void Predict_16x16_C(const NEW_GMC_DATA * const This,
uint8_t *dst,
const uint8_t *src,
int dststride,
int srcstride,
int x,
int y,
int rounding)
{
const int W = This->sW;
const int H = This->sH;
const int rho = 3 - This->accuracy;
const int Rounder = ( (1<<7) - (rounding<<(2*rho)) ) << 16;
const int dUx = This->dU[0];
const int dVx = This->dV[0];
const int dUy = This->dU[1];
const int dVy = This->dV[1];
int Uo = This->Uo + 16*(dUy*y + dUx*x);
int Vo = This->Vo + 16*(dVy*y + dVx*x);
int i, j;
dst += 16;
{
unsigned int ri = 16;
unsigned int rj = 16;
int Offset;
int u,v;
uint8_t *srctmp;
uint32_t tmpf = 0;
S32I2M(xr15,dUx);
S32I2M(xr14,dVx);
S32I2M(xr13,dUy);
S32I2M(xr12,dVy);
S32I2M(xr11,Uo); // Uo 11
S32I2M(xr10,Vo); // Vo 10
S32I2M(xr5, Rounder);
for (j = 16; j>0; --j)
{
D32SLL(xr9,xr11,xr10,xr8, 0x0); // U 9 ,V 8
D32ASUM_AA(xr11,xr13,xr12,xr10); // += dUy; +=dVy;
for (i = -16; i<0; ++i)
{
ri = 16;
rj = 16;
// ( U >> 16 ) ,( V >> 16 )
D32SAR(xr7,xr9,xr8,xr6, 0x8);
D32SAR(xr7,xr7,xr6,xr6, 0x8);
D32SLLV(xr7,xr6, rho); // << rho
u = S32M2I(xr7);
v = S32M2I(xr6);
D32ASUM_AA(xr9,xr15,xr14,xr8); // U += dUx; V += dVx;
if (u > 0 && u <= W)
{
ri = MTab[u&15];
Offset = u>>4;
}
else
{
if (u > W)
Offset = W>>4;
else
Offset = 0;
ri = MTab[0];
}
if (v > 0 && v <= H)
{
rj = MTab[v&15];
Offset += (v>>4)*srcstride;
}
static void rv40_dequant4x4(DCTELEM *block,uint32_t *dst, int n)
{
int i;
uint32_t src=block-4;
uint32_t dst_t = dst-4;
#if 0
for(i = 0; i < n; i++){
S32LDI(xr1,src,0x8);
S32LDD(xr2,src,0x4);
S32LDI(xr7,src,0x8);
S32LDD(xr8,src,0x4);
D16MUL_LW(xr4,xr12,xr1,xr3);
D16MUL_LW(xr10,xr12,xr7,xr9);
D16MUL_LW(xr14,xr12,xr8,xr15);
D32ASUM_AA(xr3,xr13,xr13,xr4);
D16MUL_LW(xr6,xr12,xr2,xr5);
D32SLR(xr3,xr3,xr4,xr4,4);
D32ASUM_AA(xr5,xr13,xr13,xr6);
D32ASUM_AA(xr9,xr13,xr13,xr10);
D32SLR(xr5,xr5,xr6,xr6,4);
D32SLR(xr9,xr9,xr10,xr10,4);
D32ASUM_AA(xr15,xr13,xr13,xr14);
S32SDI(xr3,dst_t,0x10);
S32STD(xr4,dst_t,0x4);
S32STD(xr5,dst_t,0x8);
S32STD(xr6,dst_t,0xc);
D32SLR(xr15,xr15,xr14,xr14,4);
S32SDI(xr9,dst_t,0x10);
S32STD(xr10,dst_t,0x4);
S32STD(xr15,dst_t,0x8);
S32STD(xr14,dst_t,0xc);
}
#else
/////////////////////
if(n == 1){
S32LDI(xr1,src,0x8);
S32LDI(xr2,src,0x8);
S32LDI(xr7,src,0x8);
S32LDI(xr8,src,0x8);
D16MUL_XW(xr4,xr12,xr1,xr3);
D16MUL_LW(xr10,xr12,xr7,xr9);
D16MUL_LW(xr14,xr12,xr8,xr15);
D32ASUM_AA(xr3,xr13,xr13,xr4);
D16MUL_LW(xr6,xr12,xr2,xr5);
D32SLR(xr3,xr3,xr4,xr4,4);
D32ASUM_AA(xr5,xr13,xr13,xr6);
D32ASUM_AA(xr9,xr13,xr13,xr10);
D32SLR(xr5,xr5,xr6,xr6,4);
D32SLR(xr9,xr9,xr10,xr10,4);
D32ASUM_AA(xr15,xr13,xr13,xr14);
