motion.c

avi2mpg1_src 中包含了mpeg1编码的源程序

unsigned char *blk1,*blk2;
int lx,hx,hy,h;
int distlim;
{
	unsigned char *p1,*p1a,*p2;
	int i,j;
	int s,v;

	if(fastMotionCompensationLevel)
	{
		lx <<= fastMotionCompensationLevel;
		h >>= fastMotionCompensationLevel;
	}

	if (!hx && !hy)
	{
		//针对MMX处理器
		if(cpu_MMX)
		{
			_asm
			{
				mov esi, blk1		;// esi = blk1
				mov edi, blk2		;// edi = blk2
				mov ecx, h			;// ecx = h
				mov edx, lx			;// edx = lx
				mov eax, 0			;// eax = s
dist1__l1:
				movq mm0, [esi]		;// load esi[0..7] into mm0
				movq mm1, [edi]		;// load edi[0..7] into mm1
				movq mm2, [esi+8]	;// load esi[8..15] into mm2
				movq mm3, [edi+8]	;// load edi[8..15] into mm3
				movq mm4, mm0
				movq mm5, mm1
				movq mm6, mm2
				movq mm7, mm3
				psubusb mm0, mm1
				psubusb mm2, mm3
				psubusb mm5, mm4
				psubusb mm7, mm6
				pxor mm1, mm1		;// mm1 = 0
				por mm0, mm5		;// mm0 = abs(esi[0..7] - edi[0..7])
				por mm2, mm7		;// mm2 = abs(esi[8..15] - edi[8..15])
				movq mm4, mm0
				movq mm6, mm2
				punpcklbw mm0, mm1	;// unpack the lower 4 bytes into mm0
				punpcklbw mm2, mm1	;// unpack the lower 4 bytes into mm2
				punpckhbw mm4, mm1	;// unpack the upper 4 bytes into mm4
				punpckhbw mm6, mm1	;// unpack the upper 4 bytes into mm6
				paddw mm0, mm2
				paddw mm4, mm6
				paddw mm0, mm4		;// mm0 += mm2 + mm4 + mm6
				movq mm5, mm0
				punpcklwd mm0, mm1	;// unpack the lower 2 words into mm0
				punpckhwd mm5, mm1	;// unpack the upper 2 words into mm5
				paddd mm0, mm5		;// mm0 += mm5
				movd ebx, mm0		;// load lower dword of mm0 into ebx
				add eax, ebx		;// eax += ebx
				psrlq mm0, 32		;// shift mm0 to get upper dword
				movd ebx, mm0		;// load lower dword of mm0 into ebx
				add eax, ebx		;// eax += ebx

				cmp eax, distlim	;// compare eax with distlim
				jge dist1__l2		;// terminate if eax >= distlim
				add esi, edx		;// esi += edx
				add edi, edx		;// edi += edx
				
				dec ecx				;// decrement ecx
				jnz dist1__l1		;// loop while not zero
dist1__l2:
				mov s, eax			;// s = eax
				emms				;// empty MMX state
			}
		}
		else
		{
			s = 0;
			p1 = blk1;
			p2 = blk2;

			for (j=0; j<h; j++)
			{
				if ((v = p1[0]  - p2[0])<0)  v = -v; s+= v;
				if ((v = p1[1]  - p2[1])<0)  v = -v; s+= v;
				if ((v = p1[2]  - p2[2])<0)  v = -v; s+= v;
				if ((v = p1[3]  - p2[3])<0)  v = -v; s+= v;
				if ((v = p1[4]  - p2[4])<0)  v = -v; s+= v;
				if ((v = p1[5]  - p2[5])<0)  v = -v; s+= v;
				if ((v = p1[6]  - p2[6])<0)  v = -v; s+= v;
				if ((v = p1[7]  - p2[7])<0)  v = -v; s+= v;
				if ((v = p1[8]  - p2[8])<0)  v = -v; s+= v;
				if ((v = p1[9]  - p2[9])<0)  v = -v; s+= v;
				if ((v = p1[10] - p2[10])<0) v = -v; s+= v;
				if ((v = p1[11] - p2[11])<0) v = -v; s+= v;
				if ((v = p1[12] - p2[12])<0) v = -v; s+= v;
				if ((v = p1[13] - p2[13])<0) v = -v; s+= v;
				if ((v = p1[14] - p2[14])<0) v = -v; s+= v;
				if ((v = p1[15] - p2[15])<0) v = -v; s+= v;

