From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from EUR02-VI1-obe.outbound.protection.outlook.com (mail-vi1eur02on2071.outbound.protection.outlook.com [40.107.241.71]) by sourceware.org (Postfix) with ESMTPS id 3AE0C3858D20 for ; Fri, 30 Jun 2023 07:52:55 +0000 (GMT) DMARC-Filter: OpenDMARC Filter v1.4.2 sourceware.org 3AE0C3858D20 Authentication-Results: sourceware.org; dmarc=pass (p=none dis=none) header.from=arm.com Authentication-Results: sourceware.org; spf=pass smtp.mailfrom=arm.com DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=armh.onmicrosoft.com; s=selector2-armh-onmicrosoft-com; h=From:Date:Subject:Message-ID:Content-Type:MIME-Version:X-MS-Exchange-SenderADCheck; bh=rFHU1CuraNXa/mWiKd78Mnr8FtHgyY3PbF3tZ48RbZE=; b=xEwBoVr54k/f4tZ73P9UPkORKAd00FU3hT9Qoml4FdOF409zmxiH1dI+avAJA1YnECV8GNex5r47bcnl5Zct3mUah2zaoNxMUO3dJUikWys/bTshlpc0iO3Tcx3tkajCszbEH4HkXs+AClX/75g/Z1z3DcHNHXmFvfTc0wypdjA= Received: from AS9PR01CA0041.eurprd01.prod.exchangelabs.com (2603:10a6:20b:542::15) by PAVPR08MB9818.eurprd08.prod.outlook.com (2603:10a6:102:31e::20) with Microsoft SMTP Server (version=TLS1_2, cipher=TLS_ECDHE_RSA_WITH_AES_256_GCM_SHA384) id 15.20.6521.26; Fri, 30 Jun 2023 07:52:50 +0000 Received: from AM7EUR03FT054.eop-EUR03.prod.protection.outlook.com (2603:10a6:20b:542:cafe::94) by AS9PR01CA0041.outlook.office365.com (2603:10a6:20b:542::15) with Microsoft SMTP Server (version=TLS1_2, cipher=TLS_ECDHE_RSA_WITH_AES_256_GCM_SHA384) id 15.20.6544.22 via Frontend Transport; Fri, 30 Jun 2023 07:52:50 +0000 X-MS-Exchange-Authentication-Results: spf=pass (sender IP is 63.35.35.123) smtp.mailfrom=arm.com; dkim=pass (signature was verified) header.d=armh.onmicrosoft.com;dmarc=pass action=none header.from=arm.com; Received-SPF: Pass (protection.outlook.com: domain of arm.com designates 63.35.35.123 as permitted sender) receiver=protection.outlook.com; client-ip=63.35.35.123; helo=64aa7808-outbound-1.mta.getcheckrecipient.com; pr=C Received: from 64aa7808-outbound-1.mta.getcheckrecipient.com (63.35.35.123) by AM7EUR03FT054.mail.protection.outlook.com (100.127.140.133) with Microsoft SMTP Server (version=TLS1_2, cipher=TLS_ECDHE_RSA_WITH_AES_256_GCM_SHA384) id 15.20.6544.22 via Frontend Transport; Fri, 30 Jun 2023 07:52:50 +0000 Received: ("Tessian outbound 7c913606c6e6:v142"); Fri, 30 Jun 2023 07:52:50 +0000 X-CheckRecipientChecked: true X-CR-MTA-CID: 10c3d390c40f3d2f X-CR-MTA-TID: 64aa7808 Received: from b82e9f0ee0e7.1 by 64aa7808-outbound-1.mta.getcheckrecipient.com id EA79FD33-37A1-409D-A8A2-EAFAD5396CD8.1; Fri, 30 Jun 2023 07:52:39 +0000 Received: from EUR05-AM6-obe.outbound.protection.outlook.com by 64aa7808-outbound-1.mta.getcheckrecipient.com with ESMTPS id b82e9f0ee0e7.1 (version=TLSv1.2 cipher=ECDHE-RSA-AES256-GCM-SHA384); Fri, 30 Jun 2023 07:52:38 +0000 ARC-Seal: i=1; a=rsa-sha256; s=arcselector9901; d=microsoft.com; cv=none; b=Ij5pbTwmZ5tPqychOoxr0AJgczS74Bd8nhCZZA40K1Kbsq8oVqamS0ZvP0m3mrE/DHdS0YYPzighZP0AYrmSSd02NpmJuljBAiPRG4fdGpWbXGbTP0BmsfGkhmgQ61CaE+ebULPizXKosmIIo7mngwXumGoFpEFRZsxaTz2n7jZKXex6YChFWrHEBrvP/qZbYnk6Hvj7PVewuxcdvo+m0KIx8wkZwnVPLvAenjifZmBu0am/f9SzGm55qFNIBsK1Vz7bplID+pn4jJgFU9L+MkurPz0u5X7P6mpXOr80QcxpnPwCstYRGcozt4yx6KK3gbQMZv8RZTYXmOEH6zuA/w== ARC-Message-Signature: