4
0
mirror of git://sourceware.org/git/newlib-cygwin.git synced 2025-01-23 23:47:22 +08:00
Kwok Cheung Yeung e18743072b amdgcn: Add vectorized math routines
This implements a set of vectorized math routines to be used by the
compiler auto-vectorizer.  Versions for vectors with 2 lanes up to
64 lanes (in powers of 2) are provided.

These routines are based on the scalar versions of the math routines in
libm/common, libm/math and libm/mathfp.  They make extensive use of the GCC
C vector extensions and GCN-specific builtins in GCC.
2023-01-18 13:22:58 -05:00

81 lines
2.8 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) 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.
* ====================================================
*/
/* Based on newlib/libm/common/s_modf.c in Newlib. */
#include "amdgcnmach.h"
v64si v64df_numtest (v64df);
DEF_VD_MATH_FUNC (v64df, modf, v64df x, v64df *iptr)
{
FUNCTION_INIT (v64df);
v64df ret_i;
v64si i0, i1;
EXTRACT_WORDS(i0, i1, x);
v64si j0 = ((i0 >> 20) & 0x7ff) - 0x3ff; /* exponent of x */
v64df zero;
v64si i;
INSERT_WORDS (zero, i0 & 0x80000000, VECTOR_INIT (0), NO_COND);
VECTOR_IF (j0 < 20, cond) /* integer part in x*/
VECTOR_IF2 (j0 < 0, cond2, cond) /* |x|<1 */
VECTOR_COND_MOVE (ret_i, zero, cond2);
VECTOR_RETURN (x, cond2);
VECTOR_ELSE2 (cond2, cond)
i = (0x000fffff) >> j0;
VECTOR_IF2 (((i0 & i) | i1) == 0, cond3, cond2) /* x is integral */
VECTOR_COND_MOVE (ret_i, x, cond3);
VECTOR_RETURN (zero, cond3);
VECTOR_ELSE2 (cond3, cond2)
INSERT_WORDS (ret_i, i0 & ~i, VECTOR_INIT (0), cond3);
VECTOR_RETURN (x - ret_i, cond3);
VECTOR_ENDIF
VECTOR_ENDIF
VECTOR_ELSEIF (j0 > 51, cond) /* no fraction part */
VECTOR_COND_MOVE (ret_i, x, cond);
VECTOR_IF2 (v64df_numtest (x) == NAN, cond2, cond)
VECTOR_COND_MOVE (ret_i, x + x, cond2);
VECTOR_RETURN (ret_i, cond2); /* x is NaN, return NaN */
VECTOR_ENDIF
VECTOR_RETURN (zero, cond); /* return +- 0 */
VECTOR_ELSE (cond)
i = 0xffffffff >> (j0 - 20);
VECTOR_IF2 ((i1 & i) == 0, cond2, cond)
VECTOR_COND_MOVE (ret_i, x, cond2);
INSERT_WORDS (x, i0 & 0x80000000, VECTOR_INIT (0), cond2);
VECTOR_RETURN (x, cond2);
VECTOR_ELSE2 (cond2, cond)
INSERT_WORDS (ret_i, i0, i1 & ~i, cond2);
VECTOR_RETURN (x - ret_i, cond2);
VECTOR_ENDIF
VECTOR_ENDIF
*iptr = ret_i;
FUNCTION_RETURN;
}