[15/17] math: Use log10p1f from CORE-MATH

Message ID	20241025182614.2022697-16-adhemerval.zanella@linaro.org
State	New
Headers	show Return-Path: <libc-alpha-bounces~incoming=patchwork.ozlabs.org@sourceware.org> DMARC-Filter: OpenDMARC Filter v1.4.2 sourceware.org 5490A385840F From: Adhemerval Zanella <adhemerval.zanella@linaro.org> To: libc-alpha@sourceware.org Cc: Paul Zimmermann <Paul.Zimmermann@inria.fr>, Alexei Sibidanov <sibid@uvic.ca> Subject: [PATCH 15/17] math: Use log10p1f from CORE-MATH Date: Fri, 25 Oct 2024 15:21:53 -0300 Message-ID: <20241025182614.2022697-16-adhemerval.zanella@linaro.org> In-Reply-To: <20241025182614.2022697-1-adhemerval.zanella@linaro.org> References: <20241025182614.2022697-1-adhemerval.zanella@linaro.org> MIME-Version: 1.0 Content-Transfer-Encoding: 8bit Precedence: list Errors-To: libc-alpha-bounces~incoming=patchwork.ozlabs.org@sourceware.org
Series	Add more CORE-MATH on libm \| expand [00/17] Add more CORE-MATH on libm [01/17] math: Add e_gammaf_r to glibc code and style [02/17] benchtests: Add exp10m1f benchmark [03/17] benchtests: Add exp2m1f benchmark [04/17] benchtests: Add expm1f benchmark [05/17] benchtests: Add log10f benchmark [06/17] benchtests: Add log2p1f benchmark [07/17] benchtests: Add log1p benchmark [08/17] benchtests: Add log10p1f benchmark [09/17] math: Use exp10m1f from CORE-MATH [10/17] math: Use exp2m1f from CORE-MATH [11/17] math: Use expm1f from CORE-MATH [13/17] math: Use log2p1f from CORE-MATH [15/17] math: Use log10p1f from CORE-MATH [16/17] x86_64: Add exp10m1f with FMA [17/17] x86_64: Add exp2m1f with FMA

diff --git a/sysdeps/aarch64/libm-test-ulps b/sysdeps/aarch64/libm-test-ulps index 5700c712ff..c523d45802 100644 --- a/sysdeps/aarch64/libm-test-ulps +++ b/sysdeps/aarch64/libm-test-ulps @@ -1346,22 +1346,18 @@ ldouble: 1 Function: "log10p1": double: 2 -float: 2 ldouble: 3 Function: "log10p1_downward": double: 2 -float: 3 ldouble: 4 Function: "log10p1_towardzero": double: 3 -float: 2 ldouble: 3 Function: "log10p1_upward": double: 2 -float: 3 ldouble: 4 Function: "log1p": diff --git a/sysdeps/arc/fpu/libm-test-ulps b/sysdeps/arc/fpu/libm-test-ulps index 54dc4d6760..c12e241c48 100644 --- a/sysdeps/arc/fpu/libm-test-ulps +++ b/sysdeps/arc/fpu/libm-test-ulps @@ -968,19 +968,15 @@ float: 4 Function: "log10p1": double: 2 -float: 2 Function: "log10p1_downward": double: 2 -float: 3 Function: "log10p1_towardzero": double: 3 -float: 2 Function: "log10p1_upward": double: 2 -float: 3 Function: "log1p": double: 1 diff --git a/sysdeps/arc/nofpu/libm-test-ulps b/sysdeps/arc/nofpu/libm-test-ulps index