i=1; a=rsa-sha256; c=relaxed/relaxed; d=microsoft.com; s=arcselector9901; h=From:Date:Subject:Message-ID:Content-Type:MIME-Version:X-MS-Exchange-AntiSpam-MessageData-ChunkCount:X-MS-Exchange-AntiSpam-MessageData-0:X-MS-Exchange-AntiSpam-MessageData-1; bh=rFHU1CuraNXa/mWiKd78Mnr8FtHgyY3PbF3tZ48RbZE=; b=LBeEH/JLAdXuQ0gMmRHFnRNeEdSfJwTU9ylU/U9XrvMP6mNjZGHjt4vabWiFfdM6Ok8BwCvn5XTQOlrZoH7UonnUq/nn5tuvFrdDUG01hNjME0Ls5AKA2uptclFgjqipuLQiwBR0bW9pD17+9ekkbr3aZMVDZ4AIy52JtENuj8BpIkSdqeY4Lh/Ok1aMogbhXjC8C5EXstkdzFj02zgebZcDAFvpUpwtMjzT1kjeEIFUIQe+d+Qta4b6o+xMK/9iom84brkW6RrApdsCpIrsVP3fV4agSFI06F5mMR3ObmngroRHKApCJoanx/PidwFzcS5TuKqFHNWqZUP/7S0F4A== ARC-Authentication-Results: i=1; mx.microsoft.com 1; spf=pass smtp.mailfrom=arm.com; dmarc=pass action=none header.from=arm.com; dkim=pass header.d=arm.com; arc=none DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=armh.onmicrosoft.com; s=selector2-armh-onmicrosoft-com; h=From:Date:Subject:Message-ID:Content-Type:MIME-Version:X-MS-Exchange-SenderADCheck; bh=rFHU1CuraNXa/mWiKd78Mnr8FtHgyY3PbF3tZ48RbZE=; b=xEwBoVr54k/f4tZ73P9UPkORKAd00FU3hT9Qoml4FdOF409zmxiH1dI+avAJA1YnECV8GNex5r47bcnl5Zct3mUah2zaoNxMUO3dJUikWys/bTshlpc0iO3Tcx3tkajCszbEH4HkXs+AClX/75g/Z1z3DcHNHXmFvfTc0wypdjA= Authentication-Results-Original: dkim=none (message not signed) header.d=none;dmarc=none action=none header.from=arm.com; Received: from DB9PR08MB7179.eurprd08.prod.outlook.com (2603:10a6:10:2cc::19) by DB8PR08MB5356.eurprd08.prod.outlook.com (2603:10a6:10:f9::15) with Microsoft SMTP Server (version=TLS1_2, cipher=TLS_ECDHE_RSA_WITH_AES_256_GCM_SHA384) id 15.20.6544.19; Fri, 30 Jun 2023 07:52:36 +0000 Received: from DB9PR08MB7179.eurprd08.prod.outlook.com ([fe80::43b7:3a83:5cbe:4559]) by DB9PR08MB7179.eurprd08.prod.outlook.com ([fe80::43b7:3a83:5cbe:4559%4]) with mapi id 15.20.6544.019; Fri, 30 Jun 2023 07:52:36 +0000 Date: Fri, 30 Jun 2023 08:52:18 +0100 From: Szabolcs Nagy To: Joe Ramsay , Subject: Re: [PATCH v4 2/4] aarch64: Add vector implementations of sin routines Message-ID: References: <20230628111939.48140-1-Joe.Ramsay@arm.com> <20230628111939.48140-2-Joe.Ramsay@arm.com> Content-Type: text/plain; charset=utf-8 Content-Disposition: inline In-Reply-To: <20230628111939.48140-2-Joe.Ramsay@arm.com> X-ClientProxiedBy: DS7PR05CA0091.namprd05.prod.outlook.com (2603:10b6:8:56::12) To DB9PR08MB7179.eurprd08.prod.outlook.com (2603:10a6:10:2cc::19) MIME-Version: 1.0 X-MS-TrafficTypeDiagnostic: DB9PR08MB7179:EE_|DB8PR08MB5356:EE_|AM7EUR03FT054:EE_|PAVPR08MB9818:EE_ X-MS-Office365-Filtering-Correlation-Id: 4046bca8-ab64-402c-5291-08db793f00b8 x-checkrecipientrouted: true NoDisclaimer: true X-MS-Exchange-SenderADCheck: 1 X-MS-Exchange-AntiSpam-Relay: 0 X-Microsoft-Antispam-Untrusted: BCL:0; X-Microsoft-Antispam-Message-Info-Original: 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 X-Forefront-Antispam-Report-Untrusted: CIP:255.255.255.255;CTRY:;LANG:en;SCL:1;SRV:;IPV:NLI;SFV:NSPM;H:DB9PR08MB7179.eurprd08.prod.outlook.com;PTR:;CAT:NONE;SFS:(13230028)(4636009)(346002)(39860400002)(136003)(396003)(376002)(366004)(451199021)(44832011)(66946007)(36756003)(66556008)(66476007)(478600001)(316002)(5660300002)(8676002)(8936002)(2906002)(30864003)(41300700001)(86362001)(110136005)(6486002)(6512007)(6506007)(26005)(186003)(38100700002)(6666004)(83380400001)(2616005)(2004002);DIR:OUT;SFP:1101; X-MS-Exchange-Transport-CrossTenantHeadersStamped: DB8PR08MB5356 Original-Authentication-Results: dkim=none (message not signed) header.d=none;dmarc=none action=none header.from=arm.com; X-EOPAttributedMessage: 0 