From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from mail-pl1-x62d.google.com (mail-pl1-x62d.google.com [IPv6:2607:f8b0:4864:20::62d]) by sourceware.org (Postfix) with ESMTPS id 8785C3858400 for ; Wed, 29 Dec 2021 21:26:16 +0000 (GMT) DMARC-Filter: OpenDMARC Filter v1.4.1 sourceware.org 8785C3858400 Received: by mail-pl1-x62d.google.com with SMTP id l16so4564283plg.10 for ; Wed, 29 Dec 2021 13:26:16 -0800 (PST) X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20210112; h=x-gm-message-state:date:from:to:cc:subject:message-id:references :mime-version:content-disposition:in-reply-to; bh=vRNIdb6nlV+lSkYt5VDmI3vsPmAbAuH23sILLER3uR0=; b=PBx72eDy0VkpfLA+YaTWebkMkilyb1w3nWmWwnfs8CB1HSjMtjdOr778yuXKjwqm4S ObMcjobno00UHtQokAcB0Pn2hPYM5YEHQE/EgzLAzj4j0ynfwuNvrwqdITEWl95nh8RA BAobPoUvYj9CX/BJGt4OSJVThTTuYKbQhY5N7eYvdIgVtbuNePeiOs2D0VsVOAbtEW8C A9Fokr355LDS5ykrFDakdn6/wmlQfFqJPBpIcsw4irgBvhVAPclmf6y5vvGXyZNdzyOu UFjzBUkqMrHQKu6wm5Ic/gvbd22vywdHRZ0cqJ3kkvhz4/9iMi0lPERZ8zoHGfTsojPL a7GQ== X-Gm-Message-State: AOAM532wttmpYgwo7tuPP3ECsG6pkPM1auzx+UVWerN1BHACmsmYgvLf FYj1/r05hUhReAvuFb649bk= X-Google-Smtp-Source: ABdhPJwmNks2BBZRXEW2Pc++2UfA76bLFHUy1VFcMzdI+F7b/RLgP1CqabdgFfpCL0Ha6hKL4yUQ7A== X-Received: by 2002:a17:90b:38cd:: with SMTP id nn13mr34435862pjb.37.1640813174879; Wed, 29 Dec 2021 13:26:14 -0800 (PST) Received: from gnu-tgl-3.localdomain ([172.58.35.133]) by smtp.gmail.com with ESMTPSA id w127sm25040210pfw.149.2021.12.29.13.26.14 (version=TLS1_3 cipher=TLS_AES_256_GCM_SHA384 bits=256/256); Wed, 29 Dec 2021 13:26:14 -0800 (PST) Received: by gnu-tgl-3.localdomain (Postfix, from userid 1000) id 52ADFC08AD; Wed, 29 Dec 2021 13:26:13 -0800 (PST) Date: Wed, 29 Dec 2021 13:26:13 -0800 From: "H.J. Lu" To: Sunil K Pandey Cc: libc-alpha@sourceware.org, hjl.tools@gmail.com, andrey.kolesov@intel.com, marius.cornea@intel.com Subject: Re: [PATCH v5 06/18] x86-64: Add vector cosh/coshf implementation to libmvec Message-ID: References: <20211229064000.1465621-1-skpgkp2@gmail.com> <20211229064000.1465621-7-skpgkp2@gmail.com> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20211229064000.1465621-7-skpgkp2@gmail.com> X-Spam-Status: No, score=-3028.1 required=5.0 tests=BAYES_00, DKIM_SIGNED, DKIM_VALID, DKIM_VALID_AU, DKIM_VALID_EF, FREEMAIL_FROM, GIT_PATCH_0, KAM_SHORT, KAM_STOCKGEN, RCVD_IN_BARRACUDACENTRAL, RCVD_IN_DNSWL_NONE, SPF_HELO_NONE, SPF_PASS, TXREP autolearn=ham autolearn_force=no version=3.4.4 X-Spam-Checker-Version: SpamAssassin 3.4.4 (2020-01-24) on server2.sourceware.org X-BeenThere: libc-alpha@sourceware.org X-Mailman-Version: 2.1.29 Precedence: list List-Id: Libc-alpha mailing list List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Wed, 29 Dec 2021 21:26:22 -0000 On Tue, Dec 28, 2021 at 10:39:48PM -0800, Sunil K Pandey wrote: > Implement vectorized cosh/coshf containing SSE, AVX, AVX2 and > AVX512 versions for libmvec as per vector ABI. It also contains > accuracy and ABI tests for vector cosh/coshf with regenerated ulps. > --- > bits/libm-simd-decl-stubs.h | 