math: Use asinhf from CORE-MATH

The CORE-MATH implementation is correctly rounded (for any rounding mode)
and shows slight better performance to the generic asinhf.

The code was adapted to glibc style and to use the definition of
math_config.h (to handle errno, overflow, and underflow).

Benchtest on x64_64 (Ryzen 9 5900X, gcc 14.2.1), aarch64 (Neoverse-N1,
gcc 13.3.1), and powerpc (POWER10, gcc 13.2.1):

Latency                      master        patched   improvement
x86_64                      64.5128        56.9717        11.69%
x86_64v2                    63.3065        57.2666         9.54%
x86_64v3                    62.8719        51.4170        18.22%
i686                       189.1630        137.635        27.24%
aarch64 (Neoverse)          25.3551        20.5757        18.85%
power10                     17.9712        13.3302        25.82%

reciprocal-throughput        master        patched   improvement
x86_64                      20.0844        15.4731        22.96%
x86_64v2                    19.2919        15.4000        20.17%
x86_64v3                    18.7226        11.9009        36.44%
i686                       103.7670        80.2681        22.65%
aarch64 (Neoverse)          12.5005        8.68969        30.49%
power10                      7.2220        5.03617        30.27%

Signed-off-by: Alexei Sibidanov <sibid@uvic.ca>
Signed-off-by: Paul Zimmermann <Paul.Zimmermann@inria.fr>
Signed-off-by: Adhemerval Zanella <adhemerval.zanella@linaro.org>:
Reviewed-by: DJ Delorie <dj@redhat.com>

diff --git a/SHARED-FILES b/SHARED-FILES
index 6a6ff76b31c3da1a191148213b3462031118bd9d..18b3244e44791e3981b407be09d6f73f72651da7 100644 (file)
--- a/SHARED-FILES
+++ b/SHARED-FILES
@@ -306,3 +306,7 @@ sysdeps/ieee754/flt-32/e_asinf.c:
   (src/binary32/asin/asinf.c in CORE-MATH)
   - The code was adapted to use glibc code style and internal
     functions to handle errno, overflow, and underflow.
+sysdeps/ieee754/flt-32/s_asinhf.c:
+  (src/binary32/asinh/asinhf.c in CORE-MATH)
+  - The code was adapted to use glibc code style and internal
+    functions to handle errno, overflow, and underflow.

diff --git a/sysdeps/aarch64/libm-test-ulps b/sysdeps/aarch64/libm-test-ulps
index da054b62e7560678776bca93fbb28ab4f0a708d0..44493b3f02a1ace49f7b732757788e1edec54c2c 100644 (file)
--- a/sysdeps/aarch64/libm-test-ulps
+++ b/sysdeps/aarch64/libm-test-ulps
@@ -95,7 +95,6 @@ ldouble: 2
 
 Function: "asinh":
 double: 2
-float: 2
 ldouble: 4
 
 Function: "asinh_advsimd":
@@ -104,7 +103,6 @@ float: 2
 
 Function: "asinh_downward":
 double: 3
-float: 3
 ldouble: 4
 
 Function: "asinh_sve":
@@ -113,12 +111,10 @@ float: 2
 
 Function: "asinh_towardzero":
 double: 2
-float: 2
 ldouble: 2
 
 Function: "asinh_upward":
 double: 3
-float: 3
 ldouble: 4
 
 Function: "asinpi":

diff --git a/sysdeps/alpha/fpu/libm-test-ulps b/sysdeps/alpha/fpu/libm-test-ulps
index cda6ece49ea66a4a2d8e45e14ea1561a0b8ec253..708299915de6fb3f1d78715de2cc675b3c0254c2 100644 (file)
--- a/sysdeps/alpha/fpu/libm-test-ulps
+++ b/sysdeps/alpha/fpu/libm-test-ulps
@@ -51,22 +51,18 @@ ldouble: 2
 
 Function: "asinh":
 double: 2
-float: 2
 ldouble: 4
 
 Function: "asinh_downward":
 double: 3
-float: 3
 ldouble: 4
 
 Function: "asinh_towardzero":
 double: 2
-float: 2
 ldouble: 2
 
 Function: "asinh_upward":
 double: 3
-float: 3
 ldouble: 4
 
 Function: "atan":

diff --git a/sysdeps/arc/fpu/libm-test-ulps b/sysdeps/arc/fpu/libm-test-ulps
index 3c08593d9960bb11d50bc1de469b43416ea60e55..1c34bd36d62b326ed2420fa4163255716297f5ca 100644 (file)
--- a/sysdeps/arc/fpu/libm-test-ulps
+++ b/sysdeps/arc/fpu/libm-test-ulps
@@ -39,19 +39,15 @@ double: 2
 
 Function: "asinh":
 double: 3
-float: 2
 
 Function: "asinh_downward":
 double: 3
-float: 3
 
 Function: "asinh_towardzero":
 double: 2
-float: 2
 
 Function: "asinh_upward":
 double: 3
-float: 3
 
 Function: "atan":
 double: 1

diff --git a/sysdeps/arc/nofpu/libm-test-ulps b/sysdeps/arc/nofpu/libm-test-ulps
index c3548a98078b4804b80027e6f565f3a76b4cea7b..58fc499f53b7e0cf4b2816c5ebcded09edc88763 100644 (file)
--- a/sysdeps/arc/nofpu/libm-test-ulps
+++ b/sysdeps/arc/nofpu/libm-test-ulps
@@ -12,7 +12,6 @@ double: 1
 