X-MS-Exchange-Transport-CrossTenantHeadersStripped: AM7EUR03FT054.eop-EUR03.prod.protection.outlook.com X-MS-PublicTrafficType: Email X-MS-Office365-Filtering-Correlation-Id-Prvs: a42b27d7-f9ea-42a0-837f-08db793ef877 X-Microsoft-Antispam: BCL:0; X-Microsoft-Antispam-Message-Info: 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 X-Forefront-Antispam-Report: CIP:63.35.35.123;CTRY:IE;LANG:en;SCL:1;SRV:;IPV:CAL;SFV:NSPM;H:64aa7808-outbound-1.mta.getcheckrecipient.com;PTR:ec2-63-35-35-123.eu-west-1.compute.amazonaws.com;CAT:NONE;SFS:(13230028)(4636009)(346002)(136003)(39860400002)(396003)(376002)(451199021)(40470700004)(36840700001)(46966006)(82310400005)(26005)(36756003)(36860700001)(47076005)(6486002)(336012)(83380400001)(478600001)(2616005)(110136005)(6666004)(2906002)(6512007)(186003)(6506007)(5660300002)(70206006)(30864003)(356005)(316002)(82740400003)(81166007)(44832011)(70586007)(8676002)(8936002)(41300700001)(40460700003)(40480700001)(86362001)(2004002);DIR:OUT;SFP:1101; X-OriginatorOrg: arm.com X-MS-Exchange-CrossTenant-OriginalArrivalTime: 30 Jun 2023 07:52:50.4256 (UTC) X-MS-Exchange-CrossTenant-Network-Message-Id: 4046bca8-ab64-402c-5291-08db793f00b8 X-MS-Exchange-CrossTenant-Id: f34e5979-57d9-4aaa-ad4d-b122a662184d X-MS-Exchange-CrossTenant-OriginalAttributedTenantConnectingIp: TenantId=f34e5979-57d9-4aaa-ad4d-b122a662184d;Ip=[63.35.35.123];Helo=[64aa7808-outbound-1.mta.getcheckrecipient.com] X-MS-Exchange-CrossTenant-AuthSource: AM7EUR03FT054.eop-EUR03.prod.protection.outlook.com X-MS-Exchange-CrossTenant-AuthAs: Anonymous X-MS-Exchange-CrossTenant-FromEntityHeader: HybridOnPrem X-MS-Exchange-Transport-CrossTenantHeadersStamped: PAVPR08MB9818 X-Spam-Status: No, score=-11.6 required=5.0 tests=BAYES_00,DKIM_SIGNED,DKIM_VALID,FORGED_SPF_HELO,GIT_PATCH_0,KAM_DMARC_NONE,KAM_SHORT,RCVD_IN_DNSWL_NONE,RCVD_IN_MSPIKE_H2,SPF_HELO_PASS,SPF_NONE,TXREP,T_SCC_BODY_TEXT_LINE,UNPARSEABLE_RELAY autolearn=ham autolearn_force=no version=3.4.6 X-Spam-Checker-Version: SpamAssassin 3.4.6 (2021-04-09) on server2.sourceware.org List-Id: The 06/28/2023 12:19, Joe Ramsay via Libc-alpha wrote: > Optimised implementations for single and double precision, Advanced > SIMD and SVE, copied from Arm Optimized Routines. > > As previously, data tables are used via a barrier to prevent > overly aggressive constant inlining. Special-case handlers are > marked NOINLINE to avoid incurring the penalty of switching call > standards unnecessarily. Reviewed-by: Szabolcs Nagy > --- > Changes to v3: > * Remove volatile from AdvSIMD data - use ptr_barrier instead > Thanks, > Joe > sysdeps/aarch64/fpu/Makefile | 12 +- > sysdeps/aarch64/fpu/Versions | 4 + > sysdeps/aarch64/fpu/bits/math-vector.h | 6 + > sysdeps/aarch64/fpu/sin_advsimd.c | 106 ++++++++++++++++++ > sysdeps/aarch64/fpu/sin_sve.c | 97 ++++++++++++++++ > sysdeps/aarch64/fpu/sinf_advsimd.c | 99 ++++++++++++++++ > sysdeps/aarch64/fpu/sinf_sve.c | 96 ++++++++++++++++ > .../fpu/test-double-advsimd-wrappers.c | 1 + > .../aarch64/fpu/test-double-sve-wrappers.c | 1 + > .../aarch64/fpu/test-float-advsimd-wrappers.c | 1 + > sysdeps/aarch64/fpu/test-float-sve-wrappers.c | 1 + > sysdeps/aarch64/libm-test-ulps | 8 ++ > .../unix/sysv/linux/aarch64/libmvec.abilist | 4 + > 13 files changed, 430 insertions(+), 6 deletions(-) > create mode 100644 sysdeps/aarch64/fpu/sin_advsimd.c > create mode 100644 sysdeps/aarch64/fpu/sin_sve.c > create mode 100644 