				if (s >= distlim)
					break;

				p1+= lx;
				p2+= lx;
			}
		}
	}
	else if (hx && !hy)
	{
		//针对支持3DNow的CPU
		if (cpu_3DNow)
		{
			s = 0;
			p1 = blk1;
			p2 = blk2;

			for (j=0; j<h; j++)
			{
				

				#define femmsInst __asm _emit 0x0F __asm _emit 0x0E
				#define pavgusbmm0mm2 __asm _emit 0x0F __asm _emit 0x0F __asm _emit 0xC2 __asm _emit 0xBF
				#define pavgusbmm1mm3 __asm _emit 0x0F __asm _emit 0x0F __asm _emit 0xCB __asm _emit 0xBF

				__asm
				{
					
					femmsInst
					mov              eax, p1
					movq             mm0,[eax]
					movq             mm1,[eax+8]
					movq             mm2,[eax+1]
					movq             mm3,[eax+9]
					pavgusbmm0mm2
					movq             tmpResult,mm0
					pavgusbmm1mm3
					movq             tmpResult+8,mm1
					femmsInst
				}
				s += calcDist16(tmpResult,p2);

				p1+= lx;
				p2+= lx;
			}
		}
		else if(cpu_MMX)
		{
			_asm
			{
				mov esi, blk1			;// esi = blk1
				mov edi, blk2			;// edi = blk2
				mov ecx, h				;// ecx = h
				mov edx, lx				;// edx = lx
				mov eax, 0				;// eax = s
				pxor mm7, mm7			;// mm7 = 0
dist1__l3:
				movq mm5, PACKED_1		;// mm5 = (1, 1, 1, 1)
				movd mm0, [esi]			;// lower 4 bytes in mm0 = esi[0..3]
				movd mm1, [esi+1]		;// lower 4 bytes in mm1 = esi[1..4]
				movd mm2, [esi+4]		;// lower 4 bytes in mm2 = esi[4..7]
				movd mm3, [esi+5]		;// lower 4 bytes in mm3 = esi[5..8]
				punpcklbw mm0, mm7		;// unpack the lower 4 bytes into mm0
				punpcklbw mm1, mm7		;// unpack the lower 4 bytes into mm1
				punpcklbw mm2, mm7		;// unpack the lower 4 bytes into mm2
				punpcklbw mm3, mm7		;// unpack the lower 4 bytes into mm3
				paddw mm0, mm1			;// mm0 += mm1
				paddw mm2, mm3			;// mm2 += mm3
				paddw mm0, mm5			;// mm0 += (1,1,1,1)
				paddw mm2, mm5			;// mm2 += (1,1,1,1)
				psrlw mm0, 1			;// mm0 >>= 1
				psrlw mm2, 1			;// mm2 >>= 1

				movd mm4, [edi]			;// lower 4 bytes in mm4 = edi[0..3]
				movd mm6, [edi+4]		;// lower 4 bytes in mm6 = edi[0..3]
				punpcklbw mm4, mm7		;// unpack the lower 4 bytes into mm4
				punpcklbw mm6, mm7		;// unpack the lower 4 bytes into mm6
				movq mm1, mm0
				movq mm3, mm2
				psubusw mm0, mm4
				psubusw mm2, mm6
				psubusw mm4, mm1
				psubusw mm6, mm3
				por mm0, mm4			;// mm0 = abs((esi[0..3] + esi[1..4]) >> 1) - edi[0..3])
				por mm2, mm6			;// mm2 = abs((esi[4..7] + esi[5..8]) >> 1) - edi[4..7])

