Szabolcs Nagy
e216951f50
When FLT_EVAL_METHOD!=0 (only i386 with x87 fp) the excess
precision of an expression must be removed in an assignment.
(gcc needs -fexcess-precision=standard or -std=c99 for this)
This is done by extra load/store instructions which adds code
bloat when lot of temporaries are used and it makes the result
less precise in many cases.
Using double_t and float_t avoids these issues on i386 and
it makes no difference on other archs.
For now only a few functions are modified where the excess
precision is clearly beneficial (mostly polynomial evaluations
with temporaries).
object size differences on i386, gcc-4.8:
old new
__cosdf.o 123 95
__cos.o 199 169
__sindf.o 131 95
__sin.o 225 203
__tandf.o 207 151
__tan.o 605 499
erff.o 1470 1416
erf.o 1703 1649
j0f.o 1779 1745
j0.o 2308 2274
j1f.o 1602 1568
j1.o 2286 2252
tgamma.o 1431 1424
math/*.o 64164 63635
95 lines
3.5 KiB
C
95 lines
3.5 KiB
C
/* origin: FreeBSD /usr/src/lib/msun/src/k_log.h */
|
|
/*
|
|
* ====================================================
|
|
* Copyright (C) 1993 by Sun Microsystems, Inc. All rights reserved.
|
|
*
|
|
* Developed at SunSoft, a Sun Microsystems, Inc. business.
|
|
* Permission to use, copy, modify, and distribute this
|
|
* software is freely granted, provided that this notice
|
|
* is preserved.
|
|
* ====================================================
|
|
*/
|
|
/*
|
|
* __log1p(f):
|
|
* Return log(1+f) - f for 1+f in ~[sqrt(2)/2, sqrt(2)].
|
|
*
|
|
* The following describes the overall strategy for computing
|
|
* logarithms in base e. The argument reduction and adding the final
|
|
* term of the polynomial are done by the caller for increased accuracy
|
|
* when different bases are used.
|
|
*
|
|
* Method :
|
|
* 1. Argument Reduction: find k and f such that
|
|
* x = 2^k * (1+f),
|
|
* where sqrt(2)/2 < 1+f < sqrt(2) .
|
|
*
|
|
* 2. Approximation of log(1+f).
|
|
* Let s = f/(2+f) ; based on log(1+f) = log(1+s) - log(1-s)
|
|
* = 2s + 2/3 s**3 + 2/5 s**5 + .....,
|
|
* = 2s + s*R
|
|
* We use a special Reme algorithm on [0,0.1716] to generate
|
|
* a polynomial of degree 14 to approximate R The maximum error
|
|
* of this polynomial approximation is bounded by 2**-58.45. In
|
|
* other words,
|
|
* 2 4 6 8 10 12 14
|
|
* R(z) ~ Lg1*s +Lg2*s +Lg3*s +Lg4*s +Lg5*s +Lg6*s +Lg7*s
|
|
* (the values of Lg1 to Lg7 are listed in the program)
|
|
* and
|
|
* | 2 14 | -58.45
|
|
* | Lg1*s +...+Lg7*s - R(z) | <= 2
|
|
* | |
|
|
* Note that 2s = f - s*f = f - hfsq + s*hfsq, where hfsq = f*f/2.
|
|
* In order to guarantee error in log below 1ulp, we compute log
|
|
* by
|
|
* log(1+f) = f - s*(f - R) (if f is not too large)
|
|
* log(1+f) = f - (hfsq - s*(hfsq+R)). (better accuracy)
|
|
*
|
|
* 3. Finally, log(x) = k*ln2 + log(1+f).
|
|
* = k*ln2_hi+(f-(hfsq-(s*(hfsq+R)+k*ln2_lo)))
|
|
* Here ln2 is split into two floating point number:
|
|
* ln2_hi + ln2_lo,
|
|
* where n*ln2_hi is always exact for |n| < 2000.
|
|
*
|
|
* Special cases:
|
|
* log(x) is NaN with signal if x < 0 (including -INF) ;
|
|
* log(+INF) is +INF; log(0) is -INF with signal;
|
|
* log(NaN) is that NaN with no signal.
|
|
*
|
|
* Accuracy:
|
|
* according to an error analysis, the error is always less than
|
|
* 1 ulp (unit in the last place).
|
|
*
|
|
* Constants:
|
|
* The hexadecimal values are the intended ones for the following
|
|
* constants. The decimal values may be used, provided that the
|
|
* compiler will convert from decimal to binary accurately enough
|
|
* to produce the hexadecimal values shown.
|
|
*/
|
|
|
|
static const double
|
|
Lg1 = 6.666666666666735130e-01, /* 3FE55555 55555593 */
|
|
Lg2 = 3.999999999940941908e-01, /* 3FD99999 9997FA04 */
|
|
Lg3 = 2.857142874366239149e-01, /* 3FD24924 94229359 */
|
|
Lg4 = 2.222219843214978396e-01, /* 3FCC71C5 1D8E78AF */
|
|
Lg5 = 1.818357216161805012e-01, /* 3FC74664 96CB03DE */
|
|
Lg6 = 1.531383769920937332e-01, /* 3FC39A09 D078C69F */
|
|
Lg7 = 1.479819860511658591e-01; /* 3FC2F112 DF3E5244 */
|
|
|
|
/*
|
|
* We always inline __log1p(), since doing so produces a
|
|
* substantial performance improvement (~40% on amd64).
|
|
*/
|
|
static inline double __log1p(double f)
|
|
{
|
|
double_t hfsq,s,z,R,w,t1,t2;
|
|
|
|
s = f/(2.0+f);
|
|
z = s*s;
|
|
w = z*z;
|
|
t1= w*(Lg2+w*(Lg4+w*Lg6));
|
|
t2= z*(Lg1+w*(Lg3+w*(Lg5+w*Lg7)));
|
|
R = t2+t1;
|
|
hfsq = 0.5*f*f;
|
|
return s*(hfsq+R);
|
|
}
|