5				;// mm3 += mm5
			paddw mm4, mm7				;// mm4 += mm7
			psubw mm0, PACKED_128		;// convert mm0 from unsigned to signed by substracting 128
			psubw mm1, PACKED_128		;// convert mm1 from unsigned to signed by substracting 128
			psubw mm3, PACKED_128		;// convert mm3 from unsigned to signed by substracting 128
			psubw mm4, PACKED_128		;// convert mm4 from unsigned to signed by substracting 128
			packsswb mm0, mm1			;// pack mm0 and mm1 to get only 8 signed bytes into mm0
			packsswb mm3, mm4			;// pack mm3 and mm4 to get only 8 signed bytes into mm3
			paddb mm0, PACKED_128B		;// reverse the effect of convertion to signed
			paddb mm3, PACKED_128B		;// reverse the effect of convertion to signed
			add ebx, ecx				;// ebx += 2 * lx
			movq [edi], mm0				;// store mm0
			movq [edi+edx], mm3			;// store mm3
			movq mm0, [esi+32]			;// mm0 = 4 words into esi[32..39]
			movq mm1, [esi+40]			;// mm1 = 4 words into esi[40..47]
			movq mm2, [ebx]				;// mm2 = 4 words into ebx[0..7]
			movq mm3, [esi+48]			;// mm3 = 4 words into esi[48..55]
			movq mm4, [esi+56]			;// mm4 = 4 words into esi[56..63]
			movq mm5, [ebx+edx]			;// mm5 = 4 words into (ebx + edx)[0..7]
			add edi, ecx				;// edi += 2 * lx
			movq mm6, mm2
			movq mm7, mm5
			punpcklbw mm2, PACKED_0		;// unpack the lower 4 bytes into mm2
			punpckhbw mm6, PACKED_0		;// unpack the upper 4 bytes into mm6
			punpcklbw mm5, PACKED_0		;// unpack the lower 4 bytes into mm5
			punpckhbw mm7, PACKED_0		;// unpack the upper 4 bytes into mm7
			paddw mm0, mm2				;// mm0 += mm2
			paddw mm1, mm6				;// mm1 += mm6
			paddw mm3, mm5				;// mm3 += mm5
			paddw mm4, mm7				;// mm4 += mm7
			psubw mm0, PACKED_128		;// convert mm0 from unsigned to signed by substracting 128
			psubw mm1, PACKED_128		;// convert mm1 from unsigned to signed by substracting 128
			psubw mm3, PACKED_128		;// convert mm3 from unsigned to signed by substracting 128
			psubw mm4, PACKED_128		;// convert mm4 from unsigned to signed by substracting 128
			packsswb mm0, mm1			;// pack mm0 and mm1 to get only 8 signed bytes into mm0
			packsswb mm3, mm4			;// pack mm3 and mm4 to get only 8 signed bytes into mm3
			paddb mm0, PACKED_128B		;// reverse the effect of convertion to signed
			paddb mm3, PACKED_128B		;// reverse the effect of convertion to signed
			add ebx, ecx				;// ebx += 2 * lx
			movq [edi], mm0				;// store mm0
			movq [edi+edx], mm3			;// store mm3
			movq mm0, [esi+64]			;// mm0 = 4 words into esi[64..71]
			movq mm1, [esi+72]			;// mm1 = 4 words into esi[72..79]
			movq mm2, [ebx]				;// mm2 = 4 words into ebx[0..7]
			movq mm3, [esi+80]			;// mm3 = 4 words into esi[80..87]
			movq mm4, [esi+88]			;// mm4 = 4 words into esi[88..95]
			movq mm5, [ebx+edx]			;// mm5 = 4 words into (ebx + edx)[0..7]
			add edi, ecx				;// edi += 2 * lx
			movq mm6, mm2
			movq mm7, mm5
			punpcklbw mm2, PACKED_0		;// unpack the lower 4 bytes into mm2
			punpckhbw mm6, PACKED_0		;// unpack the upper 4 bytes into mm6
			punpcklbw mm5, PACKED_0		;// unpack the lower 4 bytes into mm5
			punpckhbw mm7, PACKED_0		;// unpack the upper 4 bytes into mm7
			paddw mm0, mm2				;// mm0 += mm2
			paddw mm1, mm6				;// mm1 += mm6
			paddw mm3, mm5				;// mm3 += mm5
			paddw mm4, mm7				;// mm4 += mm7
			psubw mm0, PACKED_128		;// convert mm0 from unsigned to signed by substracting 128
			psubw mm1, PACKED_128		;// convert mm1 from unsigned to signed by substracting 128
			psubw mm3, PACKED_128		;// convert mm3 from unsigned to signed by substracting 128