double expdp_c(double a)
{
double Halfe = 0.5; //Zeroe = 0.0, Onee = 1.0, ;
double Maxe = 1.7976931348623157e+308 ;
double LnMaxe= 709.7827138470578 ;
double a0e= 0.249999999999999993;
double a1e= 0.694360001511792852e-2 ;
double a2e= 0.165203300268279130e-4;
double b0e= 0.5 ;
double b1e= 0.555538666969001188e-1 ;
double b2e= 0.495862884905441294e-3 ;
double c1e= 0.693359375 ; /* 355/512 */
double C2e=-2.1219444005469058277e-4 ;
double L2e = 1.4426950408889634074 ; /* log (base 2) of e */
double Ye,Xe,We,Re,Se,Be,Ce,De;
int Ne, upper;
Ye = a;
Ce = Ye*L2e ; /* base e --> base 2 argument */
Ne = _dpint(Ce) ; /* get unbiased exponent as an int */
Se = Ne ; /* double(int N) */
Xe = (Ye - Se*c1e) - Se*C2e ; /* range reduction */
We = Xe*Xe ;
Be = (b2e*We + b1e)*We + b0e ; /* denominator */
De = ((a2e*We + a1e)*We + a0e)*Xe ; /* numerator */
Re = Halfe + divdpMod_ci5( De, (Be-De) ) ;
upper = 1024 + Ne;
upper = _extu(upper,20,0);
Se = _itod(upper,0x00000000);
Ce = Re*Se ; /* scale by power of 2 */
if (a > LnMaxe) Ce = (Maxe) ; /* > LnMax returns MAX */
return (Ce) ;
}
static inline int divide ( int num, int den )
{
int i, sign ;
sign = num >> 31 ; // test the sign of numerator
num = _abs(num);
num += den >> 1 ;
den <<= NUM_NORM ; // align denominator
for ( i = 0 ; i <= NUM_NORM ; i++ )
num = _subc(num, den);
num = _extu(num, 31 - NUM_NORM, 31 - NUM_NORM);
/* unsigned extraction */
if ( sign )
num = -num ;
return (num);
}
/**
This function allows to get the luminance prediction of a non IDR picture when xFracl = 0 and yFracl = 1.
@param image Table of current frame.
@param refPicLXl Table of the reference decoded picture buffer.
@param PicWidthSamples Stride of the reference buffer.
@param stride Stride of the current image.
*/
void luma_sample_interp_0_1_TI(unsigned char image [], unsigned char refPicLXl[], const short PicWidthSamples, const short stride){
/* No horizontal interpolation */
unsigned int uiLine1,uiLine2,uiLine3,uiLine4,uiLine5,uiLine6,uiLine7,uiLine8,uiLine9;
unsigned int uiTmpLine12_h,uiTmpLine34_h,uiTmpLine12_l,uiTmpLine34_l,uiTmpLine1234_4,uiTmpLine1234_2,uiTmpLine1234_3,uiTmpLine1234_1;
unsigned int uiTmpLine56_h,uiTmpLine78_h,uiTmpLine56_l,uiTmpLine78_l,uiTmpLine5678_4,uiTmpLine5678_2,uiTmpLine5678_3,uiTmpLine5678_1;
unsigned int tmpc1_1,tmpc1_2,tmpc2_1,tmpc2_2,tmpc1,tmpc2;
unsigned int tmpc3_1,tmpc3_2,tmpc4_1,tmpc4_2,tmpc3,tmpc4;
unsigned int tmp12,tmp34,tmpend1,tmpend2,tmpend3,tmpend4;
unsigned int tmpl9l,tmpl9h;
unsigned int input1,input2,input3,input4;
unsigned char* pRefImgPtr;
unsigned char* pImgPtr;
pRefImgPtr = refPicLXl-(PicWidthSamples<<1);
pImgPtr = image;
uiLine1 = _mem4(pRefImgPtr);
pRefImgPtr += PicWidthSamples;
uiLine2 = _mem4(pRefImgPtr);
pRefImgPtr += PicWidthSamples;
uiLine3 = _mem4(pRefImgPtr);
pRefImgPtr += PicWidthSamples;
uiLine4 = _mem4(pRefImgPtr);
pRefImgPtr += PicWidthSamples;
uiLine5 = _mem4(pRefImgPtr);