 Function: "asinh":
 double: 2
-float: 2
 
 Function: "atan":
 double: 1

diff --git a/sysdeps/arm/libm-test-ulps b/sysdeps/arm/libm-test-ulps
index 8e55b23f41d6139b3ac0fe29ad718e4af5bbb76f..60f483ec31739972b5aa3876e112c17e1626df1f 100644 (file)
--- a/sysdeps/arm/libm-test-ulps
+++ b/sysdeps/arm/libm-test-ulps
@@ -55,19 +55,15 @@ double: 2
 
 Function: "asinh":
 double: 2
-float: 2
 
 Function: "asinh_downward":
 double: 3
-float: 3
 
 Function: "asinh_towardzero":
 double: 2
-float: 2
 
 Function: "asinh_upward":
 double: 3
-float: 3
 
 Function: "asinpi":
 double: 1

diff --git a/sysdeps/csky/fpu/libm-test-ulps b/sysdeps/csky/fpu/libm-test-ulps
index b04ed2a161e7a2b6a3fca7d634a5bb5c127d73cb..2b7b5cfc92bd629af1d39b3dfefef272fcd87274 100644 (file)
--- a/sysdeps/csky/fpu/libm-test-ulps
+++ b/sysdeps/csky/fpu/libm-test-ulps
@@ -37,19 +37,15 @@ double: 1
 
 Function: "asinh":
 double: 2
-float: 2
 
 Function: "asinh_downward":
 double: 3
-float: 3
 
 Function: "asinh_towardzero":
 double: 2
-float: 2
 
 Function: "asinh_upward":
 double: 3
-float: 3
 
 Function: "atan":
 float: 1

diff --git a/sysdeps/csky/nofpu/libm-test-ulps b/sysdeps/csky/nofpu/libm-test-ulps
index 612bd19eda0d16aa3127ddb76b479939de11c17b..0eb62de8b204d663ed2217e1246e7fdca0c284eb 100644 (file)
--- a/sysdeps/csky/nofpu/libm-test-ulps
+++ b/sysdeps/csky/nofpu/libm-test-ulps
@@ -37,19 +37,15 @@ double: 1
 
 Function: "asinh":
 double: 2
-float: 2
 
 Function: "asinh_downward":
 double: 3
-float: 3
 
 Function: "asinh_towardzero":
 double: 2
-float: 2
 
 Function: "asinh_upward":
 double: 3
-float: 3
 
 Function: "atan":
 float: 1

diff --git a/sysdeps/hppa/fpu/libm-test-ulps b/sysdeps/hppa/fpu/libm-test-ulps
index bb7e30f25576d31543324621319efd504ad4cc31..f48a12f51129435863ec1b38a90a5beb704ca56e 100644 (file)
--- a/sysdeps/hppa/fpu/libm-test-ulps
+++ b/sysdeps/hppa/fpu/libm-test-ulps
@@ -55,19 +55,15 @@ double: 2
 
 Function: "asinh":
 double: 2
-float: 2
 
 Function: "asinh_downward":
 double: 3
-float: 3
 
 Function: "asinh_towardzero":
 double: 2
-float: 2
 
 Function: "asinh_upward":
 double: 3
-float: 3
 
 Function: "asinpi":
 double: 1

diff --git a/sysdeps/i386/fpu/libm-test-ulps b/sysdeps/i386/fpu/libm-test-ulps
index 11606b93d9988958faf92f6484ab97ae21f33855..669ecbe3ac5c7e23725f7aa75867a5a0053595e4 100644 (file)
--- a/sysdeps/i386/fpu/libm-test-ulps
+++ b/sysdeps/i386/fpu/libm-test-ulps
@@ -90,19 +90,16 @@ ldouble: 3
 
