[babl] babl: include out own powf implementation
- From: Øyvind Kolås <ok src gnome org>
- To: commits-list gnome org
- Cc:
- Subject: [babl] babl: include out own powf implementation
- Date: Thu, 31 Aug 2017 03:14:54 +0000 (UTC)
commit 523fd4de8616c82ff554022ab5ab318d8c2c2a9b
Author: Øyvind Kolås <pippin gimp org>
Date: Thu Aug 31 04:46:29 2017 +0200
babl: include out own powf implementation
babl/babl-trc.c | 2 +
babl/babl-trc.h | 293 ++++++++++++++++++++++++++++++++++++++++++++++++++++++-
2 files changed, 293 insertions(+), 2 deletions(-)
---
diff --git a/babl/babl-trc.c b/babl/babl-trc.c
index b133f8d..8992200 100644
--- a/babl/babl-trc.c
+++ b/babl/babl-trc.c
@@ -50,6 +50,8 @@ babl_trc_new (const char *name,
trc.instance.id = 0;
trc.type = type;
trc.gamma = gamma;
+ if (gamma > 0.0001)
+ trc.rgamma = 1.0 / gamma;
if (n_lut )
{
diff --git a/babl/babl-trc.h b/babl/babl-trc.h
index 779ec97..caec30c 100644
--- a/babl/babl-trc.h
+++ b/babl/babl-trc.h
@@ -36,6 +36,7 @@ typedef struct
BablTRCType type;
int lut_size;
double gamma;
+ float rgamma;
char name[128];
float *lut;
float *inv_lut;
@@ -73,16 +74,304 @@ static inline float _babl_trc_linear (const Babl *trc_, float value)
return 1.0;
}
+/* origin: FreeBSD /usr/src/lib/msun/src/e_powf.c, copied from musl */
+/*
+ * Conversion to float by Ian Lance Taylor, Cygnus Support, ian cygnus com.
+ */
+/*
+ * ====================================================
+ * Copyright (C) 1993 by Sun Microsystems, Inc. All rights reserved.
+ *
+ * Developed at SunPro, a Sun Microsystems, Inc. business.
+ * Permission to use, copy, modify, and distribute this
+ * software is freely granted, provided that this notice
+ * is preserved.
+ * ====================================================
+ */
+
+//#include "libm.h"
+
+/* Get a 32 bit int from a float. */
+#define GET_FLOAT_WORD(w,d) \
+do { \
+ union {float f; uint32_t i;} __u; \
+ __u.f = (d); \
+ (w) = __u.i; \
+} while (0)
+
+/* Set a float from a 32 bit int. */
+#define SET_FLOAT_WORD(d,w) \
+do { \
+ union {float f; uint32_t i;} __u; \
+ __u.i = (w); \
+ (d) = __u.f; \
+} while (0)
+
+
+static const float
+bp[] = {1.0, 1.5,},
+dp_h[] = { 0.0, 5.84960938e-01,}, /* 0x3f15c000 */
+dp_l[] = { 0.0, 1.56322085e-06,}, /* 0x35d1cfdc */
+two24 = 16777216.0, /* 0x4b800000 */
+huge = 1.0e30,
+tiny = 1.0e-30,
+/* poly coefs for (3/2)*(log(x)-2s-2/3*s**3 */
+L1 = 6.0000002384e-01, /* 0x3f19999a */