				movq mm4, mm0
				movq mm6, mm2
				punpcklwd mm0, mm7		;// unpack the lower 4 word into mm0
				punpckhwd mm4, mm7		;// unpack the lower 4 word into mm4
				punpcklwd mm2, mm7		;// unpack the lower 4 word into mm2
				punpckhwd mm6, mm7		;// unpack the lower 4 word into mm6
				paddd mm0, mm2
				paddd mm4, mm6
				paddd mm0, mm4			;// mm0 += mm2 + mm4 + mm6
				movd ebx, mm0			;// load lower dword of mm0 into ebx
				add eax, ebx			;// eax += ebx
				psrlq mm0, 32			;// shift mm0 to get upper dword
				movd ebx, mm0			;// load lower dword of mm0 into ebx
				add eax, ebx			;// eax += ebx

				movd mm0, [esi+8]		;// lower 4 bytes in mm0 = esi[8..11]
				movd mm1, [esi+9]		;// lower 4 bytes in mm1 = esi[9..12]
				movd mm2, [esi+12]		;// lower 4 bytes in mm2 = esi[12..15]
				movd mm3, [esi+13]		;// lower 4 bytes in mm3 = esi[13..16]
				punpcklbw mm0, mm7		;// unpack the lower 4 bytes into mm0
				punpcklbw mm1, mm7		;// unpack the lower 4 bytes into mm1
				punpcklbw mm2, mm7		;// unpack the lower 4 bytes into mm2
				punpcklbw mm3, mm7		;// unpack the lower 4 bytes into mm3
				paddw mm0, mm1			;// mm0 += mm1
				paddw mm2, mm3			;// mm2 += mm3
				paddw mm0, mm5			;// mm0 += (1,1,1,1)
				paddw mm2, mm5			;// mm2 += (1,1,1,1)
				psrlw mm0, 1			;// mm0 >>= 1
				psrlw mm2, 1			;// mm2 >>= 1

				movd mm4, [edi+8]		;// lower 4 bytes in mm4 = edi[8..11]
				movd mm6, [edi+12]		;// lower 4 bytes in mm6 = edi[12..15]
				punpcklbw mm4, mm7		;// unpack the lower 4 bytes into mm4
				punpcklbw mm6, mm7		;// unpack the lower 4 bytes into mm6
				movq mm1, mm0
				movq mm3, mm2
				psubusw mm0, mm4
				psubusw mm2, mm6
				psubusw mm4, mm1
				psubusw mm6, mm3
				por mm0, mm4			;// mm0 = abs((esi[8..11] + esi[9..12]) >> 1) - edi[8..11])
				por mm2, mm6			;// mm2 = abs((esi[12..15] + esi[13..16]) >> 1) - edi[12..15])

				movq mm4, mm0
				movq mm6, mm2
				punpcklwd mm0, mm7		;// unpack the lower 4 word into mm0
				punpckhwd mm4, mm7		;// unpack the lower 4 word into mm4
				punpcklwd mm2, mm7		;// unpack the lower 4 word into mm2
				punpckhwd mm6, mm7		;// unpack the lower 4 word into mm6
				paddd mm0, mm2
				paddd mm4, mm6
				paddd mm0, mm4			;// mm0 += mm2 + mm4 + mm6
				movd ebx, mm0			;// load lower dword of mm0 into ebx
				add eax, ebx			;// eax += ebx
				psrlq mm0, 32			;// shift mm0 to get upper dword
				movd ebx, mm0			;// load lower dword of mm0 into ebx
				add eax, ebx			;// eax += ebx

				add esi, edx			;// esi += edx
				add edi, edx			;// edi += edx
				dec ecx					;// decrement ecx
				jnz dist1__l3			;// loop while not zero
				mov s, eax				;// s = eax
				emms					;// empty MMX state
			}
		}
		else
		{
			s = 0;
			p1 = blk1;
			p2 = blk2;