f6d1ef062c..d0cfa46c3d 100644 --- a/sysdeps/arc/nofpu/libm-test-ulps +++ b/sysdeps/arc/nofpu/libm-test-ulps @@ -234,7 +234,6 @@ float: 2 Function: "log10p1": double: 1 -float: 2 Function: "log1p": double: 1 diff --git a/sysdeps/arm/libm-test-ulps b/sysdeps/arm/libm-test-ulps index 502680ff1b..6cdd3d53d6 100644 --- a/sysdeps/arm/libm-test-ulps +++ b/sysdeps/arm/libm-test-ulps @@ -958,19 +958,15 @@ float: 2 Function: "log10p1": double: 1 -float: 2 Function: "log10p1_downward": double: 2 -float: 3 Function: "log10p1_towardzero": double: 3 -float: 2 Function: "log10p1_upward": double: 2 -float: 3 Function: "log1p": double: 1 diff --git a/sysdeps/hppa/fpu/libm-test-ulps b/sysdeps/hppa/fpu/libm-test-ulps index b56f512b39..021a2a482c 100644 --- a/sysdeps/hppa/fpu/libm-test-ulps +++ b/sysdeps/hppa/fpu/libm-test-ulps @@ -981,19 +981,15 @@ float: 2 Function: "log10p1": double: 2 -float: 2 Function: "log10p1_downward": double: 2 -float: 3 Function: "log10p1_towardzero": double: 3 -float: 2 Function: "log10p1_upward": double: 2 -float: 3 Function: "log1p": double: 1 diff --git a/sysdeps/i386/fpu/libm-test-ulps b/sysdeps/i386/fpu/libm-test-ulps index 63ef3569b4..a77ded2648 100644 --- a/sysdeps/i386/fpu/libm-test-ulps +++ b/sysdeps/i386/fpu/libm-test-ulps @@ -1414,25 +1414,21 @@ ldouble: 1 Function: "log10p1": double: 1 -float: 2 float128: 3 ldouble: 4 Function: "log10p1_downward": double: 2 -float: 3 float128: 4 ldouble: 8 Function: "log10p1_towardzero": double: 2 -float: 1 float128: 3 ldouble: 8 Function: "log10p1_upward": double: 2 -float: 3 float128: 4 ldouble: 6 diff --git a/sysdeps/i386/i686/fpu/multiarch/libm-test-ulps b/sysdeps/i386/i686/fpu/multiarch/libm-test-ulps index f6393dbdf1..a9cd01bf03 100644 --- a/sysdeps/i386/i686/fpu/multiarch/libm-test-ulps +++ b/sysdeps/i386/i686/fpu/multiarch/libm-test-ulps @@ -1417,25 +1417,21 @@ ldouble: 1 Function: "log10p1": double: 1 -float: 2 float128: 3 ldouble: 4 Function: "log10p1_downward": double: 2 -float: 1 float128: 4 ldouble: 8 Function: "log10p1_towardzero": double: 2 -float: 1 float128: 3 ldouble: 8 Function: "log10p1_upward": double: 2 -float: 2 float128: 4 ldouble: 6 diff --git a/sysdeps/ieee754/flt-32/s_log10p1f.c b/sysdeps/ieee754/flt-32/s_log10p1f.c new file mode 100644 index 0000000000..29dad2ea94 --- /dev/null +++ b/sysdeps/ieee754/flt-32/s_log10p1f.c @@ -0,0 +1,182 @@ +/* Correctly-rounded biased argument base-10 logarithm function for binary32 value. + +Copyright (c) 2022-2023 Alexei Sibidanov. + +This file is part of the CORE-MATH project +project (file src/binary32/log10p1/log10p1f.c revision