+L2 = 4.2857143283e-01, /* 0x3edb6db7 */
+L3 = 3.3333334327e-01, /* 0x3eaaaaab */
+L4 = 2.7272811532e-01, /* 0x3e8ba305 */
+L5 = 2.3066075146e-01, /* 0x3e6c3255 */
+L6 = 2.0697501302e-01, /* 0x3e53f142 */
+P1 = 1.6666667163e-01, /* 0x3e2aaaab */
+P2 = -2.7777778450e-03, /* 0xbb360b61 */
+P3 = 6.6137559770e-05, /* 0x388ab355 */
+P4 = -1.6533901999e-06, /* 0xb5ddea0e */
+P5 = 4.1381369442e-08, /* 0x3331bb4c */
+lg2 = 6.9314718246e-01, /* 0x3f317218 */
+lg2_h = 6.93145752e-01, /* 0x3f317200 */
+lg2_l = 1.42860654e-06, /* 0x35bfbe8c */
+ovt = 4.2995665694e-08, /* -(128-log2(ovfl+.5ulp)) */
+cp = 9.6179670095e-01, /* 0x3f76384f =2/(3ln2) */
+cp_h = 9.6191406250e-01, /* 0x3f764000 =12b cp */
+cp_l = -1.1736857402e-04, /* 0xb8f623c6 =tail of cp_h */
+ivln2 = 1.4426950216e+00, /* 0x3fb8aa3b =1/ln2 */
+ivln2_h = 1.4426879883e+00, /* 0x3fb8aa00 =16b 1/ln2*/
+ivln2_l = 7.0526075433e-06; /* 0x36eca570 =1/ln2 tail*/
+
+static inline float obabl_powf(float x, float y)
+{
+ float z,ax,z_h,z_l,p_h,p_l;
+ float y1,t1,t2,r,s,sn,t,u,v,w;
+ int32_t i,j,k,yisint,n;
+ int32_t hx,hy,ix,iy,is;
+
+ GET_FLOAT_WORD(hx, x);
+ GET_FLOAT_WORD(hy, y);
+ ix = hx & 0x7fffffff;
+ iy = hy & 0x7fffffff;
+
+#if 0
+ /* x**0 = 1, even if x is NaN */
+ if (iy == 0)
+ return 1.0f;
+#endif
+ /* 1**y = 1, even if y is NaN */
+ if (hx == 0x3f800000)
+ return 1.0f;
+ /* NaN if either arg is NaN */
+ if (ix > 0x7f800000 || iy > 0x7f800000)
+ return x + y;
+
+ /* determine if y is an odd int when x < 0
+ * yisint = 0 ... y is not an integer
+ * yisint = 1 ... y is an odd int
+ * yisint = 2 ... y is an even int
+ */
+ yisint = 0;
+ if (hx < 0) {
+ if (iy >= 0x4b800000)
+ yisint = 2; /* even integer y */
+ else if (iy >= 0x3f800000) {
+ k = (iy>>23) - 0x7f; /* exponent */
+ j = iy>>(23-k);
+ if ((j<<(23-k)) == iy)
+ yisint = 2 - (j & 1);
+ }
+ }
+#if 0
+ /* special value of y */
+ if (iy == 0x7f800000) { /* y is +-inf */
+ if (ix == 0x3f800000) /* (-1)**+-inf is 1 */
+ return 1.0f;
+ else if (ix > 0x3f800000) /* (|x|>1)**+-inf = inf,0 */
+ return hy >= 0 ? y : 0.0f;
+ else /* (|x|<1)**+-inf = 0,inf */
+ return hy >= 0 ? 0.0f: -y;
+ }
+ if (iy == 0x3f800000) /* y is +-1 */
+ return hy >= 0 ? x : 1.0f/x;
+ if (hy == 0x40000000) /* y is 2 */
+ return x*x;
+ if (hy == 0x3f000000) { /* y is 0.5 */
+ if (hx >= 0) /* x >= +0 */
+ return sqrtf(x);
+ }
+#endif
+
+ ax = fabsf(x);
+ /* special value of x */