 Function: "asinh_downward":
 double: 1
-float: 1
 float128: 4
 ldouble: 5
 
 Function: "asinh_towardzero":
 double: 1
-float: 1
 float128: 2
 ldouble: 4
 
 Function: "asinh_upward":
 double: 1
-float: 1
 float128: 4
 ldouble: 5
 

diff --git a/sysdeps/i386/fpu/s_asinhf.S b/sysdeps/i386/fpu/s_asinhf.S
deleted file mode 100644 (file)
index fe52e20..0000000
--- a/sysdeps/i386/fpu/s_asinhf.S
+++ /dev/null
@@ -1,139 +0,0 @@
-/* ix87 specific implementation of arcsinh.
-   Copyright (C) 1996-2024 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 <machine/asm.h>
-#include <libm-alias-float.h>
-
-       .section .rodata
-
-       .align ALIGNARG(4)
-       .type huge,@object
-huge:  .double 1e+36
-       ASM_SIZE_DIRECTIVE(huge)
-       .type one,@object
-one:   .double 1.0
-       ASM_SIZE_DIRECTIVE(one)
-       .type limit,@object
-limit: .double 0.29
-       ASM_SIZE_DIRECTIVE(limit)
-
-#ifdef PIC
-#define MO(op) op##@GOTOFF(%edx)
-#else
-#define MO(op) op
-#endif
-
-       .text
-ENTRY(__asinhf)
-       movl    4(%esp), %ecx
-       movl    $0x7fffffff, %eax
-       andl    %ecx, %eax
-       andl    $0x80000000, %ecx
-       movl    %eax, %edx
-       orl     $0x807fffff, %edx
-       incl    %edx
-       jz      7f                      // x in ±Inf or NaN
-       xorl    %ecx, 4(%esp)
-       flds    4(%esp)                 // |x|
-       cmpl    $0x38000000, %eax
-       jb      2f                      // |x| < 2^-14
-       fldln2                          // log(2) : |x|
-       cmpl    $0x47000000, %eax
-       fxch                            // |x| : log(2)
-       ja      3f                      // |x| > 2^14
-#ifdef PIC
-       LOAD_PIC_REG (dx)
-#endif
-       cmpl    $0x40000000, %eax
-       ja      5f                      // |x| > 2
-
-       // 2^-14 <= |x| <= 2 => y = sign(x)*log1p(|x|+|x|^2/(1+sqrt(1+|x|^2)))
-       fld     %st                     // |x| : |x| : log(2)
-       fmul    %st(1)                  // |x|^2 : |x| : log(2)
-       fld     %st                     // |x|^2 : |x|^2 : |x| : log(2)
-       faddl   MO(one)                 // 1+|x|^2 : |x|^2 : |x| : log(2)
-       fsqrt                           // sqrt(1+|x|^2) : |x|^2 : |x| : log(2)
-       faddl   MO(one)                 // 1+sqrt(1+|x|^2) : |x|^2 : |x| : log(2)
-       fdivrp                          // |x|^2/(1+sqrt(1+|x|^2)) : |x| : log(2)
-       faddp                           // |x|+|x|^2/(1+sqrt(1+|x|^2)) : log(2)
-       fcoml   MO(limit)
-       fnstsw
-       sahf
-       ja      6f
-       fyl2xp1
-       jecxz   4f
-       fchs
-4:     ret
-
-7:     flds    4(%esp)
-       ret
-
-6:     faddl   MO(one)
-       fyl2x
-       jecxz   4f
-       fchs
-4:     ret
-
-       // |x| < 2^-14 => y = x (inexact iff |x| != 0.0)
-       .align ALIGNARG(4)
-2:
-#ifdef PIC
-       LOAD_PIC_REG (dx)
-#endif
-       jecxz   4f
-       fchs                            // x
-4:     fld     %st                     // x : x
-       faddl   MO(huge)                // huge+x : x
-       fstp    %st(0)                  // x
-       cmpl    $0x00800000, %eax
-       jae     8f
-       subl    $4, %esp
-       cfi_adjust_cfa_offset (4)
-       fld     %st(0)
-       fmul    %st(0)
-       fstps   (%esp)
-       addl    $4, %esp
-       cfi_adjust_cfa_offset (-4)
-8:     ret
-
-       // |x| > 2^14 => y = sign(x) * (log(|x|) + log(2))
-       .align ALIGNARG(4)
-3:     fyl2x                           // log(|x|)
-       fldln2                          // log(2) : log(|x|)
-       faddp                           // log(|x|)+log(2)
-       jecxz   4f
-       fchs
-4:     ret
-
-       // |x| > 2 => y = sign(x) * log(2*|x| + 1/(|x|+sqrt(x*x+1)))
-       .align ALIGNARG(4)
-5:     fld     %st                     // |x| : |x| : log(2)
-       fadd    %st, %st(1)             // |x| : 2*|x| : log(2)
-       fld     %st                     // |x| : |x| : 2*|x| : log(2)
-       fmul    %st(1)                  // |x|^2 : |x| : 2*|x| : log(2)
-       faddl   MO(one)                 // 1+|x|^2 : |x| : 2*|x| : log(2)
-       fsqrt                           // sqrt(1+|x|^2) : |x| : 2*|x| : log(2)
-       faddp                           // |x|+sqrt(1+|x|^2) : 2*|x| : log(2)
-       fdivrl  MO(one)                 // 1/(|x|+sqrt(1+|x|^2)) : 2*|x| : log(2)
-       faddp                           // 2*|x|+1/(|x|+sqrt(1+|x|^2)) : log(2)
-       fyl2x                           // log(2*|x|+1/(|x|+sqrt(1+|x|^2)))
-       jecxz   4f
-       fchs
-4:     ret
-END(__asinhf)
-libm_alias_float (__asinh, asinh)

diff --git a/sysdeps/i386/i686/fpu/multiarch/libm-test-ulps b/sysdeps/i386/i686/fpu/multiarch/libm-test-ulps
index f416f8d1f0620647244a82205bb9b352cd50bce7..e5bc05eed13818093232141bbf8b40653a6c8beb 100644 (file)
--- a/sysdeps/i386/i686/fpu/multiarch/libm-test-ulps
+++ b/sysdeps/i386/i686/fpu/multiarch/libm-test-ulps
@@ -90,19 +90,16 @@ ldouble: 3
 
 Function: "asinh_downward":
 double: 1
-float: 1
 float128: 4
 ldouble: 5
 
 Function: "asinh_towardzero":
 double: 1
-float: 1
 float128: 2
 ldouble: 4
 
 Function: "asinh_upward":
 double: 1
-float: 1
 float128: 4
 ldouble: 5
 

diff --git a/sysdeps/ieee754/flt-32/s_asinhf.c b/sysdeps/ieee754/flt-32/s_asinhf.c
index a94676fbda91d9091a068cf987145be075ab6ebd..93750f5dae87399d047e261cb78c42bde2b04e8e 100644 (file)
--- a/sysdeps/ieee754/flt-32/s_asinhf.c
+++ b/sysdeps/ieee754/flt-32/s_asinhf.c
@@ -1,51 +1,190 @@
-/* s_asinhf.c -- float version of s_asinh.c.
- */
-
-/*
- * ====================================================
- * 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.
- * ====================================================
- */
-
-#include <float.h>
+/* Correctly-rounded inverse hyperbolic sine function for binary32 format.
+
+Copyright (c) 2023-2024 Alexei Sibidanov.
+
+The original version of this file was copied from the CORE-MATH
+project (file src/binary32/asinh/asinhf.c, revision e8774b3).
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.
+*/
+
 #include <math.h>
-#include <math_private.h>
-#include <math-underflow.h>
+#include <stdint.h>
 #include <libm-alias-float.h>
-
-static const float
-one =  1.0000000000e+00, /* 0x3F800000 */
-ln2 =  6.9314718246e-01, /* 0x3f317218 */
-huge=  1.0000000000e+30;
+#include "math_config.h"
 
