newlib-cygwin/newlib/libm/machine/spu/headers/expm1d2.h

136 lines
5.6 KiB
C

/* -------------------------------------------------------------- */
/* (C)Copyright 2007,2008, */
/* International Business Machines Corporation */
/* All Rights Reserved. */
/* */
/* Redistribution and use in source and binary forms, with or */
/* without modification, are permitted provided that the */
/* following conditions are met: */
/* */
/* - Redistributions of source code must retain the above copyright*/
/* notice, this list of conditions and the following disclaimer. */
/* */
/* - Redistributions in binary form must reproduce the above */
/* copyright notice, this list of conditions and the following */
/* disclaimer in the documentation and/or other materials */
/* provided with the distribution. */
/* */
/* - Neither the name of IBM Corporation nor the names of its */
/* contributors may be used to endorse or promote products */
/* derived from this software without specific prior written */
/* permission. */
/* */
/* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND */
/* CONTRIBUTORS "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, */
/* INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF */
/* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE */
/* DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR */
/* CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, */
/* SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT */
/* NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; */
/* LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) */
/* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN */
/* CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR */
/* OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, */
/* EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */
/* -------------------------------------------------------------- */
/* PROLOG END TAG zYx */
#ifdef __SPU__
#ifndef _EXPM1D2_H_
#define _EXPM1D2_H_ 1
#include <spu_intrinsics.h>
#include "expd2.h"
#include "divd2.h"
#define EXPM1_P0 0.0000000000000000000000000e+00
#define EXPM1_P1 1.0000000000000000000000000e+00
#define EXPM1_P2 9.7234232565378004697204117e-04
#define EXPM1_P3 3.3328278237299953368211192e-02
#define EXPM1_P4 3.1156225044634678993365345e-05
#define EXPM1_P5 2.1352206553343212164751408e-04
#define EXPM1_P6 1.6975135794626144795757452e-07
#define EXPM1_P7 2.7686287801334994383131629e-07
#define EXPM1_P8 1.1186114936216450015354379e-10
#define EXPM1_Q0 1.0000000000000000000000000e+00
#define EXPM1_Q1 -4.9902765767434620336473472e-01
#define EXPM1_Q2 1.1617544040780639069687652e-01
#define EXPM1_Q3 -1.6551954366467523660499950e-02
#define EXPM1_Q4 1.5864115838972218334307351e-03
#define EXPM1_Q5 -1.0534540477401370666288988e-04
#define EXPM1_Q6 4.7650003993592160620959043e-06
#define EXPM1_Q7 -1.3529198871087017840776265e-07
#define EXPM1_Q8 1.8635779407675460757658020e-09
/*
* FUNCTION
* vector double _expm1d2(vector double x)
*
* DESCRIPTION
* _expm1d2 computes the exponential - 1 for each element
* of the input vector x.
*
* This function is intended to return accurate values, even
* where exp(x) - 1 would normally produce bad results due to
* floating-point cancellation errors.
*
*/
static __inline vector double _expm1d2(vector double x)
{
vector double oned = spu_splats(1.0);
vector double range = spu_splats(1.0625);
vector unsigned long long use_exp;
vector double pr, qr;
vector double eresult;
vector double rresult;
vector double result;
/* Compiler Bug. Replace xbug with x when spu_cmp*() doesn't
* modify it's arguments! */
volatile vector double xbug = x;
use_exp = spu_cmpabsgt(xbug, range);
/*
* Calculate directly using exp(x) - 1
*/
eresult = spu_sub(_expd2(x), oned);
/*
* For x in [-1.0625,1.0625], use a rational approximation.
* The madd's are interleaved to reduce dependency stalls. Looks
* like gcc is smart enough to do this on it's own... but why
* take the chance.
*/
pr = spu_madd(x, spu_splats(EXPM1_P8), spu_splats(EXPM1_P7));
qr = spu_madd(x, spu_splats(EXPM1_Q8), spu_splats(EXPM1_Q7));
pr = spu_madd(pr, x, spu_splats(EXPM1_P6));
qr = spu_madd(qr, x, spu_splats(EXPM1_Q6));
pr = spu_madd(pr, x, spu_splats(EXPM1_P5));
qr = spu_madd(qr, x, spu_splats(EXPM1_Q5));
pr = spu_madd(pr, x, spu_splats(EXPM1_P4));
qr = spu_madd(qr, x, spu_splats(EXPM1_Q4));
pr = spu_madd(pr, x, spu_splats(EXPM1_P3));
qr = spu_madd(qr, x, spu_splats(EXPM1_Q3));
pr = spu_madd(pr, x, spu_splats(EXPM1_P2));
qr = spu_madd(qr, x, spu_splats(EXPM1_Q2));
pr = spu_madd(pr, x, spu_splats(EXPM1_P1));
qr = spu_madd(qr, x, spu_splats(EXPM1_Q1));
pr = spu_madd(pr, x, spu_splats(EXPM1_P0));
qr = spu_madd(qr, x, spu_splats(EXPM1_Q0));
rresult = _divd2(pr, qr);
/*
* Select either direct calculation or rational approximation.
*/
result = spu_sel(rresult, eresult, use_exp);
return result;
}
#endif /* _EXPM1D2_H_ */
#endif /* __SPU__ */