mirror/glibc
2
0
Fork 0
mirror of git://sourceware.org/git/glibc.git synced 2024-12-03 04:01:43 +08:00
Code Issues Packages Projects Releases Wiki Activity

i386: Move hypot implementation to C

Browse Source 
The generic hypotf is slight slower, mostly due the tricks the assembly
does to optimize the isinf/isnan/issignaling.  The generic hypot is way
slower, since the optimized implementation uses the i386 default
excessive precision to issue the operation directly.  A similar
implementation is provided instead of using the generic implementation:

Checked on i686-linux-gnu.
This commit is contained in:
Adhemerval Zanella 2021-04-06 12:32:06 -03:00
parent c212d6397e
commit a1d3c9b642
3 changed files with 48 additions and 139 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

75
sysdeps/i386/fpu/e_hypot.S
View File

@ -1,75 +0,0 @@
/* Compute the hypothenuse of X and Y.
Copyright (C) 1998-2021 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
modify it under the terms of the GNU Lesser General Public
License as published by the Free Software Foundation; either
version 2.1 of the License, or (at your option) any later version.
The GNU C Library is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
Lesser General Public License for more details.
You should have received a copy of the GNU Lesser General Public
License along with the GNU C Library; if not, see
<https://www.gnu.org/licenses/>. */
#include <sysdep.h>
#include <i386-math-asm.h>
#include <libm-alias-finite.h>
DEFINE_DBL_MIN
#ifdef PIC
# define MO(op) op##@GOTOFF(%edx)
#else
# define MO(op) op
#endif
.text
ENTRY(__ieee754_hypot)
#ifdef PIC
LOAD_PIC_REG (dx)
#endif
fldl 4(%esp) // x
fxam
fnstsw
fldl 12(%esp) // y : x
movb %ah, %ch
fxam
fnstsw
movb %ah, %al
orb %ch, %ah
sahf
jc 1f
fmul %st(0) // y * y : x
fxch // x : y * y
fmul %st(0) // x * x : y * y
faddp // x * x + y * y
fsqrt
DBL_NARROW_EVAL_UFLOW_NONNEG
2: ret
// We have to test whether any of the parameters is Inf.
// In this case the result is infinity.
1: andb $0x45, %al
cmpb $5, %al
je 3f // jump if y is Inf
andb $0x45, %ch
cmpb $5, %ch
jne 4f // jump if x is not Inf
fxch
3: fstp %st(1)
fabs
jmp 2b
4: testb $1, %al
jnz 5f // y is NaN
fxch
5: fstp %st(1)
jmp 2b
END(__ieee754_hypot)
libm_alias_finite (__ieee754_hypot, __hypot)

48
sysdeps/i386/fpu/e_hypot.c Normal file
View File

@ -0,0 +1,48 @@
/* Euclidean distance function. Double/Binary64 i386 version.
Copyright (C) 2021 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
modify it under the terms of the GNU Lesser General Public
License as published by the Free Software Foundation; either
version 2.1 of the License, or (at your option) any later version.
The GNU C Library is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
Lesser General Public License for more details.
You should have received a copy of the GNU Lesser General Public
License along with the GNU C Library; if not, see
<https://www.gnu.org/licenses/>. */
#include <math.h>
#include <math_private.h>
#include <math-narrow-eval.h>
#include <math-underflow.h>
#include <math-svid-compat.h>
#include <libm-alias-finite.h>
#include <libm-alias-double.h>
#include <errno.h>
/* The i386 allows to use the default excess of precision to optimize the
hypot implementation, since internal multiplication and sqrt is carried
with 80-bit FP type. */
double
__ieee754_hypot (double x, double y)
{
if (!isfinite (x) || !isfinite (y))
{
if ((isinf (x) || isinf (y))
&& !issignaling (x) && !issignaling (y))
return INFINITY;
return x + y;
}
long double lx = x;
long double ly = y;
double r = math_narrow_eval ((double) sqrtl (lx * lx + ly * ly));
math_check_force_underflow_nonneg (r);
return r;
}
libm_alias_finite (__ieee754_hypot, __hypot)

64
sysdeps/i386/fpu/e_hypotf.S
View File

@ -1,64 +0,0 @@
/* Compute the hypothenuse of X and Y.
Copyright (C) 1998-2021 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
modify it under the terms of the GNU Lesser General Public
License as published by the Free Software Foundation; either
version 2.1 of the License, or (at your option) any later version.
The GNU C Library is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
Lesser General Public License for more details.
You should have received a copy of the GNU Lesser General Public
License along with the GNU C Library; if not, see
<https://www.gnu.org/licenses/>. */
#include <sysdep.h>
#include <i386-math-asm.h>
#include <libm-alias-finite.h>
.text
ENTRY(__ieee754_hypotf)
flds 4(%esp) // x
fxam
fnstsw
flds 8(%esp) // y : x
movb %ah, %ch
fxam
fnstsw
movb %ah, %al
orb %ch, %ah
sahf
jc 1f
fmul %st(0) // y * y : x
fxch // x : y * y
fmul %st(0) // x * x : y * y
faddp // x * x + y * y
fsqrt
FLT_NARROW_EVAL
2: ret
// We have to test whether any of the parameters is Inf.
// In this case the result is infinity.
1: andb $0x45, %al
cmpb $5, %al
je 3f // jump if y is Inf
andb $0x45, %ch
cmpb $5, %ch
jne 4f // jump if x is not Inf
fxch
3: fstp %st(1)
fabs
jmp 2b
4: testb $1, %al
jnz 5f // y is NaN
fxch
5: fstp %st(1)
jmp 2b
END(__ieee754_hypotf)
libm_alias_finite (__ieee754_hypotf, __hypotf)