+ if (ix == 0x7f800000 || ix == 0 || ix == 0x3f800000) { /* x is +-0,+-inf,+-1 */
+ z = ax;
+ if (hy < 0) /* z = (1/|x|) */
+ z = 1.0f/z;
+ if (hx < 0) {
+ if (((ix-0x3f800000)|yisint) == 0) {
+ z = (z-z)/(z-z); /* (-1)**non-int is NaN */
+ } else if (yisint == 1)
+ z = -z; /* (x<0)**odd = -(|x|**odd) */
+ }
+ return z;
+ }
+
+ sn = 1.0f; /* sign of result */
+ if (hx < 0) {
+ if (yisint == 0) /* (x<0)**(non-int) is NaN */
+ return (x-x)/(x-x);
+ if (yisint == 1) /* (x<0)**(odd int) */
+ sn = -1.0f;
+ }
+
+#if 0
+ /* |y| is huge */
+ if (iy > 0x4d000000) { /* if |y| > 2**27 */
+ /* over/underflow if x is not close to one */
+ if (ix < 0x3f7ffff8)
+ return hy < 0 ? sn*huge*huge : sn*tiny*tiny;
+ if (ix > 0x3f800007)
+ return hy > 0 ? sn*huge*huge : sn*tiny*tiny;
+ /* now |1-x| is tiny <= 2**-20, suffice to compute
+ log(x) by x-x^2/2+x^3/3-x^4/4 */
+ t = ax - 1; /* t has 20 trailing zeros */
+ w = (t*t)*(0.5f - t*(0.333333333333f - t*0.25f));
+ u = ivln2_h*t; /* ivln2_h has 16 sig. bits */
+ v = t*ivln2_l - w*ivln2;
+ t1 = u + v;
+ GET_FLOAT_WORD(is, t1);
+ SET_FLOAT_WORD(t1, is & 0xfffff000);
+ t2 = v - (t1-u);
+ } else
+#endif
+{
+ float s2,s_h,s_l,t_h,t_l;
+ n = 0;
+ /* take care subnormal number */
+ if (ix < 0x00800000) {
+ ax *= two24;
+ n -= 24;
+ GET_FLOAT_WORD(ix, ax);
+ }
+ n += ((ix)>>23) - 0x7f;
+ j = ix & 0x007fffff;
+ /* determine interval */
+ ix = j | 0x3f800000; /* normalize ix */
+ if (j <= 0x1cc471) /* |x|<sqrt(3/2) */
+ k = 0;
+ else if (j < 0x5db3d7) /* |x|<sqrt(3) */
+ k = 1;
+ else {
+ k = 0;
+ n += 1;
+ ix -= 0x00800000;
+ }
+ SET_FLOAT_WORD(ax, ix);
+
+ /* compute s = s_h+s_l = (x-1)/(x+1) or (x-1.5)/(x+1.5) */
+ u = ax - bp[k]; /* bp[0]=1.0, bp[1]=1.5 */
+ v = 1.0f/(ax+bp[k]);
+ s = u*v;
+ s_h = s;
+ GET_FLOAT_WORD(is, s_h);
+ SET_FLOAT_WORD(s_h, is & 0xfffff000);
+ /* t_h=ax+bp[k] High */
+ is = ((ix>>1) & 0xfffff000) | 0x20000000;
+ SET_FLOAT_WORD(t_h, is + 0x00400000 + (k<<21));
+ t_l = ax - (t_h - bp[k]);
+ s_l = v*((u - s_h*t_h) - s_h*t_l);
+ /* compute log(ax) */
+ s2 = s*s;
+ r = s2*s2*(L1+s2*(L2+s2*(L3+s2*(L4+s2*(L5+s2*L6)))));
+ r += s_l*(s_h+s);
+ s2 = s_h*s_h;
+ t_h = 3.0f + s2 + r;
+ GET_FLOAT_WORD(is, t_h);
+ SET_FLOAT_WORD(t_h, is & 0xfffff000);
+ t_l = r - ((t_h - 3.0f) - s2);
+ /* u+v = s*(1+...) */
+ u = s_h*t_h;
+ v = s_l*t_h + t_l*s;
+ /* 2/(3log2)*(s+...) */
+ p_h = u + v;
+ GET_FLOAT_WORD(is, p_h);
+ SET_FLOAT_WORD(p_h, is & 0xfffff000);