 float
-__asinhf(float x)
+__asinhf (float x)
 {
-       float w;
-       int32_t hx,ix;
-       GET_FLOAT_WORD(hx,x);
-       ix = hx&0x7fffffff;
-       if(__builtin_expect(ix< 0x38000000, 0)) {       /* |x|<2**-14 */
-           math_check_force_underflow (x);
-           if(huge+x>one) return x;    /* return x inexact except 0 */
+  static const double ix[] =
+    {
+      0x1p+0,          0x1.fc07f01fcp-1, 0x1.f81f81f82p-1, 0x1.f44659e4ap-1,
+      0x1.f07c1f07cp-1, 0x1.ecc07b302p-1, 0x1.e9131abfp-1,  0x1.e573ac902p-1,
+      0x1.e1e1e1e1ep-1, 0x1.de5d6e3f8p-1, 0x1.dae6076bap-1, 0x1.d77b654b8p-1,
+      0x1.d41d41d42p-1, 0x1.d0cb58f6ep-1, 0x1.cd8568904p-1, 0x1.ca4b3055ep-1,
+      0x1.c71c71c72p-1, 0x1.c3f8f01c4p-1, 0x1.c0e070382p-1, 0x1.bdd2b8994p-1,
+      0x1.bacf914c2p-1, 0x1.b7d6c3ddap-1, 0x1.b4e81b4e8p-1, 0x1.b2036406cp-1,
+      0x1.af286bca2p-1, 0x1.ac5701ac6p-1, 0x1.a98ef606ap-1, 0x1.a6d01a6dp-1,
+      0x1.a41a41a42p-1, 0x1.a16d3f97ap-1, 0x1.9ec8e951p-1,  0x1.9c2d14ee4p-1,
+      0x1.99999999ap-1, 0x1.970e4f80cp-1, 0x1.948b0fcd6p-1, 0x1.920fb49dp-1,
+      0x1.8f9c18f9cp-1, 0x1.8d3018d3p-1,  0x1.8acb90f6cp-1, 0x1.886e5f0acp-1,
+      0x1.861861862p-1, 0x1.83c977ab2p-1, 0x1.818181818p-1, 0x1.7f405fd02p-1,
+      0x1.7d05f417ep-1, 0x1.7ad2208ep-1,  0x1.78a4c8178p-1, 0x1.767dce434p-1,
+      0x1.745d1745ep-1, 0x1.724287f46p-1, 0x1.702e05c0cp-1, 0x1.6e1f76b44p-1,
+      0x1.6c16c16c2p-1, 0x1.6a13cd154p-1, 0x1.681681682p-1, 0x1.661ec6a52p-1,
+      0x1.642c8590cp-1, 0x1.623fa7702p-1, 0x1.605816058p-1, 0x1.5e75bb8dp-1,
+      0x1.5c9882b94p-1, 0x1.5ac056b02p-1, 0x1.58ed23082p-1, 0x1.571ed3c5p-1,
+      0x1.555555556p-1, 0x1.5390948f4p-1, 0x1.51d07eae2p-1, 0x1.501501502p-1,
+      0x1.4e5e0a73p-1,  0x1.4cab88726p-1, 0x1.4afd6a052p-1, 0x1.49539e3b2p-1,
+      0x1.47ae147aep-1, 0x1.460cbc7f6p-1, 0x1.446f86562p-1, 0x1.42d6625d6p-1,
+      0x1.414141414p-1, 0x1.3fb013fbp-1,  0x1.3e22cbce4p-1, 0x1.3c995a47cp-1,
+      0x1.3b13b13b2p-1, 0x1.3991c2c18p-1, 0x1.381381382p-1, 0x1.3698df3dep-1,
+      0x1.3521cfb2cp-1, 0x1.33ae45b58p-1, 0x1.323e34a2cp-1, 0x1.30d19013p-1,
+      0x1.2f684bda2p-1, 0x1.2e025c04cp-1, 0x1.2c9fb4d82p-1, 0x1.2b404ad02p-1,
+      0x1.29e4129e4p-1, 0x1.288b01288p-1, 0x1.27350b882p-1, 0x1.25e22708p-1,
+      0x1.24924924ap-1, 0x1.23456789ap-1, 0x1.21fb78122p-1, 0x1.20b470c68p-1,
+      0x1.1f7047dc2p-1, 0x1.1e2ef3b4p-1,  0x1.1cf06ada2p-1, 0x1.1bb4a4046p-1,
+      0x1.1a7b9611ap-1, 0x1.19453808cp-1, 0x1.181181182p-1, 0x1.16e068942p-1,
+      0x1.15b1e5f76p-1, 0x1.1485f0e0ap-1, 0x1.135c81136p-1, 0x1.12358e75ep-1,
+      0x1.111111112p-1, 0x1.0fef010fep-1, 0x1.0ecf56be6p-1, 0x1.0db20a89p-1,
+      0x1.0c9714fbcp-1, 0x1.0b7e6ec26p-1, 0x1.0a6810a68p-1, 0x1.0953f3902p-1,
+      0x1.084210842p-1, 0x1.073260a48p-1, 0x1.0624dd2f2p-1, 0x1.05197f7d8p-1,
+      0x1.041041042p-1, 0x1.03091b52p-1,  0x1.020408102p-1, 0x1.01010101p-1,
+      0x1p-1
+    };
+
+  static const double lix[] =
+    {
+      0x0p+0,               0x1.fe02a6b146789p-8, 0x1.fc0a8b0fa03e4p-7,
+      0x1.7b91b07de311bp-6, 0x1.f829b0e7c33p-6,   0x1.39e87b9fd7d6p-5,
+      0x1.77458f63edcfcp-5, 0x1.b42dd7117b1bfp-5, 0x1.f0a30c01362a6p-5,
+      0x1.16536eea7fae1p-4, 0x1.341d7961791d1p-4, 0x1.51b073f07983fp-4,
+      0x1.6f0d28ae3eb4cp-4, 0x1.8c345d6383b21p-4, 0x1.a926d3a475563p-4,
+      0x1.c5e548f63a743p-4, 0x1.e27076e28f2e6p-4, 0x1.fec9131dbaabbp-4,
+      0x1.0d77e7ccf6e59p-3, 0x1.1b72ad52f87ap-3,  0x1.29552f81eb523p-3,
+      0x1.371fc201f7f74p-3, 0x1.44d2b6ccbfd1ep-3, 0x1.526e5e3a41438p-3,
+      0x1.5ff3070a613d4p-3, 0x1.6d60fe717221dp-3, 0x1.7ab890212b909p-3,
+      0x1.87fa065214911p-3, 0x1.9525a9cf296b4p-3, 0x1.a23bc1fe42563p-3,
+      0x1.af3c94e81bff3p-3, 0x1.bc2867430acd6p-3, 0x1.c8ff7c7989a22p-3,
+      0x1.d5c216b535b91p-3, 0x1.e27076e2f92e6p-3, 0x1.ef0adcbe0d936p-3,
+      0x1.fb9186d5ebe2bp-3, 0x1.0402594b51041p-2, 0x1.0a324e27370e3p-2,
+      0x1.1058bf9ad7ad5p-2, 0x1.1675cabaa660ep-2, 0x1.1c898c16b91fbp-2,
+      0x1.22941fbcfb966p-2, 0x1.2895a13dd2ea3p-2, 0x1.2e8e2bade7d31p-2,
+      0x1.347dd9a9afd55p-2, 0x1.3a64c556b05eap-2, 0x1.40430868877e4p-2,
+      0x1.4618bc219dec2p-2, 0x1.4be5f9579e0a1p-2, 0x1.51aad872c982dp-2,
+      0x1.5767717432a6cp-2, 0x1.5d1bdbf5669cap-2, 0x1.62c82f2b83795p-2,
+      0x1.686c81e9964afp-2, 0x1.6e08eaa2929e4p-2, 0x1.739d7f6b95007p-2,
+      0x1.792a55fdb7fa2p-2, 0x1.7eaf83b82efc3p-2, 0x1.842d1da1ecb17p-2,
+      0x1.89a3386be825bp-2, 0x1.8f11e87347ac7p-2, 0x1.947941c1f26fbp-2,
+      0x1.99d958119208bp-2, 0x1.9f323ecbd984cp-2, 0x1.a484090e5eb0ap-2,
+      0x1.a9cec9a9cf84ap-2, 0x1.af1293245606bp-2, 0x1.b44f77bc98f63p-2,