S32SDI(xr3,dst_t,0x10);
S32STD(xr4,dst_t,0x4);
S32STD(xr0,dst_t,0x8);
S32STD(xr0,dst_t,0xc);
S32SDI(xr5,dst_t,0x10);
S32STD(xr6,dst_t,0x4);
S32STD(xr0,dst_t,0x8);
S32STD(xr0,dst_t,0xc);
D32SLR(xr15,xr15,xr14,xr14,4);
S32SDI(xr9,dst_t,0x10);
S32STD(xr10,dst_t,0x4);
S32STD(xr0,dst_t,0x8);
S32STD(xr0,dst_t,0xc);
//S32STD(xr15,dst_t,0x8);
//S32STD(xr14,dst_t,0xc);
S32SDI(xr15,dst_t,0x10);
S32STD(xr14,dst_t,0x4);
S32STD(xr0,dst_t,0x8);
S32STD(xr0,dst_t,0xc);
}
else if(n==2)
{
S32LDI(xr1,src,0x8);
S32LDD(xr2,src,0x4);
S32LDI(xr7,src,0x8);
S32LDD(xr8,src,0x4);
D16MUL_XW(xr4,xr12,xr1,xr3);
D16MUL_LW(xr10,xr12,xr7,xr9);
D16MUL_LW(xr14,xr12,xr8,xr15);
D32ASUM_AA(xr3,xr13,xr13,xr4);
D16MUL_LW(xr6,xr12,xr2,xr5);
D32SLR(xr3,xr3,xr4,xr4,4);
D32ASUM_AA(xr5,xr13,xr13,xr6);
D32ASUM_AA(xr9,xr13,xr13,xr10);
D32SLR(xr5,xr5,xr6,xr6,4);
D32SLR(xr9,xr9,xr10,xr10,4);
D32ASUM_AA(xr15,xr13,xr13,xr14);
S32SDI(xr3,dst_t,0x10);
S32STD(xr4,dst_t,0x4);
S32STD(xr5,dst_t,0x8);
S32STD(xr6,dst_t,0xc);
D32SLR(xr15,xr15,xr14,xr14,4);
S32SDI(xr9,dst_t,0x10);
S32STD(xr10,dst_t,0x4);
S32STD(xr15,dst_t,0x8);
S32STD(xr14,dst_t,0xc);
S32SDI(xr0,dst_t,0x10);
S32STD(xr0,dst_t,0x4);
S32STD(xr0,dst_t,0x8);
S32STD(xr0,dst_t,0xc);
S32SDI(xr0,dst_t,0x10);
S32STD(xr0,dst_t,0x4);
S32STD(xr0,dst_t,0x8);
S32STD(xr0,dst_t,0xc);
}
else
{
S32LDI(xr1,src,0x8);
S32LDD(xr2,src,0x4);
S32LDI(xr7,src,0x8);
S32LDD(xr8,src,0x4);
D16MUL_XW(xr4,xr12,xr1,xr3);
D16MUL_LW(xr10,xr12,xr7,xr9);
D16MUL_LW(xr14,xr12,xr8,xr15);
D32ASUM_AA(xr3,xr13,xr13,xr4);
D16MUL_LW(xr6,xr12,xr2,xr5);
D32SLR(xr3,xr3,xr4,xr4,4);
D32ASUM_AA(xr5,xr13,xr13,xr6);
D32ASUM_AA(xr9,xr13,xr13,xr10);
D32SLR(xr5,xr5,xr6,xr6,4);
D32SLR(xr9,xr9,xr10,xr10,4);
D32ASUM_AA(xr15,xr13,xr13,xr14);
S32SDI(xr3,dst_t,0x10);
S32STD(xr4,dst_t,0x4);
S32STD(xr5,dst_t,0x8);
S32STD(xr6,dst_t,0xc);
D32SLR(xr15,xr15,xr14,xr14,4);
S32SDI(xr9,dst_t,0x10);
S32STD(xr10,dst_t,0x4);
S32STD(xr15,dst_t,0x8);
S32STD(xr14,dst_t,0xc);
S32LDI(xr1,src,0x8);
S32LDD(xr2,src,0x4);
S32LDI(xr7,src,0x8);
S32LDD(xr8,src,0x4);
D16MUL_LW(xr4,xr12,xr1,xr3);
D16MUL_LW(xr10,xr12,xr7,xr9);
D16MUL_LW(xr14,xr12,xr8,xr15);
D32ASUM_AA(xr3,xr13,xr13,xr4);
D16MUL_LW(xr6,xr12,xr2,xr5);
D32SLR(xr3,xr3,xr4,xr4,4);
D32ASUM_AA(xr5,xr13,xr13,xr6);
D32ASUM_AA(xr9,xr13,xr13,xr10);
D32SLR(xr5,xr5,xr6,xr6,4);
D32SLR(xr9,xr9,xr10,xr10,4);
D32ASUM_AA(xr15,xr13,xr13,xr14);
S32SDI(xr3,dst_t,0x10);
S32STD(xr4,dst_t,0x4);
S32STD(xr5,dst_t,0x8);
S32STD(xr6,dst_t,0xc);
D32SLR(xr15,xr15,xr14,xr14,4);
S32SDI(xr9,dst_t,0x10);
S32STD(xr10,dst_t,0x4);
S32STD(xr15,dst_t,0x8);
S32STD(xr14,dst_t,0xc);
}
#endif
}