pRefImgPtr += PicWidthSamples;
uiLine6 = _mem4(pRefImgPtr);
pRefImgPtr += PicWidthSamples;
uiLine7 = _mem4(pRefImgPtr);
pRefImgPtr += PicWidthSamples;
uiLine8 = _mem4(pRefImgPtr);
pRefImgPtr += PicWidthSamples;
uiLine9 = _mem4(pRefImgPtr);
input1 = uiLine3;
input2 = uiLine4;
input3 = uiLine5;
input4 = uiLine6;
uiTmpLine12_h = _packh4(uiLine1,uiLine2);
uiTmpLine34_h = _packh4(uiLine3,uiLine4);
uiTmpLine12_l = _packl4(uiLine1,uiLine2);
uiTmpLine34_l = _packl4(uiLine3,uiLine4);
uiTmpLine1234_4 = _packh4(uiTmpLine12_h,uiTmpLine34_h);
uiTmpLine1234_2 = _packl4(uiTmpLine12_h,uiTmpLine34_h);
uiTmpLine1234_3 = _packh4(uiTmpLine12_l,uiTmpLine34_l);
uiTmpLine1234_1 = _packl4(uiTmpLine12_l,uiTmpLine34_l);
uiTmpLine56_h = _packh4(uiLine5,uiLine6);
uiTmpLine78_h = _packh4(uiLine7,uiLine8);
uiTmpLine56_l = _packl4(uiLine5,uiLine6);
uiTmpLine78_l = _packl4(uiLine7,uiLine8);
uiTmpLine5678_4 = _packh4(uiTmpLine56_h,uiTmpLine78_h);
uiTmpLine5678_2 = _packl4(uiTmpLine56_h,uiTmpLine78_h);
uiTmpLine5678_3 = _packh4(uiTmpLine56_l,uiTmpLine78_l);
uiTmpLine5678_1 = _packl4(uiTmpLine56_l,uiTmpLine78_l);
tmpc1_1 = _dotpsu4(0x01FB1414,uiTmpLine1234_1);
tmpc1_2 = _dotpsu4(0xFB010000,uiTmpLine5678_1);
tmpc2_1 = _dotpsu4(0x01FB1414,uiTmpLine1234_2);
tmpc2_2 = _dotpsu4(0xFB010000,uiTmpLine5678_2);
tmpc1 = _spack2(tmpc1_1,tmpc2_1);
tmpc2 = _spack2(tmpc1_2,tmpc2_2);
tmp12 = _sadd2(tmpc1,tmpc2);
tmp12 = _shr2(_sadd2(tmp12,0x00100010),5);
tmpc3_1 = _dotpsu4(0x01FB1414,uiTmpLine1234_3);
tmpc3_2 = _dotpsu4(0xFB010000,uiTmpLine5678_3);
tmpc4_1 = _dotpsu4(0x01FB1414,uiTmpLine1234_4);
tmpc4_2 = _dotpsu4(0xFB010000,uiTmpLine5678_4);
tmpc3 = _spack2(tmpc3_1,tmpc4_1);
tmpc4 = _spack2(tmpc3_2,tmpc4_2);
tmp34 = _sadd2(tmpc3,tmpc4);
tmp34 = _shr2(_sadd2(tmp34,0x00100010),5);
tmpend1 = _spacku4(tmp34,tmp12);
tmpend1 = _swap4(tmpend1);
_amem4(pImgPtr) = _avgu4(tmpend1,input1);
pImgPtr += stride;
tmpc1_1 = _dotpsu4(0x0001FB14,uiTmpLine1234_1);
tmpc1_2 = _dotpsu4(0x14FB0100,uiTmpLine5678_1);
tmpc2_1 = _dotpsu4(0x0001FB14,uiTmpLine1234_2);
tmpc2_2 = _dotpsu4(0x14FB0100,uiTmpLine5678_2);
tmpc1 = _spack2(tmpc1_1,tmpc2_1);
tmpc2 = _spack2(tmpc1_2,tmpc2_2);
tmp12 = _sadd2(tmpc1,tmpc2);
tmp12 = _shr2(_sadd2(tmp12,0x00100010),5);
tmpc3_1 = _dotpsu4(0x0001FB14,uiTmpLine1234_3);
tmpc3_2 = _dotpsu4(0x14FB0100,uiTmpLine5678_3);
tmpc4_1 = _dotpsu4(0x0001FB14,uiTmpLine1234_4);
tmpc4_2 = _dotpsu4(0x14FB0100,uiTmpLine5678_4);
tmpc3 = _spack2(tmpc3_1,tmpc4_1);
tmpc4 = _spack2(tmpc3_2,tmpc4_2);
tmp34 = _sadd2(tmpc3,tmpc4);
tmp34 = _shr2(_sadd2(tmp34,0x00100010),5);
tmpend2 = _spacku4(tmp34,tmp12);
tmpend2 = _swap4(tmpend2);
_amem4(pImgPtr) = _avgu4(tmpend2,input2);
pImgPtr += stride;
tmpc1_1 = _dotpsu4(0x000001FB,uiTmpLine1234_1);
tmpc1_2 = _dotpsu4(0x1414FB01,uiTmpLine5678_1);
tmpc2_1 = _dotpsu4(0x000001FB,uiTmpLine1234_2);
tmpc2_2 = _dotpsu4(0x1414FB01,uiTmpLine5678_2);