+      0x1.b9858969218fbp-2, 0x1.beb4d9da96b7cp-2, 0x1.c3dd7a7d0354dp-2,
+      0x1.c8ff7c79ada22p-2, 0x1.ce1af0b855bebp-2, 0x1.d32fe7e039bd5p-2,
+      0x1.d83e72587673ep-2, 0x1.dd46a04c204a1p-2, 0x1.e24881a7cac26p-2,
+      0x1.e744261d8a788p-2, 0x1.ec399d2457ccp-2,  0x1.f128f5fac86edp-2,
+      0x1.f6123fa71c8acp-2, 0x1.faf588f76631fp-2, 0x1.ffd2e08580c98p-2,
+      0x1.02552a5a4f0ffp-1, 0x1.04bdf9da8b6d2p-1, 0x1.0723e5c1b4f4p-1,
+      0x1.0986f4f589521p-1, 0x1.0be72e423ca83p-1, 0x1.0e44985d0f48cp-1,
+      0x1.109f39e2be497p-1, 0x1.12f71959283bcp-1, 0x1.154c3d2f4f5eap-1,
+      0x1.179eabbd9c9a1p-1, 0x1.19ee6b466516fp-1, 0x1.1c3b81f723c25p-1,
+      0x1.1e85f5e6ec0dp-1,  0x1.20cdcd193f76ep-1, 0x1.23130d7beb743p-1,
+      0x1.2555bce9887cbp-1, 0x1.2795e1288211bp-1, 0x1.29d37fec2308bp-1,
+      0x1.2c0e9ed45768cp-1, 0x1.2e47436e5ae68p-1, 0x1.307d7334ff0bep-1,
+      0x1.32b1339134571p-1, 0x1.34e289d9b39d3p-1, 0x1.37117b5481bb6p-1,
+      0x1.393e0d3549a1ap-1, 0x1.3b6844a017823p-1, 0x1.3d9026a70eefbp-1,
+      0x1.3fb5b84cfeb42p-1, 0x1.41d8fe844b2aep-1, 0x1.43f9fe2fb9267p-1,
+      0x1.4618bc21d86c2p-1, 0x1.48353d1e928dfp-1, 0x1.4a4f85db1debbp-1,
+      0x1.4c679afcc323ap-1, 0x1.4e7d811b77bb1p-1, 0x1.50913cbff8c6bp-1,
+      0x1.52a2d265be5abp-1, 0x1.54b2467998498p-1, 0x1.56bf9d5b34b99p-1,
+      0x1.58cadb5cbe989p-1, 0x1.5ad404c33af2dp-1, 0x1.5cdb1dc6ad765p-1,
+      0x1.5ee02a9241e75p-1, 0x1.60e32f447a8d9p-1, 0x1.62e42fefa39efp-1
+    };
+
+  uint32_t t = asuint (x);
+  t &= ~0u >> 1;
+  double xs = x;
+  if (__glibc_unlikely (t <= 0x3e815667u))
+    {
+      if (__glibc_unlikely (t <= 0x39ddb3d7u))
+       {
+         if (__glibc_unlikely (t == 0))
+           return x;
+         return fmaf (x, -0x1p-25f, x);
        }
-       if(__builtin_expect(ix>0x47000000, 0)) {        /* |x| > 2**14 */
-           if(ix>=0x7f800000) return x+x;      /* x is inf or NaN */
-           w = __ieee754_logf(fabsf(x))+ln2;
-       } else {
-           float xa = fabsf(x);
-           if (ix>0x40000000) {        /* 2**14 > |x| > 2.0 */
-               w = __ieee754_logf(2.0f*xa+one/(sqrtf(xa*xa+one)+xa));
-           } else {            /* 2.0 > |x| > 2**-14 */
-               float t = xa*xa;
-               w =__log1pf(xa+t/(one+sqrtf(one+t)));
+      static const double c[] =
+       {
+          0x1.5555555555553p-3, -0x1.3333333330e9dp-4,  0x1.6db6db67cb37ap-5,
+         -0x1.f1c71699375dp-6,   0x1.6e8a374c39ff9p-6, -0x1.1c1e98f9d01e1p-6,
+          0x1.c277e96d84026p-7, -0x1.329ff5faf02abp-7
+       };
+      double x2 = xs * xs;
+      double x4 = x2 * x2;
+      double  x8 = x4 * x4;
+      double f = x2 * (((c[0] + x2 * c[1]) + x4 * (c[2] + x2 * c[3]))
+                      + x8 * ((c[4] + x2 * c[5]) + x4 * (c[6] + x2 * c[7])));
+      double r = xs - xs * f;
+      return r;
+    }
+  else
+    {
+      if (__glibc_unlikely (t >= 0x7f800000u))
+       return x + x; /* +-inf or nan */
+      double xd = fabs (xs);
+      double x2 = xd * xd;
+      uint64_t tp = asuint64 (xd + sqrt (x2 + 1));
+      uint64_t m = tp & (~UINT64_C(0) >> 12);
+      int j = (m + (UINT64_C(1) << (52 - 8))) >> (52 - 7);
+      int e = (tp >> 52) - 0x3ff;
+      double w = asdouble (m | UINT64_C(0x3ff) << 52);
+      double z = w * ix[j] - 1.0;
+      static const double c[] =
+       {
+         0x1.0000000066947p+0, -0x1.00007f053d8cbp-1, 0x1.555280111d914p-2
+       };
+      double z2 = z * z;
+      double r = ((lix[128] * e + lix[j]) + z * c[0]) + z2 * (c[1] + z * c[2]);
+      if (__glibc_unlikely (((asuint64 (r) + 259000) & INT64_C(0xfffffff))
+                           < 260000)) /* accurate path */
+       {
+         static const double cp[] =
+           {
+              0x1p+0,               -0x1p-1,
+              0x1.55555555030bcp-2, -0x1.ffffffff2b4e5p-3,
+              0x1.999b5076a42f2p-3, -0x1.55570c45a647dp-3
+           };
+         z2 = z * z;
+         double c0 = cp[0] + z * cp[1];
+         double c2 = cp[2] + z * cp[3];
+         double c4 = cp[4] + z * cp[5];
+         c0 += z2 * (c2 + z2 * c4);
+         const double ln2l = 0x1.7f7d1cf79abcap-20;
+         const double ln2h = 0x1.62e4p-1;
+         double Lh = ln2h * e;
+         double Ll = ln2l * e;
+         r = fma (z, c0, Ll + lix[j]) + Lh;
+         if (__glibc_unlikely ((asuint64 (r) & INT64_C(0xfffffff)) == 0))
+           {
+             double h = fma (z, c0, Ll + lix[j]) + (Lh - r);
+             r = r + 64 * h;
            }
        }
-       return copysignf(w, x);
+      return copysign (r, xs);
+    }
 }
 libm_alias_float (__asinh, asinh)

