128 lines
4.0 KiB
C
128 lines
4.0 KiB
C
/*
|
|
* Copyright 2023 Siemens
|
|
*
|
|
* The authors hereby grant permission to use, copy, modify, distribute,
|
|
* and license this software and its documentation for any purpose, provided
|
|
* that existing copyright notices are retained in all copies and that this
|
|
* notice is included verbatim in any distributions. No written agreement,
|
|
* license, or royalty fee is required for any of the authorized uses.
|
|
* Modifications to this software may be copyrighted by their authors
|
|
* and need not follow the licensing terms described here, provided that
|
|
* the new terms are clearly indicated on the first page of each file where
|
|
* they apply.
|
|
*/
|
|
|
|
/*
|
|
* Copyright (c) 1994-2009 Red Hat, Inc. All rights reserved.
|
|
*
|
|
* This copyrighted material is made available to anyone wishing to use,
|
|
* modify, copy, or redistribute it subject to the terms and conditions
|
|
* of the BSD License. This program is distributed in the hope that
|
|
* it will be useful, but WITHOUT ANY WARRANTY expressed or implied,
|
|
* including the implied warranties of MERCHANTABILITY or FITNESS FOR
|
|
* A PARTICULAR PURPOSE. A copy of this license is available at
|
|
* http://www.opensource.org/licenses. Any Red Hat trademarks that are
|
|
* incorporated in the source code or documentation are not subject to
|
|
* the BSD License and may only be used or replicated with the express
|
|
* permission of Red Hat, Inc.
|
|
*/
|
|
|
|
/******************************************************************
|
|
* The following routines are coded directly from the algorithms
|
|
* and coefficients given in "Software Manual for the Elementary
|
|
* Functions" by William J. Cody, Jr. and William Waite, Prentice
|
|
* Hall, 1980.
|
|
******************************************************************/
|
|
|
|
/* Based in newlib/libm/mathfp/sf_sineh.c in Newlib. */
|
|
|
|
#include "amdgcnmach.h"
|
|
|
|
v64sf v64sf_expf_aux (v64sf, v64si);
|
|
v64si v64sf_numtestf (v64sf);
|
|
v64si v64sf_isposf (v64sf);
|
|
|
|
static const float q[] = { -0.428277109e+2 };
|
|
static const float p[] = { -0.713793159e+1,
|
|
-0.190333399 };
|
|
static const float LNV = 0.6931610107;
|
|
static const float INV_V2 = 0.2499930850;
|
|
static const float V_OVER2_MINUS1 = 0.1383027787e-4;
|
|
|
|
#if defined (__has_builtin) && __has_builtin (__builtin_gcn_fabsvf)
|
|
|
|
DEF_VS_MATH_FUNC (v64sf, sinehf, v64sf x, int cosineh)
|
|
{
|
|
const float WBAR = 18.55;
|
|
|
|
FUNCTION_INIT (v64sf);
|
|
|
|
v64si sgn = VECTOR_INIT (0);
|
|
v64si v_cosineh = VECTOR_INIT (cosineh ? -1 : 0);
|
|
|
|
/* Check for special values. */
|
|
v64si num_type = v64sf_numtestf (x);
|
|
VECTOR_IF (num_type == NAN, cond)
|
|
errno = EDOM;
|
|
VECTOR_RETURN (x, cond);
|
|
VECTOR_ELSEIF (num_type == INF, cond)
|
|
errno = ERANGE;
|
|
VECTOR_RETURN (VECTOR_MERGE (VECTOR_INIT (z_infinity_f.f),
|
|
VECTOR_INIT (-z_infinity_f.f),
|
|
v64sf_isposf (x)),
|
|
cond);
|
|
VECTOR_ENDIF
|
|
|
|
v64sf y = __builtin_gcn_fabsvf (x);
|
|
|
|
if (!cosineh)
|
|
VECTOR_COND_MOVE (sgn, VECTOR_INIT (-1), x < 0.0f);
|
|
|
|
v64sf res;
|
|
|
|
VECTOR_IF (((y > 1.0f) & ~v_cosineh) | v_cosineh, cond)
|
|
VECTOR_IF2 (y > (float) BIGX, cond2, cond)
|
|
v64sf w = y - LNV;
|
|
|
|
/* Check for w > maximum here. */
|
|
VECTOR_IF2 (w > (float) BIGX, cond3, cond2)
|
|
errno = ERANGE;
|
|
VECTOR_RETURN (x, cond3);
|
|
VECTOR_ENDIF
|
|
|
|
v64sf z = v64sf_expf_aux (w, __mask);
|
|
|
|
VECTOR_COND_MOVE (res, z * (V_OVER2_MINUS1 + 1.0f),
|
|
cond2 & (w > WBAR));
|
|
VECTOR_ELSE2 (cond2, cond)
|
|
v64sf z = v64sf_expf_aux (y, __mask);
|
|
if (cosineh) {
|
|
VECTOR_COND_MOVE (res, (z + 1 / z) * 0.5f, cond2);
|
|
} else {
|
|
VECTOR_COND_MOVE (res, (z - 1 / z) * 0.5f, cond2);
|
|
}
|
|
VECTOR_ENDIF
|
|
|
|
VECTOR_COND_MOVE (res, -res, sgn);
|
|
VECTOR_ELSE (cond)
|
|
/* Check for y being too small. */
|
|
VECTOR_IF2 (y < z_rooteps_f, cond2, cond);
|
|
VECTOR_COND_MOVE (res, x, cond2);
|
|
VECTOR_ELSE2 (cond2, cond)
|
|
/* Calculate the Taylor series. */
|
|
v64sf f = x * x;
|
|
v64sf Q = f + q[0];
|
|
v64sf P = p[1] * f + p[0];
|
|
v64sf R = f * (P / Q);
|
|
|
|
VECTOR_COND_MOVE (res, x + x * R, cond2);
|
|
VECTOR_ENDIF
|
|
VECTOR_ENDIF
|
|
|
|
VECTOR_RETURN (res, NO_COND);
|
|
|
|
FUNCTION_RETURN;
|
|
}
|
|
|
|
#endif
|