			for (j=0; j<h; j++)
			{
				for (i=0; i<16; i++)
				{
					v = ((unsigned int)(p1[i]+p1[i+1]+1)>>1) - p2[i];

					if (v>=0)
						s+= v;
					else
						s-= v;
				}
				p1+= lx;
				p2+= lx;
			}
		}
	}
	else if (!hx && hy)
	{
		if(cpu_3DNow)
		{
			s = 0;
			p1 = blk1;
			p2 = blk2;
			p1a = p1 + lx;
			for (j=0; j<h; j++)
			{
				__asm
				{
				// femms
				femmsInst
				mov              eax, p1
				movq             mm0,[eax]
				movq             mm1,[eax+8]
				mov              eax,p1a
				movq             mm2,[eax]
				movq             mm3,[eax+8]
				pavgusbmm0mm2
				movq             tmpResult,mm0
				pavgusbmm1mm3
				movq             tmpResult+8,mm1
				femmsInst
				}
				s += calcDist16(tmpResult,p2);
				p1 = p1a;
				p1a+= lx;
				p2+= lx;
			}
		}
		else if(cpu_MMX)
		{
			_asm
			{
				mov esi, blk1			;// esi = blk1
				mov edi, blk2			;// edi = blk2
				mov edx, lx				;// edx = lx
				mov ecx, h				;// ecx = h
				mov eax, 0				;// eax = s
				pxor mm7, mm7			;// mm7 = 0
dist1__l4:
				movq mm5, PACKED_1		;// mm5 = (1, 1, 1, 1)
				movd mm0, [esi]			;// lower 4 bytes in mm0 = esi[0..3]
				movd mm1, [esi+edx]		;// lower 4 bytes in mm1 = (esi + edx)[0..3]
				movd mm2, [esi+4]		;// lower 4 bytes in mm2 = esi[4..7]
				movd mm3, [esi+edx+4]	;// lower 4 bytes in mm3 = (esi + edx)[4..7]
				punpcklbw mm0, mm7		;// unpack the lower 4 bytes into mm0
				punpcklbw mm1, mm7		;// unpack the lower 4 bytes into mm1
				punpcklbw mm2, mm7		;// unpack the lower 4 bytes into mm2
				punpcklbw mm3, mm7		;// unpack the lower 4 bytes into mm3
				paddw mm0, mm1			;// mm0 += mm1
				paddw mm2, mm3			;// mm2 += mm3
				paddw mm0, mm5			;// mm0 += (1, 1, 1, 1)
				paddw mm2, mm5			;// mm2 += (1, 1, 1, 1)
				psrlw mm0, 1			;// mm0 >>= 1
				psrlw mm2, 1			;// mm0 >>= 1

				movd mm4, [edi]			;// lower 4 bytes in mm4 = edi[0..3]
				movd mm6, [edi+4]		;// lower 4 bytes in mm6 = edi[4..7]
				punpcklbw mm4, mm7		;// unpack the lower 4 bytes into mm4
				punpcklbw mm6, mm7		;// unpack the lower 4 bytes into mm6
				movq mm1, mm0
				movq mm3, mm2
				psubusw mm0, mm4
				psubusw mm2, mm6
				psubusw mm4, mm1
				psubusw mm6, mm3
				por mm0, mm4			;// mm0 = abs((esi[0..3] + (esi + edx)[0..3]) >> 1) - edi[0..3])
				por mm2, mm6			;// mm2 = abs((esi[4..7] + (esi + edx)[4..7]) >> 1) - edi[4..7])