diff --git a/sysdeps/loongarch/lp64/libm-test-ulps b/sysdeps/loongarch/lp64/libm-test-ulps
index 67beca8ff6ca63bafa262f43b2f84da67d9cdcb2..b24bc582ea4e5a60f81485bc3a241400229a2617 100644 (file)
--- a/sysdeps/loongarch/lp64/libm-test-ulps
+++ b/sysdeps/loongarch/lp64/libm-test-ulps
@@ -51,22 +51,18 @@ ldouble: 2
 
 Function: "asinh":
 double: 2
-float: 2
 ldouble: 4
 
 Function: "asinh_downward":
 double: 3
-float: 3
 ldouble: 4
 
 Function: "asinh_towardzero":
 double: 2
-float: 2
 ldouble: 2
 
 Function: "asinh_upward":
 double: 3
-float: 3
 ldouble: 4
 
 Function: "atan":

diff --git a/sysdeps/m68k/m680x0/fpu/libm-test-ulps b/sysdeps/m68k/m680x0/fpu/libm-test-ulps
index 73019999e0adf6675ad74442f301ab036abed083..7b3e67efaf57543dc204c5f5654d4b20511fb37f 100644 (file)
--- a/sysdeps/m68k/m680x0/fpu/libm-test-ulps
+++ b/sysdeps/m68k/m680x0/fpu/libm-test-ulps
@@ -34,22 +34,18 @@ float: 1
 
