backend: add double version of remquo

cp from fdlibm, need to be refined to pass cft

Signed-off-by: rander <rander.wang@intel.com>
Tested-by: Yang Rong <rong.r.yang@intel.com>

4 files changed, 199 insertions, 0 deletions

diff --git a/backend/src/libocl/tmpl/ocl_math.tmpl.cl b/backend/src/libocl/tmpl/ocl_math.tmpl.cl
index 3ea97510..850d198c 100644
--- a/backend/src/libocl/tmpl/ocl_math.tmpl.cl
+++ b/backend/src/libocl/tmpl/ocl_math.tmpl.cl
@@ -4465,5 +4465,150 @@ OVERLOADABLE double lgamma_r(double x, private int *signgamp)
 	return r;
 }
 
+OVERLOADABLE double remquo(double x, double p, global int *quo)
+{
+	int hx,hp;
+	unsigned sx,lx,lp;
+	double p_half, zero = 0.0;
+	double xx = x;
+
+	hx = __HI(x);		/* high word of x */
+	lx = __LO(x);		/* low  word of x */
+	hp = __HI(p);		/* high word of p */
+	lp = __LO(p);		/* low  word of p */
+	sx = hx&0x80000000;
+	hp &= 0x7fffffff;
+	hx &= 0x7fffffff;
+
+	/* purge off exception values */
+	if((hp|lp)==0) return (x*p)/(x*p); 	/* p = 0 */
+	if((hx>=0x7ff00000)||			/* x not finite */
+	  ((hp>=0x7ff00000)&&			/* p is NaN */
+	  (((hp-0x7ff00000)|lp)!=0)))
+		return (x*p)/(x*p);
+
+
+	if (hp<=0x7fdfffff) x = fmod(x,p+p);	/* now x < 2p */
+	if (((hx-hp)|(lx-lp))==0) return zero*x;
+	x  = fabs(x);
+	p  = fabs(p);
+	if (hp<0x00200000)
+		{
+		if(x+x>p) {
+		x-=p;
+		if(x+x>=p) x -= p;
+		}
+	}
+		else
+		{
+		p_half = 0.5*p;
+		if(x>p_half)
+			{
+		x-=p;
+		if(x>=p_half) x -= p;
+		}
+	}
+	__setHigh(&x,  __HI(x) ^sx);
+		*quo = (xx -x)/p;
+	return x;
+}
+
+OVERLOADABLE double remquo(double x, double p, local int *quo)
+{
+	int hx,hp;
+	unsigned sx,lx,lp;
+	double p_half, zero = 0.0;
+	double xx = x;
+
+	hx = __HI(x);		/* high word of x */
+	lx = __LO(x);		/* low  word of x */
+	hp = __HI(p);		/* high word of p */
+	lp = __LO(p);		/* low  word of p */
+	sx = hx&0x80000000;
+	hp &= 0x7fffffff;
+	hx &= 0x7fffffff;
+
+	/* purge off exception values */
+	if((hp|lp)==0) return (x*p)/(x*p); 	/* p = 0 */
+	if((hx>=0x7ff00000)||			/* x not finite */
+	  ((hp>=0x7ff00000)&&			/* p is NaN */
+	  (((hp-0x7ff00000)|lp)!=0)))
+		return (x*p)/(x*p);
+
+
+	if (hp<=0x7fdfffff) x = fmod(x,p+p);	/* now x < 2p */
+	if (((hx-hp)|(lx-lp))==0) return zero*x;
+	x  = fabs(x);
+	p  = fabs(p);
+	if (hp<0x00200000)
+		{
+		if(x+x>p) {
+		x-=p;
+		if(x+x>=p) x -= p;
+		}
+	}
+		else
+		{
+		p_half = 0.5*p;
+		if(x>p_half)
+			{
+		x-=p;
+		if(x>=p_half) x -= p;
+		}
+	}
+	__setHigh(&x,  __HI(x) ^sx);
+		*quo = (xx -x)/p;
+	return x;
+
+}
+
+OVERLOADABLE double remquo(double x, double p, private int *quo)
+{
+	int hx,hp;
+	unsigned sx,lx,lp;
+	double p_half, zero = 0.0;
+	double xx = x;
+
+	hx = __HI(x);		/* high word of x */
+	lx = __LO(x);		/* low  word of x */
+	hp = __HI(p);		/* high word of p */
+	lp = __LO(p);		/* low  word of p */
+	sx = hx&0x80000000;
+	hp &= 0x7fffffff;
+	hx &= 0x7fffffff;
+
+	/* purge off exception values */
+	if((hp|lp)==0) return (x*p)/(x*p); 	/* p = 0 */
+	if((hx>=0x7ff00000)||			/* x not finite */
+	  ((hp>=0x7ff00000)&&			/* p is NaN */
+	  (((hp-0x7ff00000)|lp)!=0)))
+		return (x*p)/(x*p);
+
+
+	if (hp<=0x7fdfffff) x = fmod(x,p+p);	/* now x < 2p */
+	if (((hx-hp)|(lx-lp))==0) return zero*x;
+	x  = fabs(x);
+	p  = fabs(p);
+	if (hp<0x00200000)
+		{
+		if(x+x>p) {
+		x-=p;
+		if(x+x>=p) x -= p;
+		}
+	}
+		else
+		{
+		p_half = 0.5*p;
+		if(x>p_half)
+			{
+		x-=p;
+		if(x>=p_half) x -= p;
+		}
+	}
+	__setHigh(&x,  __HI(x) ^sx);
+		*quo = (xx -x)/p;
+	return x;
+
+}
 