+ p_l = v - (p_h - u);
+ z_h = cp_h*p_h; /* cp_h+cp_l = 2/(3*log2) */
+ z_l = cp_l*p_h + p_l*cp+dp_l[k];
+ /* log2(ax) = (s+..)*2/(3*log2) = n + dp_h + z_h + z_l */
+ t = (float)n;
+ t1 = (((z_h + z_l) + dp_h[k]) + t);
+ GET_FLOAT_WORD(is, t1);
+ SET_FLOAT_WORD(t1, is & 0xfffff000);
+ t2 = z_l - (((t1 - t) - dp_h[k]) - z_h);
+ }
+
+ /* split up y into y1+y2 and compute (y1+y2)*(t1+t2) */
+ GET_FLOAT_WORD(is, y);
+ SET_FLOAT_WORD(y1, is & 0xfffff000);
+ p_l = (y-y1)*t1 + y*t2;
+ p_h = y1*t1;
+ z = p_l + p_h;
+ GET_FLOAT_WORD(j, z);
+ if (j > 0x43000000) /* if z > 128 */
+ return sn*huge*huge; /* overflow */
+ else if (j == 0x43000000) { /* if z == 128 */
+ if (p_l + ovt > z - p_h)
+ return sn*huge*huge; /* overflow */
+ } else if ((j&0x7fffffff) > 0x43160000) /* z < -150 */ // FIXME: check should be (uint32_t)j >
0xc3160000
+ return sn*tiny*tiny; /* underflow */
+ else if (j == 0xc3160000) { /* z == -150 */
+ if (p_l <= z-p_h)
+ return sn*tiny*tiny; /* underflow */
+ }
+ /*
+ * compute 2**(p_h+p_l)
+ */
+ i = j & 0x7fffffff;
+ k = (i>>23) - 0x7f;
+ n = 0;
+ if (i > 0x3f000000) { /* if |z| > 0.5, set n = [z+0.5] */
+ n = j + (0x00800000>>(k+1));
+ k = ((n&0x7fffffff)>>23) - 0x7f; /* new k for n */
+ SET_FLOAT_WORD(t, n & ~(0x007fffff>>k));
+ n = ((n&0x007fffff)|0x00800000)>>(23-k);
+ if (j < 0)
+ n = -n;
+ p_h -= t;
+ }
+ t = p_l + p_h;
+ GET_FLOAT_WORD(is, t);
+ SET_FLOAT_WORD(t, is & 0xffff8000);
+ u = t*lg2_h;
+ v = (p_l-(t-p_h))*lg2 + t*lg2_l;
+ z = u + v;
+ w = v - (z - u);
+ t = z*z;
+ t1 = z - t*(P1+t*(P2+t*(P3+t*(P4+t*P5))));
+ r = (z*t1)/(t1-2.0f) - (w+z*w);
+ z = 1.0f - (r - z);
+ GET_FLOAT_WORD(j, z);
+ j += n<<23;
+ if ((j>>23) <= 0) /* subnormal output */
+ z = scalbnf(z, n);
+ else
+ SET_FLOAT_WORD(z, j);
+ return sn*z;
+}
+
+static inline float babl_powf (float x, float y)
+{
+ return expf (y * logf (x));
+}
+
static inline float _babl_trc_gamma_to_linear (const Babl *trc_, float value)
{
BablTRC *trc = (void*)trc_;
- return powf (value, trc->gamma);
+ return babl_powf (value, trc->gamma);
}
static inline float _babl_trc_gamma_from_linear (const Babl *trc_, float value)
{
BablTRC *trc = (void*)trc_;
- return powf (value, 1.0f/trc->gamma);
+ return babl_powf (value, trc->rgamma);
}
static inline float _babl_trc_srgb_to_linear (const Babl *trc_, float value)
[
Date Prev][
Date Next] [
Thread Prev][
Thread Next]
[
Thread Index]
[
Date Index]
[
Author Index]