sysdeps/aarch64/fpu/sinf_advsimd.c > create mode 100644 sysdeps/aarch64/fpu/sinf_sve.c > > diff --git a/sysdeps/aarch64/fpu/Makefile b/sysdeps/aarch64/fpu/Makefile > index 850cfb9012..9ceea35148 100644 > --- a/sysdeps/aarch64/fpu/Makefile > +++ b/sysdeps/aarch64/fpu/Makefile > @@ -1,10 +1,10 @@ > -float-advsimd-funcs = cos > +libmvec-supported-funcs = cos \ > + sin > > -double-advsimd-funcs = cos > - > -float-sve-funcs = cos > - > -double-sve-funcs = cos > +float-advsimd-funcs = $(libmvec-supported-funcs) > +double-advsimd-funcs = $(libmvec-supported-funcs) > +float-sve-funcs = $(libmvec-supported-funcs) > +double-sve-funcs = $(libmvec-supported-funcs) > > ifeq ($(subdir),mathvec) > libmvec-support = $(addsuffix f_advsimd,$(float-advsimd-funcs)) \ > diff --git a/sysdeps/aarch64/fpu/Versions b/sysdeps/aarch64/fpu/Versions > index 5222a6f180..d26b3968a9 100644 > --- a/sysdeps/aarch64/fpu/Versions > +++ b/sysdeps/aarch64/fpu/Versions > @@ -1,8 +1,12 @@ > libmvec { > GLIBC_2.38 { > _ZGVnN2v_cos; > + _ZGVnN2v_sin; > _ZGVnN4v_cosf; > + _ZGVnN4v_sinf; > _ZGVsMxv_cos; > _ZGVsMxv_cosf; > + _ZGVsMxv_sin; > + _ZGVsMxv_sinf; > } > } > diff --git a/sysdeps/aarch64/fpu/bits/math-vector.h b/sysdeps/aarch64/fpu/bits/math-vector.h > index a2f2277591..ad9c9945e8 100644 > --- a/sysdeps/aarch64/fpu/bits/math-vector.h > +++ b/sysdeps/aarch64/fpu/bits/math-vector.h > @@ -50,7 +50,10 @@ typedef __SVBool_t __sv_bool_t; > # define __vpcs __attribute__ ((__aarch64_vector_pcs__)) > > __vpcs __f32x4_t _ZGVnN4v_cosf (__f32x4_t); > +__vpcs __f32x4_t _ZGVnN4v_sinf (__f32x4_t); > + > __vpcs __f64x2_t _ZGVnN2v_cos (__f64x2_t); > +__vpcs __f64x2_t _ZGVnN2v_sin (__f64x2_t); > > # undef __ADVSIMD_VEC_MATH_SUPPORTED > #endif /* __ADVSIMD_VEC_MATH_SUPPORTED */ > @@ -58,7 +61,10 @@ __vpcs __f64x2_t _ZGVnN2v_cos (__f64x2_t); > #ifdef __SVE_VEC_MATH_SUPPORTED > > __sv_f32_t _ZGVsMxv_cosf (__sv_f32_t, __sv_bool_t); > +__sv_f32_t _ZGVsMxv_sinf (__sv_f32_t, __sv_bool_t); > + > __sv_f64_t _ZGVsMxv_cos (__sv_f64_t, __sv_bool_t); > +__sv_f64_t _ZGVsMxv_sin (__sv_f64_t, __sv_bool_t); > > # undef __SVE_VEC_MATH_SUPPORTED > #endif /* __SVE_VEC_MATH_SUPPORTED */ > diff --git a/sysdeps/aarch64/fpu/sin_advsimd.c b/sysdeps/aarch64/fpu/sin_advsimd.c > new file mode 100644 > index 0000000000..ddc4142599 > --- /dev/null > +++ b/sysdeps/aarch64/fpu/sin_advsimd.c > @@ -0,0 +1,106 @@ > +/* Double-precision vector (Advanced SIMD) sin function. > + > + Copyright (C) 2023 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 > + . */ > + > +#include "v_math.h" > + > +static const struct data > +{ > + float64x2_t poly[7]; > + float64x2_t range_val, inv_pi, shift, pi_1, pi_2, pi_3; > +} data = { > + /* Worst-case error is 2.8 ulp in [-pi/2, pi/2]. */ > + .poly = { V2 (-0x1.555555555547bp-3), V2 (0x1.1111111108a4dp-7), > + V2 (-0x1.a01a019936f27p-13), V2 (0x1.71de37a97d93ep-19), > + V2 (-0x1.ae633919987c6p-26), V2 (0x1.60e277ae07cecp-33), > + V2 (-0x1.9e9540300a1p-41) }, > + > + .range_val = V2 (0x1p23), > + .inv_pi = V2 (0x1.45f306dc9c883p-2), > + .pi_1 = V2 (0x1.921fb54442d18p+1), > + .pi_2 = V2 (0x1.1a62633145c06p-53), > + .pi_3 = V2 (0x1.c1cd129024e09p-106), > + .shift = V2 (0x1.8p52), > +}; > + > +#if WANT_SIMD_EXCEPT > +# define TinyBound v_u64 (0x3000000000000000) /* asuint64 (0x1p-255). */ > +# define Thresh v_u64 (0x1160000000000000) /* RangeVal - TinyBound. */ > +#endif > + > +#define C(i) d->poly[i] > + > +static