				movq mm4, mm0
				movq mm6, mm2
				punpcklwd mm0, mm7		;// unpack lower 4 words into mm0
				punpckhwd mm4, mm7		;// unpack upper 4 words into mm4
				punpcklwd mm2, mm7		;// unpack lower 4 words into mm2
				punpckhwd mm6, mm7		;// unpack upper 4 words into mm6
				paddd mm0, mm2			
				paddd mm4, mm6			
				paddd mm0, mm4			;// mm0 += mm2 + mm4 + mm6
				movd ebx, mm0			;// load lower dword of mm0 into ebx
				add eax, ebx			;// eax += ebx
				psrlq mm0, 32			;// shift mm0 to get upper dword
				movd ebx, mm0			;// load lower dword of mm0 into ebx
				add eax, ebx			;// eax += ebx

				movd mm0, [esi+8]		;// lower 4 bytes in mm0 = esi[8..11]
				movd mm1, [esi+edx+8]	;// lower 4 bytes in mm1 = (esi + edx)[8..11]
				movd mm2, [esi+12]		;// lower 4 bytes in mm2 = esi[12..15]
				movd mm3, [esi+edx+12]	;// lower 4 bytes in mm3 = (esi + edx)[12..15]
				punpcklbw mm0, mm7		;// unpack the lower 4 bytes into mm0
				punpcklbw mm1, mm7		;// unpack the lower 4 bytes into mm1
				punpcklbw mm2, mm7		;// unpack the lower 4 bytes into mm2
				punpcklbw mm3, mm7		;// unpack the lower 4 bytes into mm3
				paddw mm0, mm1			;// mm0 += mm1
				paddw mm2, mm3			;// mm2 += mm3
				paddw mm0, mm5			;// mm0 += (1, 1, 1, 1)
				paddw mm2, mm5			;// mm2 += (1, 1, 1, 1)
				psrlw mm0, 1			;// mm0 >>= 1
				psrlw mm2, 1			;// mm0 >>= 1

				movd mm4, [edi+8]		;// lower 4 bytes in mm4 = edi[8..11]
				movd mm6, [edi+12]		;// lower 4 bytes in mm6 = edi[12..15]
				punpcklbw mm4, mm7		;// unpack the lower 4 bytes into mm4
				punpcklbw mm6, mm7		;// unpack the lower 4 bytes into mm6
				movq mm1, mm0
				movq mm3, mm2
				psubusw mm0, mm4
				psubusw mm2, mm6
				psubusw mm4, mm1
				psubusw mm6, mm3
				por mm0, mm4			;// mm0 = abs((esi[8..11] + (esi + edx)[8..11]) >> 1) - edi[8..11])
				por mm2, mm6			;// mm2 = abs((esi[12..15] + (esi + edx)[12..15]) >> 1) - edi[12..15])

				movq mm4, mm0
				movq mm6, mm2
				punpcklwd mm0, mm7		;// unpack the lower 2 words into mm0
				punpckhwd mm4, mm7		;// unpack the upper 2 words into mm4
				punpcklwd mm2, mm7		;// unpack the lower 2 words into mm2
				punpckhwd mm6, mm7		;// unpack the upper 2 words into mm6
				paddd mm0, mm2
				paddd mm4, mm6
				paddd mm0, mm4			;// mm0 += mm2 + mm4 + mm6
				movd ebx, mm0			;// load lower dword of mm0 into ebx
				add eax, ebx			;// eax += ebx
				psrlq mm0, 32			;// shift mm0 to get upper dword
				movd ebx, mm0			;// load lower dword of mm0 into ebx
				add eax, ebx			;// eax += ebx

				add esi, edx			;// esi += edx
				add edi, edx			;// edi += edx

				dec ecx					;// decrement ecx
				jnz dist1__l4			;// loop while not zero
				mov s, eax				;// s = eax
				emms					;// empty MMX state
			}
		}
		else
		{
			s = 0;
			p1 = blk1;
			p2 = blk2;
			p1a = p1 + lx;
			for (j=0; j<h; j++)
			{
				for (i=0; i<16; i++)
				{
					v = ((unsigned int)(p1[i]+p1a[i]+1)>>1) - p2[i];