bc385c2). + +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 <stdint.h> +#include <errno.h> +#include <fenv.h> +#include <libm-alias-float.h> +#include "math_config.h" + +static __attribute__ ((noinline)) float +as_special (float x) +{ + uint32_t ux = asuint (x); + if (ux == 0x7f800000u) + return x; /* +inf */ + uint32_t ax = ux << 1; + if (ax == 0x17fu << 24) + /* x+1 = 0.0 */ + return __math_divzerof (1); + if (ax > 0xff000000u) + return x + x; /* nan */ + return __math_invalidf (x); +} + +float +__log10p1f (float x) +{ + static const double tr[] = + { + 0x1p+0, 0x1.f81f82p-1, 0x1.f07c1fp-1, 0x1.e9131acp-1, + 0x1.e1e1e1ep-1, 0x1.dae6077p-1, 0x1.d41d41dp-1, 0x1.cd85689p-1, + 0x1.c71c71cp-1, 0x1.c0e0704p-1, 0x1.bacf915p-1, 0x1.b4e81b5p-1, + 0x1.af286bdp-1, 0x1.a98ef6p-1, 0x1.a41a41ap-1, 0x1.9ec8e95p-1, + 0x1.999999ap-1, 0x1.948b0fdp-1, 0x1.8f9c19p-1, 0x1.8acb90fp-1, + 0x1.8618618p-1, 0x1.8181818p-1, 0x1.7d05f41p-1, 0x1.78a4c81p-1, + 0x1.745d174p-1, 0x1.702e05cp-1, 0x1.6c16c17p-1, 0x1.6816817p-1, + 0x1.642c859p-1, 0x1.605816p-1, 0x1.5c9882cp-1, 0x1.58ed231p-1, + 0x1.5555555p-1, 0x1.51d07ebp-1, 0x1.4e5e0a7p-1, 0x1.4afd6ap-1, + 0x1.47ae148p-1, 0x1.446f865p-1, 0x1.4141414p-1, 0x1.3e22cbdp-1, + 0x1.3b13b14p-1, 0x1.3813814p-1, 0x1.3521cfbp-1, 0x1.323e34ap-1, + 0x1.2f684bep-1, 0x1.2c9fb4ep-1, 0x1.29e412ap-1, 0x1.27350b9p-1, + 0x1.2492492p-1, 0x1.21fb781p-1, 0x1.1f7047ep-1, 0x1.1cf06aep-1, + 0x1.1a7b961p-1, 0x1.1811812p-1, 0x1.15b1e5fp-1, 0x1.135c811p-1, + 0x1.1111111p-1, 0x1.0ecf56cp-1, 0x1.0c9715p-1, 0x1.0a6810ap-1, + 0x1.0842108p-1, 0x1.0624dd3p-1, 0x1.041041p-1, 0x1.0204081p-1, + 0.5 + }; + static const double tl[] = + { + 0x1.562ec497ef351p-43, 0x1.b9476892ea99cp-8, 0x1.b5e909c959eecp-7, + 0x1.45f4f59ec84fp-6, 0x1.af5f92cbcf2aap-6, 0x1.0ba01a6069052p-5, + 0x1.3ed119b99dd41p-5, 0x1.714834298a088p-5, 0x1.a30a9d98309c1p-5, + 0x1.d41d51266b9d9p-5, 0x1.02428c0f62dfcp-4, 0x1.1a23444eea521p-4, + 0x1.31b30543f2597p-4, 0x1.48f3ed39bd5e7p-4, 0x1.5fe8049a0bd06p-4, + 0x1.769140a6a78eap-4, 0x1.8cf1836c96595p-4, 0x1.a30a9d5551a84p-4, + 0x1.b8de4d1ee5b21p-4, 0x1.ce6e4202c7bc9p-4, 0x1.e3bc1accaa6eap-4, + 0x1.f8c9683b584b7p-4, 0x1.06cbd68ca86ep-3, 0x1.11142f19de3a2p-3, + 0x1.1b3e71fa795fp-3, 0x1.254b4d37a3354p-3, 0x1.2f3b6912cab79p-3, + 0x1.390f6831144f7p-3, 0x1.42c7e7fffb21ap-3, 0x1.4c65808c779aep-3, + 0x1.55e8c507508c7p-3, 0x1.5f52445deb049p-3, 0x1.68a288c3efe72p-3, + 0x1.71da17bdef98bp-3, 0x1.7af9736089c4bp-3, 0x1.84011952a11ebp-3, + 0x1.8cf1837a7d6d1p-3, 0x1.95cb2891e3048p-3, 0x1.9e8e7b0f85651p-3, + 0x1.a73beaa5d9dfep-3, 0x1.afd3e39454544p-3, 0x1.b856cf060c662p-3, + 0x1.c0c5134de0c6dp-3, 