diff options
Diffstat (limited to 'newlib/libm/machine/spu/headers/log2d2.h')
| -rw-r--r-- | newlib/libm/machine/spu/headers/log2d2.h | 145 |
1 files changed, 145 insertions, 0 deletions
diff --git a/newlib/libm/machine/spu/headers/log2d2.h b/newlib/libm/machine/spu/headers/log2d2.h new file mode 100644 index 000000000..9bdbb772d --- /dev/null +++ b/newlib/libm/machine/spu/headers/log2d2.h @@ -0,0 +1,145 @@ +/* -------------------------------------------------------------- */ +/* (C)Copyright 2006,2007, */ +/* International Business Machines Corporation, */ +/* Sony Computer Entertainment, Incorporated, */ +/* Toshiba 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. */ +/* 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 _LOG2D2_H_ +#define _LOG2D2_H_ 1 + +#include <spu_intrinsics.h> + +/* + * FUNCTION + * vector double _log2d2(vector double x) + * + * DESCRIPTION + * The function _log2d2 computes log base 2 of the input x for each + * of the double word elements of x. The log2 is decomposed + * into two parts, log2 of the exponent and log2 of the + * fraction. The log2 of the fraction is approximated + * using a 21st order polynomial of the form: + * + * __20_ + * \ + * log(x) = x * (1 + \ (Ci * x^i)) + * / + * /____ + * i=0 + * + * for x in the range 0-1. + */ +#define LOG_C00 +#define LOG_C01 +#define LOG_C02 + +static __inline vector double _log2d2(vector double vx) +{ + vec_int4 addval; + vec_ullong2 exp_mask = spu_splats(0x7FF0000000000000ULL); + vec_double2 vy, vxw; + vec_double2 v1 = spu_splats(1.0); + vec_double2 x2, x4, x8, x10, p1, p2; + + /* Extract the fraction component of input by forcing + * its exponent so that input is in the range [1.0, 2.0) + * and then subtract 1.0 to force it in the range + * [0.0, 1.0). + */ + vxw = spu_sub(spu_sel(vx, v1, exp_mask), v1); + + /* Compute the log2 of the exponent as exp - 1023. + */ + addval = spu_add(spu_rlmask((vec_int4)vx, -20), -1023); + + /* Compute the log2 of the fractional component using a 21st + * order polynomial. The polynomial is evaluated in two halves + * to improve efficiency. + */ + p1 = spu_madd(spu_splats(3.61276447184348752E-05), vxw, spu_splats(-4.16662127033480827E-04)); + p2 = spu_madd(spu_splats(-1.43988260692073185E-01), vxw, spu_splats(1.60245637034704267E-01)); + p1 = spu_madd(vxw, p1, spu_splats(2.28193656337578229E-03)); + p2 = spu_madd(vxw, p2, spu_splats(-1.80329036970820794E-01)); + p1 = spu_madd(vxw, p1, spu_splats(-7.93793829370930689E-03)); + p2 = spu_madd(vxw, p2, spu_splats(2.06098446037376922E-01)); + p1 = spu_madd(vxw, p1, spu_splats(1.98461565426430164E-02)); + p2 = spu_madd(vxw, p2, spu_splats(-2.40449108727688962E-01)); + p1 = spu_madd(vxw, p1, spu_splats(-3.84093543662501949E-02)); + p2 = spu_madd(vxw, p2, spu_splats(2.88539004851839364E-01)); + p1 = spu_madd(vxw, p1, spu_splats(6.08335872067172597E-02)); + p2 = spu_madd(vxw, p2, spu_splats(-3.60673760117245982E-01)); + p1 = spu_madd(vxw, p1, spu_splats(-8.27937055456904317E-02)); + p2 = spu_madd(vxw, p2, spu_splats(4.80898346961226595E-01)); + p1 = spu_madd(vxw, p1, spu_splats(1.01392360727236079E-01)); + p2 = spu_madd(vxw, p2, spu_splats(-7.21347520444469934E-01)); + p1 = spu_madd(vxw, p1, spu_splats(-1.16530490533844182E-01)); + p2 = spu_madd(vxw, p2, spu_splats(0.44269504088896339E+00)); + p1 = spu_madd(vxw, p1, spu_splats(1.30009193360025350E-01)); + + x2 = spu_mul(vxw, vxw); + x4 = spu_mul(x2, x2); + x8 = spu_mul(x4, x4); + x10 = spu_mul(x8, x2); + + vy = spu_madd(spu_madd(x10, p1, p2), vxw, vxw); + + /* Add the log2(exponent) and the log2(fraction) to + * compute the final result. + */ + vy = spu_add(vy, spu_extend(spu_convtf(addval, 0))); + + vxw = spu_extend(spu_convtf(addval, 20)); + + return(vy); +} + +#endif /* _LOG2D2_H_ */ +#endif /* __SPU__ */ |