diff options
author | Ulrich Drepper <drepper@redhat.com> | 2004-05-26 04:47:00 +0000 |
---|---|---|
committer | Ulrich Drepper <drepper@redhat.com> | 2004-05-26 04:47:00 +0000 |
commit | ffdd5e50e18b0cb212acad135e421d932cf3d3a2 (patch) | |
tree | 0a2b8dfc93270b99dbbb314cc68152ea2c5b230a /sysdeps/powerpc/fpu/w_sqrtf.c | |
parent | a752d0cc542a891a086d486654a43212f1821360 (diff) | |
download | glibc-ffdd5e50e18b0cb212acad135e421d932cf3d3a2.zip glibc-ffdd5e50e18b0cb212acad135e421d932cf3d3a2.tar.gz glibc-ffdd5e50e18b0cb212acad135e421d932cf3d3a2.tar.bz2 |
Update.
2004-05-25 Steven Munroe <sjmunroe@us.ibm.com>
* sysdeps/powerpc/fpu/Makefile: Make ld.so a dependency of libm.so.
* sysdeps/powerpc/fpu/bits/mathinline.h [__LIBC_INERNAL_MATH_INLINES]
(__ieee754_sqrt): Define as __MATH_INLINE using fsqrt instruction.
(__ieee754_sqrtf): Define as __MATH_INLINE using fsqrts instruction.
* sysdeps/powerpc/fpu/e_sqrt.c (__slow_ieee754_sqrt): Moved
implementation from w_sqrt.c.
* sysdeps/powerpc/fpu/e_sqrtf.c (__slow_ieee754_sqrtf): Moved
implementation from w_sqrtf.c.
* sysdeps/powerpc/fpu/w_sqrt.c (__sqrt): Wrapper implementation
using inline __ieee754_sqrt().
* sysdeps/powerpc/fpu/w_sqrtf.c (__sqrtf): Wrapper implementation
using inline __ieee754_sqrtf().
* sysdeps/powerpc/powerpc32/sysdep.h [__ASSEMBLER__]: Include
<sysdeps/powerpc/sysdep.h> independent of __ASSEMBLER__.
* sysdeps/powerpc/sysdep.h [__ASSEMBLER__] (PPC_FEATURE_*): Define
PPC_FEATURE_* independent of __ASSEMBLER__.
2004-05-25 Jakub Jelinek <jakub@redhat.com>
* sysdeps/pthread/aio_notify.c: Use <> instead of "" for aio_misc.h
include.
(aio_start_notify_thread): Define if not defined.
(notify_func_wrapper): Use it.
* sysdeps/pthread/aio_misc.c: Use <> instead of "" for aio_misc.h
include.
(aio_create_helper_thread): Define if not defined.
(__aio_create_helper_thread): New function.
(__aio_enqueue_request): Use aio_create_helper_thread.
* nis/ypclnt.c (ypall_data, ypall_foreach): Remove.
(struct ypresp_all_data): New type.
(__xdr_ypresp_all): Change second argument to
struct ypresp_all_data *. Replace ypall_foreach and
ypall_data with objp->foreach and objp->data.
(yp_all): Remove status variable, add data. Replace
all uses of status with data.status. Initialize data.foreach
and data.data instead of ypall_foreach and ypall_data.
2004-05-24 Jakub Jelinek <jakub@redhat.com>
* elf/dl-lookup.c (add_dependency): Set DF_1_NODELETE bit
in l_flags_1, not in l_flags.