0x1.c91f1371bb611p-3, 0x1.d1652ffcd2bc5p-3, + 0x1.d997c6f634ae6p-3, 0x1.e1b733ab8fbadp-3, 0x1.e9c3ceadab4c8p-3, + 0x1.f1bdeec438f77p-3, 0x1.f9a5e7a5f906fp-3, 0x1.00be05ac02564p-2, + 0x1.04a054d81990cp-2, 0x1.087a083594e33p-2, 0x1.0c4b457098b4fp-2, + 0x1.101431aa1f48ap-2, 0x1.13d4f08b98411p-2, 0x1.178da53edaecbp-2, + 0x1.1b3e71e9f9391p-2, 0x1.1ee777defd526p-2, 0x1.2288d7b48d874p-2, + 0x1.2622b0f52dad8p-2, 0x1.29b522a4c594cp-2, 0x1.2d404b0e305b9p-2, + 0x1.30c4478f3f21dp-2, 0x1.34413509f6f4dp-2 + }; + static const union + { + float f; + uint32_t u; + } st[] = + { + { 0x0p+0 }, { 0x1.2p+3 }, { 0x1.8cp+6 }, + { 0x1.f38p+9 }, { 0x1.3878p+13 }, { 0x1.869fp+16 }, + { 0x1.e847ep+19 }, { 0x1.312cfep+23 } + }; + double z = x; + uint32_t ux = asuint (x); + if (__glibc_unlikely (ux >= 0x17fu << 23)) /* x <= -1 */ + return as_special (x); + uint32_t ax = ux & (~0u >> 1); + if (__glibc_unlikely (ax == 0)) + return copysign (0, x); + if (__glibc_unlikely (ax >= (0xff << 23))) /* +inf, nan */ + return as_special (x); + int ie = ux; + ie >>= 23; + unsigned int je = ie - 126; + je = (je * 0x9a209a8) >> 29; + if (__glibc_unlikely (ux == st[je].u)) + return je; + + uint64_t tz = asuint64 (z + 1.0); + uint64_t m = tz & (~(uint64_t) 0 >> 12); + int32_t e = (tz >> 52) - 1023, j = ((m + ((int64_t) 1 << 45)) >> 46); + tz = m | ((uint64_t) 0x3ff << 52); + double ix = tr[j], l = tl[j]; + double off = e * 0x1.34413509f79ffp-2 + l; + double v = asdouble (tz) * ix - 1; + + static const double h[] = + { + 0x1.bcb7b150bf6d8p-2, -0x1.bcb7b1738c07ep-3, + 0x1.287de19e795c5p-3, -0x1.bca44edc44bc4p-4 + }; + double v2 = v * v; + double f = (h[0] + v * h[1]) + v2 * (h[2] + v * h[3]); + double r = off + v * f; + float ub = r; + float lb = r + 0x1.5cp-42; + if (__glibc_unlikely (ub != lb)) + { + if (__glibc_unlikely (ax < 0x3d32743eu)) + { /* 0x1.64e87cp-5f */ + if (__glibc_unlikely (ux == 0xa6aba8afu)) + return -0x1.2a33bcp-51f + 0x1p-76f; + if (__glibc_unlikely (ux == 0xaf39b9a7u)) + return -0x1.42a342p-34f + 0x1p-59f; + if (__glibc_unlikely (ux == 0x399a7c00u)) + return 0x1.0c53cap-13f + 0x1p-38f; + z /= 2.0 + z; + double z2 = z * z, z4 = z2 * z2; + static const double c[] = + { + 0x1.bcb7b1526e50fp-1, 0x1.287a76370129dp-2, + 0x1.63c62378fa3dbp-3, 0x1.fca4139a42374p-4 + }; + float ret = z * ((c[0] + z2 * c[1]) + z4 * (c[2] + z2 * c[3])); + if (x != 0.0f && ret == 0.0) + __set_errno (ERANGE); + return ret; + } + if (__glibc_unlikely (ux == 0x7956ba5eu)) + return 0x1.16bebap+5f + 0x1p-20f; + if (__glibc_unlikely (ux == 0xbd86ffb9u)) + return -0x1.e53536p-6f + 0x1p-31f; + static const double c[] = + { + 