 Function: "asinh":
 double: 1
-float: 1
 ldouble: 1
 
 Function: "asinh_downward":
 double: 2
-float: 2
 ldouble: 2
 
 Function: "asinh_towardzero":
 double: 2
-float: 2
 ldouble: 2
 
 Function: "asinh_upward":
 double: 2
-float: 2
 ldouble: 3
 
 Function: "atan2":

diff --git a/sysdeps/microblaze/libm-test-ulps b/sysdeps/microblaze/libm-test-ulps
index a0e8b5bd507c1df1b01f39b6f0ef49592ff5225a..b7e73db063ca57bf9964239c763777ae43c79224 100644 (file)
--- a/sysdeps/microblaze/libm-test-ulps
+++ b/sysdeps/microblaze/libm-test-ulps
@@ -10,7 +10,6 @@ Function: "asin":
 
 Function: "asinh":
 double: 1
-float: 1
 
 Function: "atan":
 float: 1

diff --git a/sysdeps/mips/mips32/libm-test-ulps b/sysdeps/mips/mips32/libm-test-ulps
index 7eca3dc8dfd550016a4a9165fa2f22c95ee83cce..ca4eac5090f73d3d536a1783adea8dee63e400e7 100644 (file)
--- a/sysdeps/mips/mips32/libm-test-ulps
+++ b/sysdeps/mips/mips32/libm-test-ulps
@@ -39,19 +39,15 @@ double: 2
 
 Function: "asinh":
 double: 2
-float: 2
 
 Function: "asinh_downward":
 double: 3
-float: 3
 
 Function: "asinh_towardzero":
 double: 2
-float: 2
 
 Function: "asinh_upward":
 double: 3
-float: 3
 
 Function: "atan":
 double: 1

diff --git a/sysdeps/mips/mips64/libm-test-ulps b/sysdeps/mips/mips64/libm-test-ulps
index 4cabe88af10806db4fccce014499160ea9d39ccf..30e8d46c68b272f22f3a8af6b9cec894f75bba6a 100644 (file)
--- a/sysdeps/mips/mips64/libm-test-ulps
+++ b/sysdeps/mips/mips64/libm-test-ulps
@@ -51,22 +51,18 @@ ldouble: 2
 
 Function: "asinh":
 double: 2
-float: 2
 ldouble: 4
 
 Function: "asinh_downward":
 double: 3
-float: 3
 ldouble: 4
 
 Function: "asinh_towardzero":
 double: 2
-float: 2
 ldouble: 2
 
 Function: "asinh_upward":
 double: 3
-float: 3
 ldouble: 4
 
 Function: "atan":

diff --git a/sysdeps/or1k/fpu/libm-test-ulps b/sysdeps/or1k/fpu/libm-test-ulps
index f95e13886479c427dd53dbc09aed8088c7fd0fb4..dd972b3063c998dd7c83887005a0e223be7fd11f 100644 (file)
--- a/sysdeps/or1k/fpu/libm-test-ulps
+++ b/sysdeps/or1k/fpu/libm-test-ulps
@@ -39,19 +39,15 @@ double: 2
 
 Function: "asinh":
 double: 2
-float: 2
 
 Function: "asinh_downward":
 double: 3
-float: 3
 
 Function: "asinh_towardzero":
 double: 2
-float: 2
 
 Function: "asinh_upward":
 double: 3
-float: 3
 
 Function: "atan":
 double: 1

diff --git a/sysdeps/or1k/nofpu/libm-test-ulps b/sysdeps/or1k/nofpu/libm-test-ulps
index 5ec5b1f3f3ad1f3294df19a19585f7abca0b86b5..d5d1c8dc8097764c7e5b68b7e2b10bead381ed66 100644 (file)
--- a/sysdeps/or1k/nofpu/libm-test-ulps
+++ b/sysdeps/or1k/nofpu/libm-test-ulps
@@ -43,19 +43,15 @@ double: 1
 
 Function: "asinh":
 double: 2
-float: 2
 
 Function: "asinh_downward":
 double: 3
-float: 3
 
 Function: "asinh_towardzero":
 double: 2
-float: 2
 
 Function: "asinh_upward":
 double: 3
-float: 3
 
 Function: "asinpi":
 double: 1

diff --git a/sysdeps/powerpc/fpu/libm-test-ulps b/sysdeps/powerpc/fpu/libm-test-ulps
index a11a5894e0e1d91431fe3ffd3868f7985d38f7ee..edb6678fdb53f3b18c359ba35133ec931a2fd035 100644 (file)
--- a/sysdeps/powerpc/fpu/libm-test-ulps
+++ b/sysdeps/powerpc/fpu/libm-test-ulps
@@ -91,25 +91,21 @@ ldouble: 2
 
 Function: "asinh":
 double: 2
-float: 2
 float128: 4
 ldouble: 2
 
 Function: "asinh_downward":
 double: 3
-float: 3
 float128: 4
 ldouble: 5
 
 Function: "asinh_towardzero":
 double: 2
-float: 2
 float128: 2
 ldouble: 5
 
 Function: "asinh_upward":
 double: 3
-float: 3
 float128: 4
 ldouble: 7
 

diff --git a/sysdeps/powerpc/nofpu/libm-test-ulps b/sysdeps/powerpc/nofpu/libm-test-ulps
index d03d25e54a420f303681056c89953bb9f19b6084..939468399cd0d91a0c39458dc5e8e079081096d1 100644 (file)
--- a/sysdeps/powerpc/nofpu/libm-test-ulps
+++ b/sysdeps/powerpc/nofpu/libm-test-ulps
@@ -55,22 +55,18 @@ ldouble: 2
 