float64x2_t VPCS_ATTR NOINLINE > +special_case (float64x2_t x, float64x2_t y, uint64x2_t odd, uint64x2_t cmp) > +{ > + y = vreinterpretq_f64_u64 (veorq_u64 (vreinterpretq_u64_f64 (y), odd)); > + return v_call_f64 (sin, x, y, cmp); > +} > + > +float64x2_t VPCS_ATTR V_NAME_D1 (sin) (float64x2_t x) > +{ > + const struct data *d = ptr_barrier (&data); > + float64x2_t n, r, r2, r3, r4, y, t1, t2, t3; > + uint64x2_t odd, cmp, eqz; > + > +#if WANT_SIMD_EXCEPT > + /* Detect |x| <= TinyBound or |x| >= RangeVal. If fenv exceptions are to be > + triggered correctly, set any special lanes to 1 (which is neutral w.r.t. > + fenv). These lanes will be fixed by special-case handler later. */ > + uint64x2_t ir = vreinterpretq_u64_f64 (vabsq_f64 (x)); > + cmp = vcgeq_u64 (vsubq_u64 (ir, TinyBound), Thresh); > + r = vbslq_f64 (cmp, vreinterpretq_f64_u64 (cmp), x); > +#else > + r = x; > + cmp = vcageq_f64 (d->range_val, x); > + cmp = vceqzq_u64 (cmp); /* cmp = ~cmp. */ > +#endif > + eqz = vceqzq_f64 (x); > + > + /* n = rint(|x|/pi). */ > + n = vfmaq_f64 (d->shift, d->inv_pi, r); > + odd = vshlq_n_u64 (vreinterpretq_u64_f64 (n), 63); > + n = vsubq_f64 (n, d->shift); > + > + /* r = |x| - n*pi (range reduction into -pi/2 .. pi/2). */ > + r = vfmsq_f64 (r, d->pi_1, n); > + r = vfmsq_f64 (r, d->pi_2, n); > + r = vfmsq_f64 (r, d->pi_3, n); > + > + /* sin(r) poly approx. */ > + r2 = vmulq_f64 (r, r); > + r3 = vmulq_f64 (r2, r); > + r4 = vmulq_f64 (r2, r2); > + > + t1 = vfmaq_f64 (C (4), C (5), r2); > + t2 = vfmaq_f64 (C (2), C (3), r2); > + t3 = vfmaq_f64 (C (0), C (1), r2); > + > + y = vfmaq_f64 (t1, C (6), r4); > + y = vfmaq_f64 (t2, y, r4); > + y = vfmaq_f64 (t3, y, r4); > + y = vfmaq_f64 (r, y, r3); > + > + /* Sign of 0 is discarded by polynomial, so copy it back here. */ > + if (__glibc_unlikely (v_any_u64 (eqz))) > + y = vbslq_f64 (eqz, x, y); > + > + if (__glibc_unlikely (v_any_u64 (cmp))) > + return special_case (x, y, odd, cmp); > + return vreinterpretq_f64_u64 (veorq_u64 (vreinterpretq_u64_f64 (y), odd)); > +} > diff --git a/sysdeps/aarch64/fpu/sin_sve.c b/sysdeps/aarch64/fpu/sin_sve.c > new file mode 100644 > index 0000000000..c3f450d0ea > --- /dev/null > +++ b/sysdeps/aarch64/fpu/sin_sve.c > @@ -0,0 +1,97 @@ > +/* Double-precision vector (SVE) sin function. > + > + Copyright (C) 2023 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 > + . */ > + > +#include "sv_math.h" > + > +static const struct data > +{ > + double inv_pi, half_pi, inv_pi_over_2, pi_over_2_1, pi_over_2_2, pi_over_2_3, > + shift; > +} data = { > + /* Polynomial coefficients are hard-wired in the FTMAD instruction. */ > + .inv_pi = 0x1.45f306dc9c883p-2, > + .half_pi = 0x1.921fb54442d18p+0, > + .inv_pi_over_2 = 0x1.45f306dc9c882p-1, > + .pi_over_2_1 = 0x1.921fb50000000p+0, > + .pi_over_2_2 = 0x1.110b460000000p-26, > + .pi_over_2_3 = 0x1.1a62633145c07p-54, > + .shift = 0x1.8p52 > +}; > + > +#define RangeVal 0x4160000000000000 /* asuint64 (0x1p23). */ > + > +static svfloat64_t NOINLINE > +special_case (svfloat64_t x, svfloat64_t y, svbool_t cmp) > +{ > + return sv_call_f64 (sin, x, y, cmp); > +} > + > +/* A fast SVE implementation of sin based on trigonometric > + instructions (FTMAD, FTSSEL, FTSMUL). > + Maximum observed error in 2.52 ULP: > + SV_NAME_D1 (sin)(0x1.2d2b00df69661p+19) got 0x1.10ace8f3e786bp-40 > + want 0x1.10ace8f3e7868p-40. */ > +svfloat64_t SV_NAME_D1 (sin) (svfloat64_t x, const