Diffstat (limited to 'sysdeps/powerpc/fpu/w_sqrtf.c')
-rw-r--r-- | sysdeps/powerpc/fpu/w_sqrtf.c | 138 |
1 files changed, 28 insertions, 110 deletions
diff --git a/sysdeps/powerpc/fpu/w_sqrtf.c b/sysdeps/powerpc/fpu/w_sqrtf.c index 8eb94d8..e3f3c99 100644 --- a/sysdeps/powerpc/fpu/w_sqrtf.c +++ b/sysdeps/powerpc/fpu/w_sqrtf.c @@ -1,5 +1,5 @@ -/* Single-precision floating point square root. - Copyright (C) 1997, 2003 Free Software Foundation, Inc. +/* Single-precision floating point square root wrapper. + Copyright (C) 2004 Free Software Foundation, Inc. This file is part of the GNU C Library. The GNU C Library is free software; you can redistribute it and/or @@ -17,120 +17,38 @@ Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA. */ -#include <math.h> -#include <math_private.h> +#include "math.h" +#include "math_private.h" #include <fenv_libc.h> -#include <inttypes.h> -static const float almost_half = 0.50000006; /* 0.5 + 2^-24 */ -static const ieee_float_shape_type a_nan = { .word = 0x7fc00000 }; -static const ieee_float_shape_type a_inf = { .word = 0x7f800000 }; -static const float two48 = 281474976710656.0; -static const float twom24 = 5.9604644775390625e-8; -extern const float __t_sqrt[1024]; +#include <sysdep.h> +#include <ldsodefs.h> +#include <dl-procinfo.h> -/* The method is based on a description in - Computation of elementary functions on the IBM RISC System/6000 processor, - P. W. Markstein, IBM J. Res. Develop, 34(1) 1990. - Basically, it consists of two interleaved Newton-Rhapson approximations, - one to find the actual square root, and one to find its reciprocal - without the expense of a division operation. The tricky bit here - is the use of the POWER/PowerPC multiply-add operation to get the - required accuracy with high speed. - - The argument reduction works by a combination of table lookup to - obtain the initial guesses, and some careful modification of the - generated guesses (which mostly runs on the integer unit, while the - Newton-Rhapson is running on the FPU). */ +#ifdef __STDC__ float -__sqrtf(float x) -{ - const float inf = a_inf.value; - /* x = f_washf(x); *//* This ensures only one exception for SNaN. */ - if (x > 0) - { - if (x != inf) - { - /* Variables named starting with 's' exist in the - argument-reduced space, so that 2 > sx >= 0.5, - 1.41... > sg >= 0.70.., 0.70.. >= sy > 0.35... . - Variables named ending with 'i' are integer versions of - floating-point values. */ - float sx; /* The value of which we're trying to find the square - root. */ - float sg,g; /* Guess of the square root of x. */ - float sd,d; /* Difference between the square of the guess and x. */ - float sy; /* Estimate of 1/2g (overestimated by 1ulp). */ - float sy2; /* 2*sy */ - float e; /* Difference between y*g and 1/2 (note that e==se). */ - float shx; /* == sx * fsg */ - float fsg; /* sg*fsg == g. */ - fenv_t fe; /* Saved floating-point environment (stores rounding - mode and whether the inexact exception is - enabled). */ - uint32_t xi, sxi, fsgi; - const float *t_sqrt; - - GET_FLOAT_WORD (xi, x); - fe = fegetenv_register (); - relax_fenv_state (); - sxi = (xi & 0x3fffffff) | 0x3f000000; - SET_FLOAT_WORD (sx, sxi); - t_sqrt = __t_sqrt + (xi >> (23-8-1) & 0x3fe); - sg = t_sqrt[0]; - sy = t_sqrt[1]; - - /* Here we have three Newton-Rhapson iterations each of a - division and a square root and the remainder of the - argument reduction, all interleaved. */ - sd = -(sg*sg - sx); - fsgi = (xi + 0x40000000) >> 1 & 0x7f800000; - sy2 = sy + sy; - sg = sy*sd + sg; /* 16-bit approximation to sqrt(sx). */ - e = -(sy*sg - almost_half); - SET_FLOAT_WORD (fsg, fsgi); - sd = -(sg*sg - sx); - sy = sy + e*sy2; - if ((xi & 0x7f800000) == 0) - goto denorm; - shx = sx * fsg; - sg = sg + sy*sd; /* 32-bit approximation to sqrt(sx), - but perhaps rounded incorrectly. */ - sy2 = sy + sy; - g = sg * fsg; - e = -(sy*sg - almost_half); - d = -(g*sg - shx); - sy = sy + e*sy2; - fesetenv_register (fe); - return g + sy*d; - denorm: - /* For denormalised numbers, we normalise, calculate the - square root, and return an adjusted result. */ - fesetenv_register (fe); - return __sqrtf(x * two48) * twom24; - } - } - else if (x < 0) - { -#ifdef FE_INVALID_SQRT - feraiseexcept (FE_INVALID_SQRT); - /* For some reason, some PowerPC processors don't implement - FE_INVALID_SQRT. I guess no-one ever thought they'd be - used for square roots... :-) */ - if (!fetestexcept (FE_INVALID)) +__sqrtf (float x) /* wrapper sqrtf */ +#else +float +__sqrtf (x) /* wrapper sqrtf */ + float x; #endif - feraiseexcept (FE_INVALID); -#ifndef _IEEE_LIBM - if (_LIB_VERSION != _IEEE_) - x = __kernel_standard(x,x,126); - else +{ +#ifdef _IEEE_LIBM + return __ieee754_sqrtf (x); +#else + float z; + z = __ieee754_sqrtf (x); + + if (_LIB_VERSION == _IEEE_ || (x != x)) + return z; + + if (x < (float) 0.0) + /* sqrtf(negative) */ + return (float) __kernel_standard ((double) x, (double) x, 126); + else + return z; #endif - x = a_nan.value; - } - return f_washf(x); } weak_alias (__sqrtf, sqrtf) -/* Strictly, this is wrong, but the only places where _ieee754_sqrt is - used will not pass in a negative result. */ -strong_alias(__sqrtf,__ieee754_sqrtf) |