 
diff --git a/backend/src/libocl/tmpl/ocl_math.tmpl.h b/backend/src/libocl/tmpl/ocl_math.tmpl.h
index 936f72f6..1619210e 100644
--- a/backend/src/libocl/tmpl/ocl_math.tmpl.h
+++ b/backend/src/libocl/tmpl/ocl_math.tmpl.h
@@ -243,4 +243,8 @@ OVERLOADABLE double frexp(double x, private int *exp);
 OVERLOADABLE double lgamma_r(double x, global int *signgamp);
 OVERLOADABLE double lgamma_r(double x, local int *signgamp);
 OVERLOADABLE double lgamma_r(double x, private int *signgamp);
+OVERLOADABLE double remquo(double x, double y, global int *quo);
+OVERLOADABLE double remquo(double x, double y, local int *quo);
+OVERLOADABLE double remquo(double x, double y, private int *quo);
+
 
diff --git a/backend/src/libocl/tmpl/ocl_math_20.tmpl.cl b/backend/src/libocl/tmpl/ocl_math_20.tmpl.cl
index 50fbff28..dd35b84b 100644
--- a/backend/src/libocl/tmpl/ocl_math_20.tmpl.cl
+++ b/backend/src/libocl/tmpl/ocl_math_20.tmpl.cl
@@ -4020,4 +4020,53 @@ OVERLOADABLE double lgamma_r(double x, int *signgamp)
 	return r;
 }
 
+OVERLOADABLE double remquo(double x, double p, int *quo)
+{
+	int hx,hp;
+	unsigned sx,lx,lp;
+	double p_half, zero = 0.0;
+	double xx = x;
+
+	hx = __HI(x);		/* high word of x */
+	lx = __LO(x);		/* low  word of x */
+	hp = __HI(p);		/* high word of p */
+	lp = __LO(p);		/* low  word of p */
+	sx = hx&0x80000000;
+	hp &= 0x7fffffff;
+	hx &= 0x7fffffff;
+
+	/* purge off exception values */
+	if((hp|lp)==0) return (x*p)/(x*p); 	/* p = 0 */
+	if((hx>=0x7ff00000)||			/* x not finite */
+	  ((hp>=0x7ff00000)&&			/* p is NaN */
+	  (((hp-0x7ff00000)|lp)!=0)))
+		return (x*p)/(x*p);
+
+
+	if (hp<=0x7fdfffff) x = fmod(x,p+p);	/* now x < 2p */
+	if (((hx-hp)|(lx-lp))==0) return zero*x;
+	x  = fabs(x);
+	p  = fabs(p);
+	if (hp<0x00200000)
+		{
+		if(x+x>p) {
+		x-=p;
+		if(x+x>=p) x -= p;
+		}
+	}
+		else
+		{
+		p_half = 0.5*p;
+		if(x>p_half)
+			{
+		x-=p;
+		if(x>=p_half) x -= p;
+		}
+	}
+	__setHigh(&x,  __HI(x) ^sx);
+		*quo = (xx -x)/p;
+	return x;
+
+}
+
 
diff --git a/backend/src/libocl/tmpl/ocl_math_20.tmpl.h b/backend/src/libocl/tmpl/ocl_math_20.tmpl.h
index 419f06e1..79383ef5 100644
--- a/backend/src/libocl/tmpl/ocl_math_20.tmpl.h
+++ b/backend/src/libocl/tmpl/ocl_math_20.tmpl.h
@@ -214,4 +214,5 @@ OVERLOADABLE float half_tan(float x);
 OVERLOADABLE double fract(double x, double *p);
 OVERLOADABLE double frexp(double x, int *exp);
 OVERLOADABLE double lgamma_r(double x, int *signgamp);
+OVERLOADABLE double remquo(double x, double y, int *quo);