tmpc1 = _spack2(tmpc1_1,tmpc2_1);
tmpc2 = _spack2(tmpc1_2,tmpc2_2);
tmp12 = _sadd2(tmpc1,tmpc2);
tmp12 = _shr2(_sadd2(tmp12,0x00100010),5);
tmpc3_1 = _dotpsu4(0x000001FB,uiTmpLine1234_3);
tmpc3_2 = _dotpsu4(0x1414FB01,uiTmpLine5678_3);
tmpc4_1 = _dotpsu4(0x000001FB,uiTmpLine1234_4);
tmpc4_2 = _dotpsu4(0x1414FB01,uiTmpLine5678_4);
tmpc3 = _spack2(tmpc3_1,tmpc4_1);
tmpc4 = _spack2(tmpc3_2,tmpc4_2);
tmp34 = _sadd2(tmpc3,tmpc4);
tmp34 = _shr2(_sadd2(tmp34,0x00100010),5);
tmpend3 = _spacku4(tmp34,tmp12);
tmpend3 = _swap4(tmpend3);
_amem4(pImgPtr) = _avgu4(tmpend3,input3);
pImgPtr += stride;
uiLine9 = _swap4(uiLine9);
tmpl9h = _unpkhu4 (uiLine9);
tmpl9l = _unpklu4 (uiLine9);
tmpc1_1 = _extu(uiTmpLine1234_1,24,24);//_dotpsu4(0x00000001,uiTmpLine1234_1);
tmpc1_2 = _dotpsu4(0xFB1414FB,uiTmpLine5678_1);
tmpc2_1 = _extu(uiTmpLine1234_2,24,24);//_dotpsu4(0x00000001,uiTmpLine1234_2);
tmpc2_2 = _dotpsu4(0xFB1414FB,uiTmpLine5678_2);
tmpc1 = _spack2(tmpc1_1,tmpc2_1);
tmpc2 = _spack2(tmpc1_2,tmpc2_2);
tmp12 = _sadd2(tmpc1,tmpc2);
tmp12 = _sadd2(tmp12,tmpl9l);
tmp12 = _shr2(_sadd2(tmp12,0x00100010),5);
tmpc3_1 = _extu(uiTmpLine1234_3,24,24);//_dotpsu4(0x00000001,uiTmpLine1234_3);
tmpc3_2 = _dotpsu4(0xFB1414FB,uiTmpLine5678_3);
tmpc4_1 = _extu(uiTmpLine1234_4,24,24);//_dotpsu4(0x00000001,uiTmpLine1234_4);
tmpc4_2 = _dotpsu4(0xFB1414FB,uiTmpLine5678_4);
tmpc3 = _spack2(tmpc3_1,tmpc4_1);
tmpc4 = _spack2(tmpc3_2,tmpc4_2);
tmp34 = _sadd2(tmpc3,tmpc4);
tmp34 = _sadd2(tmp34,tmpl9h);
tmp34 = _shr2(_sadd2(tmp34,0x00100010),5);
tmpend4 = _spacku4(tmp34,tmp12);
tmpend4 = _swap4(tmpend4);
_amem4(pImgPtr) = _avgu4(tmpend4,input4);
}
double logdp_c(double a)
{
// double One=1.0 ;
double Zero = 0.0, Half = 0.5 ;
double srHalf= 0.70710678118654752440 ; /* sqrt(0.5) */
double MIN = 2.2250738585072014e-308 ;
double LnMin = -708.3964185322641 ;
double MAX = 1.7976931348623157e+308 ;
double LnMax = 709.7827138470578 ;
double a0 =-0.64124943423745581147e+2 ;
double a1 = 0.16383943563021534222e+2 ;
double a2 =-0.78956112887491257267e+0 ;
double b0 =-0.76949932108494879777e+3 ;
double b1 = 0.31203222091924532844e+3 ;
double b2 =-0.35667977739034646171e+2 ; /* Note b3 = 1.0 */
double c1 = 0.693359375 ; /* 355/512 */
double c2 =-2.121944400546905827679e-4 ;
double Y,Z,zn,zd,X,W,Rz,Sa,Bd,Cn,Da ;
int N, exp, upper;
Y = a ;
exp = _extu(_hi(Y),1,21);
N = exp - 1022;
upper = _clr(_hi(Y),20,31);
upper = 0x3fe00000 | upper;
Z = _itod(upper, _lo(Y));
if (exp == 0) Z = 0;
if (Z > srHalf)
{ zn=(Z-Half)-Half ; zd=Z*Half+Half ; }
else
{ zn=Z-Half; zd=zn*Half+Half ; N=N-1 ; }
//X = zn/zd;
X = divdpMod_clog(zn,zd);
W = X*X ;
Bd= ( (W+b2) * W + b1) * W + b0 ;
Cn= (W*a2 + a1) * W + a0 ;
Rz= W * divdpMod_clog(Cn,Bd);//Cn / Bd ;
Sa = X + X*Rz ;
Cn= N ;
Da = (Cn*c2 + Sa) + Cn*c1 ;
if (Y <= Zero) Da = 0;//{Z = setnmx(Y) ; Da = (Z) ; }
if (Y < MIN) Da = (LnMin) ;
if (Y > MAX) Da = (LnMax) ;
return (Da) ;
}