svbool_t pg) > +{ > + const struct data *d = ptr_barrier (&data); > + > + svfloat64_t r = svabs_f64_x (pg, x); > + svuint64_t sign > + = sveor_u64_x (pg, svreinterpret_u64_f64 (x), svreinterpret_u64_f64 (r)); > + svbool_t cmp = svcmpge_n_u64 (pg, svreinterpret_u64_f64 (r), RangeVal); > + > + /* Load first two pio2-related constants to one vector. */ > + svfloat64_t invpio2_and_pio2_1 > + = svld1rq_f64 (svptrue_b64 (), &d->inv_pi_over_2); > + > + /* n = rint(|x|/(pi/2)). */ > + svfloat64_t q = svmla_lane_f64 (sv_f64 (d->shift), r, invpio2_and_pio2_1, 0); > + svfloat64_t n = svsub_n_f64_x (pg, q, d->shift); > + > + /* r = |x| - n*(pi/2) (range reduction into -pi/4 .. pi/4). */ > + r = svmls_lane_f64 (r, n, invpio2_and_pio2_1, 1); > + r = svmls_n_f64_x (pg, r, n, d->pi_over_2_2); > + r = svmls_n_f64_x (pg, r, n, d->pi_over_2_3); > + > + /* Final multiplicative factor: 1.0 or x depending on bit #0 of q. */ > + svfloat64_t f = svtssel_f64 (r, svreinterpret_u64_f64 (q)); > + > + /* sin(r) poly approx. */ > + svfloat64_t r2 = svtsmul_f64 (r, svreinterpret_u64_f64 (q)); > + svfloat64_t y = sv_f64 (0.0); > + y = svtmad_f64 (y, r2, 7); > + y = svtmad_f64 (y, r2, 6); > + y = svtmad_f64 (y, r2, 5); > + y = svtmad_f64 (y, r2, 4); > + y = svtmad_f64 (y, r2, 3); > + y = svtmad_f64 (y, r2, 2); > + y = svtmad_f64 (y, r2, 1); > + y = svtmad_f64 (y, r2, 0); > + > + /* Apply factor. */ > + y = svmul_f64_x (pg, f, y); > + > + /* sign = y^sign. */ > + y = svreinterpret_f64_u64 ( > + sveor_u64_x (pg, svreinterpret_u64_f64 (y), sign)); > + > + if (__glibc_unlikely (svptest_any (pg, cmp))) > + return special_case (x, y, cmp); > + return y; > +} > diff --git a/sysdeps/aarch64/fpu/sinf_advsimd.c b/sysdeps/aarch64/fpu/sinf_advsimd.c > new file mode 100644 > index 0000000000..b67d37f2fd > --- /dev/null > +++ b/sysdeps/aarch64/fpu/sinf_advsimd.c > @@ -0,0 +1,99 @@ > +/* Single-precision vector (Advanced SIMD) sin function. > + > + Copyright (C) 2023 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 > + . */ > + > +#include "v_math.h" > + > +static const struct data > +{ > + float32x4_t poly[4]; > + float32x4_t range_val, inv_pi, shift, pi_1, pi_2, pi_3; > +} data = { > + /* 1.886 ulp error. */ > + .poly = { V4 (-0x1.555548p-3f), V4 (0x1.110df4p-7f), V4 (-0x1.9f42eap-13f), > + V4 (0x1.5b2e76p-19f) }, > + > + .pi_1 = V4 (0x1.921fb6p+1f), > + .pi_2 = V4 (-0x1.777a5cp-24f), > + .pi_3 = V4 (-0x1.ee59dap-49f), > + > + .inv_pi = V4 (0x1.45f306p-2f), > + .shift = V4 (0x1.8p+23f), > + .range_val = V4 (0x1p20f) > +}; > + > +#if WANT_SIMD_EXCEPT > +# define TinyBound v_u32 (0x21000000) /* asuint32(0x1p-61f). */ > +# define Thresh v_u32 (0x28800000) /* RangeVal - TinyBound. */ > +#endif > + > +#define C(i) d->poly[i] > + > +static float32x4_t VPCS_ATTR NOINLINE > +special_case (float32x4_t x, float32x4_t y, uint32x4_t odd, uint32x4_t cmp) > +{ > + /* Fall back to scalar code. */ > + y = vreinterpretq_f32_u32 (veorq_u32 (vreinterpretq_u32_f32 (y), odd)); > + return v_call_f32 (sinf, x, y, cmp); > +} > + > +float32x4_t VPCS_ATTR V_NAME_F1 (sin) (float32x4_t x) > +{ > + const struct data *d = ptr_barrier (&data); > + float32x4_t n, r, r2, y; > + uint32x4_t odd, cmp, eqz; > + > +#if WANT_SIMD_EXCEPT > + uint32x4_t ir = vreinterpretq_u32_f32 (vabsq_f32 (x)); > + cmp = vcgeq_u32 (vsubq_u32 (ir, TinyBound), Thresh); > + /* If fenv exceptions are to be triggered correctly, set any special lanes > + to 1 (which is neutral w.r.t. fenv). These lanes will be fixed by > + special-case handler later. */ > + r = vbslq_f32 (cmp, vreinterpretq_f32_u32 (cmp), x); > +#else > + r = x; > + cmp = vcageq_f32 (d->range_val, x); > + cmp = vceqzq_u32 (cmp); /* cmp = ~cmp. */ > +#endif > + eqz = vceqzq_f32 (x); > + > + /* n = rint(|x|/pi) */ > + n = vfmaq_f32 (d->shift, d->inv_pi, r); > + odd = vshlq_n_u32 (vreinterpretq_u32_f32 (n), 31); > + n = vsubq_f32 (n, d->shift); > + > + /* r = |x| - n*pi (range reduction into -pi/2 .. pi/2) */ > + r = vfmsq_f32 (r, d->pi_1, n); > + r = vfmsq_f32 (r, d->pi_2, n); > + r = vfmsq_f32 (r, d->pi_3, n); > + > + /* y = sin(r) */ > + r2 = vmulq_f32 (r, r); > + y = vfmaq_f32 (C (2), C (3), r2); > + y = vfmaq_f32 (C (1), y, r2); > + y = vfmaq_f32 (C (0), y, r2); > + y = vfmaq_f32 (r, vmulq_f32 (y, r2), r); > + > + /* Sign of 0 is discarded by polynomial, so copy it back here. */ > + if (__glibc_unlikely (v_any_u32 (eqz))) > + y = vbslq_f32 (eqz, x, y); > + > + if (__glibc_unlikely (v_any_u32 (cmp))) > + return special_case (x, y, odd, cmp); > + return vreinterpretq_f32_u32 (veorq_u32 (vreinterpretq_u32_f32 (y), odd)); > +} > diff --git a/sysdeps/aarch64/fpu/sinf_sve.c b/sysdeps/aarch64/fpu/sinf_sve.c > new file mode 100644 > index 0000000000..4d2ce7a846 > --- /dev/null > +++ b/sysdeps/aarch64/fpu/sinf_sve.c > @@ -0,0 +1,96 @@ > +/* Single-precision vector (SVE) sin function. > + > + Copyright (C) 2023 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 > + . */ > + > +#include "sv_math.h" > + > +static const struct data > +{ > + float poly[4]; > + /* Pi-related values to be loaded as one quad-word and used with > + svmla_lane_f32. */ > + float negpi1, negpi2, negpi3, invpi; > + float shift; > +} data = { > + .poly = { > + /* Non-zero coefficients from the degree 9 Taylor series expansion of > + sin. */ > + -0x1.555548p-3f, 0x1.110df4p-7f, -0x1.9f42eap-13f, 0x1.5b2e76p-19f > + }, > + .negpi1 = -0x1.921fb6p+1f, > + .negpi2 = 0x1.777a5cp-24f, > + .negpi3 = 0x1.ee59dap-49f, > + .invpi = 0x1.45f306p-2f, > + .shift = 0x1.8p+23f > +}; > + > +#define RangeVal 0x49800000 /* asuint32 (0x1p20f). */ > +#define C(i) sv_f32 (d->poly[i]) > + > +static svfloat32_t NOINLINE > +special_case (svfloat32_t x, svfloat32_t y, svbool_t cmp) > +{ > + return sv_call_f32 (sinf, x, y, cmp); > +} > + > +/* A fast SVE implementation of sinf. > + Maximum error: 1.89 ULPs. > + This maximum error is achieved at multiple values in [-2^18, 2^18] > + but one example is: > + SV_NAME_F1 (sin)(0x1.9247a4p+0) got 0x1.fffff6p-1 want 0x1.fffffap-1. */ > +svfloat32_t SV_NAME_F1 (sin) (svfloat32_t x, const svbool_t pg) > +{ > + const struct data *d = ptr_barrier (&data); > + > + svfloat32_t ax = svabs_f32_x (pg, x); > + svuint32_t sign = sveor_u32_x (pg, svreinterpret_u32_f32 (x), > + svreinterpret_u32_f32 (ax)); > + svbool_t cmp = svcmpge_n_u32 (pg, svreinterpret_u32_f32 (ax), RangeVal); > + > + /* pi_vals are a quad-word of helper values - the first 3 elements contain > + -pi in extended precision, the last contains 1 / pi. */ > + svfloat32_t pi_vals = svld1rq_f32 (svptrue_b32 (), &d->negpi1); > + > + /* n = rint(|x|/pi). */ > + svfloat32_t n = svmla_lane_f32 (sv_f32 (d->shift), ax, pi_vals, 3); > + svuint32_t odd = svlsl_n_u32_x (pg, svreinterpret_u32_f32 (n), 31); > + n = svsub_n_f32_x (pg, n, d->shift); > + > + /* r = |x| - n*pi (range reduction into -pi/2 .. pi/2). */ > + svfloat32_t r; > + r = svmla_lane_f32 (ax, n, pi_vals, 0); > + r = svmla_lane_f32 (r, n, pi_vals, 1); > + r = svmla_lane_f32 (r, n, pi_vals, 2); > + > + /* sin(r) approx using a degree 9 polynomial from the Taylor series > + expansion. Note that only the odd terms of this are non-zero. */ > + svfloat32_t r2 = svmul_f32_x (pg, r, r); > + svfloat32_t y; > + y = svmla_f32_x (pg, C (2), r2, C (3)); > + y = svmla_f32_x (pg, C (1), r2, y); > + y = svmla_f32_x (pg, C (0), r2, y); > + y = svmla_f32_x (pg, r, r, svmul_f32_x (pg, y, r2)); > + > + /* sign = y^sign^odd. */ > + y = svreinterpret_f32_u32 (sveor_u32_x (pg, svreinterpret_u32_f32 (y), > + sveor_u32_x (pg, sign, odd))); > + > + if (__glibc_unlikely (svptest_any (pg, cmp))) > + return special_case (x, y, cmp); > + return y; > +} > diff --git a/sysdeps/aarch64/fpu/test-double-advsimd-wrappers.c b/sysdeps/aarch64/fpu/test-double-advsimd-wrappers.c > index cb45fd3298..4af97a25a2 100644 > --- a/sysdeps/aarch64/fpu/test-double-advsimd-wrappers.c > +++ b/sysdeps/aarch64/fpu/test-double-advsimd-wrappers.c > @@ -24,3 +24,4 @@ > #define VEC_TYPE float64x2_t > > VPCS_VECTOR_WRAPPER (cos_advsimd, _ZGVnN2v_cos) > +VPCS_VECTOR_WRAPPER (sin_advsimd, _ZGVnN2v_sin) > diff --git a/sysdeps/aarch64/fpu/test-double-sve-wrappers.c b/sysdeps/aarch64/fpu/test-double-sve-wrappers.c > index cf72ef83b7..64c790adc5 100644 > --- a/sysdeps/aarch64/fpu/test-double-sve-wrappers.c > +++ b/sysdeps/aarch64/fpu/test-double-sve-wrappers.c > @@ -33,3 +33,4 @@ > } > > SVE_VECTOR_WRAPPER (cos_sve, _ZGVsMxv_cos) > +SVE_VECTOR_WRAPPER (sin_sve, _ZGVsMxv_sin) > diff --git a/sysdeps/aarch64/fpu/test-float-advsimd-wrappers.c b/sysdeps/aarch64/fpu/test-float-advsimd-wrappers.c > index fa146862b0..50e776b952 100644 > --- a/sysdeps/aarch64/fpu/test-float-advsimd-wrappers.c > +++ b/sysdeps/aarch64/fpu/test-float-advsimd-wrappers.c > @@ -24,3 +24,4 @@ > #define VEC_TYPE float32x4_t > > VPCS_VECTOR_WRAPPER (cosf_advsimd, _ZGVnN4v_cosf) > +VPCS_VECTOR_WRAPPER (sinf_advsimd, _ZGVnN4v_sinf) > diff --git a/sysdeps/aarch64/fpu/test-float-sve-wrappers.c b/sysdeps/aarch64/fpu/test-float-sve-wrappers.c > index bc26558c62..7355032929 100644 > --- a/sysdeps/aarch64/fpu/test-float-sve-wrappers.c > +++ b/sysdeps/aarch64/fpu/test-float-sve-wrappers.c > @@ -33,3 +33,4 @@ > } > > SVE_VECTOR_WRAPPER (cosf_sve, _ZGVsMxv_cosf) > +SVE_VECTOR_WRAPPER (sinf_sve, _ZGVsMxv_sinf) > diff --git a/sysdeps/aarch64/libm-test-ulps b/sysdeps/aarch64/libm-test-ulps > index 07da4ab843..4145662b2d 100644 > --- a/sysdeps/aarch64/libm-test-ulps > +++ b/sysdeps/aarch64/libm-test-ulps > @@ -1257,11 +1257,19 @@ double: 1 > float: 1 > ldouble: 2 > > +Function: "sin_advsimd": > +double: 2 > +float: 1 > + > Function: "sin_downward": > double: 1 > float: 1 > ldouble: 3 > > +Function: "sin_sve": > +double: 2 > +float: 1 > + > Function: "sin_towardzero": > double: 1 > float: 1 > diff --git a/sysdeps/unix/sysv/linux/aarch64/libmvec.abilist b/sysdeps/unix/sysv/linux/aarch64/libmvec.abilist > index 13af421af2..a4c564859c 100644 > --- a/sysdeps/unix/sysv/linux/aarch64/libmvec.abilist > +++ b/sysdeps/unix/sysv/linux/aarch64/libmvec.abilist > @@ -1,4 +1,8 @@ > GLIBC_2.38 _ZGVnN2v_cos F > +GLIBC_2.38 _ZGVnN2v_sin F > GLIBC_2.38 _ZGVnN4v_cosf F > +GLIBC_2.38 _ZGVnN4v_sinf F > GLIBC_2.38 _ZGVsMxv_cos F > GLIBC_2.38 _ZGVsMxv_cosf F > +GLIBC_2.38 _ZGVsMxv_sin F > +GLIBC_2.38 _ZGVsMxv_sinf F > -- > 2.27.0 >