 Function: "asinh":
 double: 2
-float: 2
 ldouble: 2
 
 Function: "asinh_downward":
 double: 3
-float: 3
 ldouble: 5
 
 Function: "asinh_towardzero":
 double: 2
-float: 2
 ldouble: 5
 
 Function: "asinh_upward":
 double: 3
-float: 3
 ldouble: 7
 
 Function: "atan":

diff --git a/sysdeps/riscv/nofpu/libm-test-ulps b/sysdeps/riscv/nofpu/libm-test-ulps
index edaa91a4c198165584b73b4750e0e3e0993c0ff3..7c89f7915ba0e75539dfa91672381be7e4b91aec 100644 (file)
--- a/sysdeps/riscv/nofpu/libm-test-ulps
+++ b/sysdeps/riscv/nofpu/libm-test-ulps
@@ -51,22 +51,18 @@ ldouble: 2
 
 Function: "asinh":
 double: 2
-float: 2
 ldouble: 4
 
 Function: "asinh_downward":
 double: 3
-float: 3
 ldouble: 4
 
 Function: "asinh_towardzero":
 double: 2
-float: 2
 ldouble: 2
 
 Function: "asinh_upward":
 double: 3
-float: 3
 ldouble: 4
 
 Function: "atan":

diff --git a/sysdeps/riscv/rvd/libm-test-ulps b/sysdeps/riscv/rvd/libm-test-ulps
index 4db2d0231b4f11041af8c55ad80d7810ec7bd479..e1c1a6aee5453ded6dacbd406da943bea2bfe027 100644 (file)
--- a/sysdeps/riscv/rvd/libm-test-ulps
+++ b/sysdeps/riscv/rvd/libm-test-ulps
@@ -51,22 +51,18 @@ ldouble: 2
 
 Function: "asinh":
 double: 2
-float: 2
 ldouble: 4
 
 Function: "asinh_downward":
 double: 3
-float: 3
 ldouble: 4
 
 Function: "asinh_towardzero":
 double: 2
-float: 2
 ldouble: 2
 
 Function: "asinh_upward":
 double: 3
-float: 3
 ldouble: 4
 
 Function: "atan":

diff --git a/sysdeps/s390/fpu/libm-test-ulps b/sysdeps/s390/fpu/libm-test-ulps
index 22dfe602940d1c9bfd877440e3e35e33905b753f..038e3eb9d72bc8e7a798d690d39e0d5cc1648dd7 100644 (file)
--- a/sysdeps/s390/fpu/libm-test-ulps
+++ b/sysdeps/s390/fpu/libm-test-ulps
@@ -51,22 +51,18 @@ ldouble: 2
 
 Function: "asinh":
 double: 2
-float: 2
 ldouble: 4
 
 Function: "asinh_downward":
 double: 3
-float: 3
 ldouble: 4
 
 Function: "asinh_towardzero":
 double: 2
-float: 2
 ldouble: 2
 
 Function: "asinh_upward":
 double: 3
-float: 3
 ldouble: 4
 
 Function: "atan":

diff --git a/sysdeps/sh/libm-test-ulps b/sysdeps/sh/libm-test-ulps
index e225f7893604c37bd88a0d32822c47cba0620f9e..4bd1ff1f988daa868848ada106673a2e9e635731 100644 (file)
--- a/sysdeps/sh/libm-test-ulps
+++ b/sysdeps/sh/libm-test-ulps
@@ -19,11 +19,9 @@ double: 1
 
 Function: "asinh":
 double: 2
-float: 2
 
 Function: "asinh_towardzero":
 double: 2
-float: 2
 
 Function: "atan":
 float: 1

diff --git a/sysdeps/sparc/fpu/libm-test-ulps b/sysdeps/sparc/fpu/libm-test-ulps
index f24b64a944c5dfb2986304d877f7d2a17346da86..a4d678da03b466818f649a0ee59b424307b655fb 100644 (file)
--- a/sysdeps/sparc/fpu/libm-test-ulps
+++ b/sysdeps/sparc/fpu/libm-test-ulps
@@ -51,22 +51,18 @@ ldouble: 2
 
 Function: "asinh":
 double: 2
-float: 2
 ldouble: 4
 
 Function: "asinh_downward":
 double: 3
-float: 3
 ldouble: 4
 
 Function: "asinh_towardzero":
 double: 2
-float: 2
 ldouble: 2
 
 Function: "asinh_upward":
 double: 3
-float: 3
 ldouble: 4
 
 Function: "atan":

diff --git a/sysdeps/x86_64/fpu/libm-test-ulps b/sysdeps/x86_64/fpu/libm-test-ulps
index ec2b3629accddb17f424e0afb60cff8f763a3cde..23c40d059080fce995ccdee2720dc20107f382d4 100644 (file)
--- a/sysdeps/x86_64/fpu/libm-test-ulps
+++ b/sysdeps/x86_64/fpu/libm-test-ulps
@@ -147,25 +147,21 @@ float: 1
 
 Function: "asinh":
 double: 2
-float: 2
 float128: 4
 ldouble: 3
 
 Function: "asinh_downward":
 double: 3
-float: 3
 float128: 4
 ldouble: 5
 
 Function: "asinh_towardzero":
 double: 2
-float: 2
 float128: 2
 ldouble: 4
 
 Function: "asinh_upward":
 double: 3
-float: 3
 float128: 4
 ldouble: 5