0x1.bcb7b1526e50ep-2, -0x1.bcb7b1526e53dp-3, 0x1.287a7636f3fa2p-3, + -0x1.bcb7b146a14b3p-4, 0x1.63c627d5219cbp-4, -0x1.2880736c8762dp-4, + 0x1.fc1ecf913961ap-5 + }; + f = v + * ((c[0] + v * c[1]) + + v2 * ((c[2] + v * c[3]) + v2 * (c[4] + v * c[5] + v2 * c[6]))); + f += l - tl[0]; + double el = e * 0x1.34413509f79ffp-2; + r = el + f; + ub = r; + } + return ub; +} +libm_alias_float (__log10p1, log10p1) diff --git a/sysdeps/loongarch/lp64/libm-test-ulps b/sysdeps/loongarch/lp64/libm-test-ulps index 846ddb4b3e..ecd9cc5873 100644 --- a/sysdeps/loongarch/lp64/libm-test-ulps +++ b/sysdeps/loongarch/lp64/libm-test-ulps @@ -1195,22 +1195,18 @@ ldouble: 1 Function: "log10p1": double: 2 -float: 2 ldouble: 3 Function: "log10p1_downward": double: 2 -float: 3 ldouble: 4 Function: "log10p1_towardzero": double: 3 -float: 2 ldouble: 3 Function: "log10p1_upward": double: 2 -float: 3 ldouble: 4 Function: "log1p": diff --git a/sysdeps/mips/mips32/libm-test-ulps b/sysdeps/mips/mips32/libm-test-ulps index 2ba46a0749..c319e0642c 100644 --- a/sysdeps/mips/mips32/libm-test-ulps +++ b/sysdeps/mips/mips32/libm-test-ulps @@ -955,19 +955,15 @@ float: 2 Function: "log10p1": double: 1 -float: 2 Function: "log10p1_downward": double: 2 -float: 3 Function: "log10p1_towardzero": double: 3 -float: 2 Function: "log10p1_upward": double: 2 -float: 3 Function: "log1p": double: 1 diff --git a/sysdeps/mips/mips64/libm-test-ulps b/sysdeps/mips/mips64/libm-test-ulps index f73f95cdad..365b860c54 100644 --- a/sysdeps/mips/mips64/libm-test-ulps +++ b/sysdeps/mips/mips64/libm-test-ulps @@ -1200,22 +1200,18 @@ ldouble: 1 Function: "log10p1": double: 1 -float: 2 ldouble: 3 Function: "log10p1_downward": double: 2 -float: 3 ldouble: 4 Function: "log10p1_towardzero": double: 3 -float: 2 ldouble: 3 Function: "log10p1_upward": double: 2 -float: 3 ldouble: 4 Function: "log1p": diff --git a/sysdeps/powerpc/fpu/libm-test-ulps b/sysdeps/powerpc/fpu/libm-test-ulps index 48a3ea10db..8d0c18eed1 100644 --- a/sysdeps/powerpc/fpu/libm-test-ulps +++ b/sysdeps/powerpc/fpu/libm-test-ulps @@ -1497,25 +1497,21 @@ ldouble: 2 Function: "log10p1": double: 2 -float: 2 float128: 3 ldouble: 3 Function: "log10p1_downward": double: 2 -float: 3 float128: 4 ldouble: 6 Function: "log10p1_towardzero": double: 3 -float: 2 float128: 3 ldouble: 6 Function: "log10p1_upward": double: 2 -float: 3 float128: 4 ldouble: 6 diff --git a/sysdeps/powerpc/nofpu/libm-test-ulps b/sysdeps/powerpc/nofpu/libm-test-ulps index ec9104806a..20036c779c 100644 --- a/sysdeps/powerpc/nofpu/libm-test-ulps +++ b/sysdeps/powerpc/nofpu/libm-test-ulps @@ -1257,22 +1257,18 @@ ldouble: 2 Function: "log10p1": double: 1 -float: 2 ldouble: 3 Function: "log10p1_downward": double: 2 -float: 3 ldouble: 6 Function: "log10p1_towardzero": double: 3 -float: 2 ldouble: 6 Function: "log10p1_upward": double: 2 -float: 3 ldouble: 6 Function: "log1p": diff --git a/sysdeps/riscv/nofpu/libm-test-ulps b/sysdeps/riscv/nofpu/libm-test-ulps index 0189fc5bd0..cccc864a7a 100644 --- a/sysdeps/riscv/nofpu/libm-test-ulps +++ b/sysdeps/riscv/nofpu/libm-test-ulps @@ -1167,7 +1167,6 @@ ldouble: 1 Function: "log10p1": double: 1 -float: 1 ldouble: 3 Function: "log1p": diff --git a/sysdeps/riscv/rvd/libm-test-ulps b/sysdeps/riscv/rvd/libm-test-ulps index a0b43761a2..14fc7633af 100644 --- a/sysdeps/riscv/rvd/libm-test-ulps +++ b/sysdeps/riscv/rvd/libm-test-ulps @@ -1195,22 +1195,18 @@ ldouble: 1 Function: "log10p1": double: 2 -float: 2 ldouble: 3 Function: "log10p1_downward": double: 2 -float: 3 ldouble: 4 Function: "log10p1_towardzero": double: 3 -float: 2 ldouble: 3 Function: "log10p1_upward": double: 2 -float: 3 ldouble: 4 Function: "log1p": diff --git a/sysdeps/s390/fpu/libm-test-ulps b/sysdeps/s390/fpu/libm-test-ulps index 658752c41e..288054aafc 100644 --- a/sysdeps/s390/fpu/libm-test-ulps +++ b/sysdeps/s390/fpu/libm-test-ulps @@ -1196,22 +1196,18 @@ ldouble: 1 Function: "log10p1": double: 2 -float: 2 ldouble: 3 Function: "log10p1_downward": double: 2 -float: 3 ldouble: 4 Function: "log10p1_towardzero": double: 3 -float: 2 ldouble: 3 Function: "log10p1_upward": double: 2 -float: 3 ldouble: 4 Function: "log1p": diff --git a/sysdeps/sparc/fpu/libm-test-ulps b/sysdeps/sparc/fpu/libm-test-ulps index 4c4d8ff238..6ea02058e9 100644 --- a/sysdeps/sparc/fpu/libm-test-ulps +++ b/sysdeps/sparc/fpu/libm-test-ulps @@ -1200,22 +1200,18 @@ ldouble: 1 Function: "log10p1": double: 1 -float: 2 ldouble: 3 Function: "log10p1_downward": double: 2 -float: 3 ldouble: 4 Function: "log10p1_towardzero": double: 3 -float: 2 ldouble: 3 Function: "log10p1_upward": double: 2 -float: 3 ldouble: 4 Function: "log1p": diff --git a/sysdeps/x86_64/fpu/libm-test-ulps b/sysdeps/x86_64/fpu/libm-test-ulps index ef9ef51e5e..e3c811549c 100644 --- a/sysdeps/x86_64/fpu/libm-test-ulps +++ b/sysdeps/x86_64/fpu/libm-test-ulps @@ -1798,25 +1798,21 @@ float: 1 Function: "log10p1": double: 2 -float: 2 float128: 3 ldouble: 4 Function: "log10p1_downward": double: 2 -float: 3 float128: 4 ldouble: 8 Function: "log10p1_towardzero": double: 3 -float: 2 float128: 3 ldouble: 8 Function: "log10p1_upward": double: 2 -float: 3 float128: 4 ldouble: 6

[15/17] math: Use log10p1f from CORE-MATH

Commit Message

Patch