11 + > math/bits/mathcalls.h | 2 +- > .../unix/sysv/linux/x86_64/libmvec.abilist | 8 + > sysdeps/x86/fpu/bits/math-vector.h | 4 + > .../x86/fpu/finclude/math-vector-fortran.h | 4 + > sysdeps/x86_64/fpu/Makeconfig | 1 + > sysdeps/x86_64/fpu/Versions | 2 + > sysdeps/x86_64/fpu/libm-test-ulps | 20 + > .../fpu/multiarch/svml_d_cosh2_core-sse2.S | 20 + > .../x86_64/fpu/multiarch/svml_d_cosh2_core.c | 27 ++ > .../fpu/multiarch/svml_d_cosh2_core_sse4.S | 396 +++++++++++++++++ > .../fpu/multiarch/svml_d_cosh4_core-sse.S | 20 + > .../x86_64/fpu/multiarch/svml_d_cosh4_core.c | 27 ++ > .../fpu/multiarch/svml_d_cosh4_core_avx2.S | 412 ++++++++++++++++++ > .../fpu/multiarch/svml_d_cosh8_core-avx2.S | 20 + > .../x86_64/fpu/multiarch/svml_d_cosh8_core.c | 27 ++ > .../fpu/multiarch/svml_d_cosh8_core_avx512.S | 323 ++++++++++++++ > .../fpu/multiarch/svml_s_coshf16_core-avx2.S | 20 + > .../fpu/multiarch/svml_s_coshf16_core.c | 28 ++ > .../multiarch/svml_s_coshf16_core_avx512.S | 321 ++++++++++++++ > .../fpu/multiarch/svml_s_coshf4_core-sse2.S | 20 + > .../x86_64/fpu/multiarch/svml_s_coshf4_core.c | 28 ++ > .../fpu/multiarch/svml_s_coshf4_core_sse4.S | 305 +++++++++++++ > .../fpu/multiarch/svml_s_coshf8_core-sse.S | 20 + > .../x86_64/fpu/multiarch/svml_s_coshf8_core.c | 28 ++ > .../fpu/multiarch/svml_s_coshf8_core_avx2.S | 308 +++++++++++++ > sysdeps/x86_64/fpu/svml_d_cosh2_core.S | 29 ++ > sysdeps/x86_64/fpu/svml_d_cosh4_core.S | 29 ++ > sysdeps/x86_64/fpu/svml_d_cosh4_core_avx.S | 25 ++ > sysdeps/x86_64/fpu/svml_d_cosh8_core.S | 25 ++ > sysdeps/x86_64/fpu/svml_s_coshf16_core.S | 25 ++ > sysdeps/x86_64/fpu/svml_s_coshf4_core.S | 29 ++ > sysdeps/x86_64/fpu/svml_s_coshf8_core.S | 29 ++ > sysdeps/x86_64/fpu/svml_s_coshf8_core_avx.S | 25 ++ > .../x86_64/fpu/test-double-libmvec-cosh-avx.c | 1 + > .../fpu/test-double-libmvec-cosh-avx2.c | 1 + > .../fpu/test-double-libmvec-cosh-avx512f.c | 1 + > sysdeps/x86_64/fpu/test-double-libmvec-cosh.c | 3 + > .../x86_64/fpu/test-double-vlen2-wrappers.c | 1 + > .../fpu/test-double-vlen4-avx2-wrappers.c | 1 + > .../x86_64/fpu/test-double-vlen4-wrappers.c | 1 + > .../x86_64/fpu/test-double-vlen8-wrappers.c | 1 + > .../x86_64/fpu/test-float-libmvec-coshf-avx.c | 1 + > .../fpu/test-float-libmvec-coshf-avx2.c | 1 + > .../fpu/test-float-libmvec-coshf-avx512f.c | 1 + > sysdeps/x86_64/fpu/test-float-libmvec-coshf.c | 3 + > .../x86_64/fpu/test-float-vlen16-wrappers.c | 1 + > .../x86_64/fpu/test-float-vlen4-wrappers.c | 1 + > .../fpu/test-float-vlen8-avx2-wrappers.c | 1 + > .../x86_64/fpu/test-float-vlen8-wrappers.c | 1 + > 50 files changed, 2637 insertions(+), 1 deletion(-) > create mode 100644 sysdeps/x86_64/fpu/multiarch/svml_d_cosh2_core-sse2.S > create mode 100644 sysdeps/x86_64/fpu/multiarch/svml_d_cosh2_core.c > create mode 100644 sysdeps/x86_64/fpu/multiarch/svml_d_cosh2_core_sse4.S > create mode 100644 sysdeps/x86_64/fpu/multiarch/svml_d_cosh4_core-sse.S > create mode 100644 sysdeps/x86_64/fpu/multiarch/svml_d_cosh4_core.c > create mode 100644 sysdeps/x86_64/fpu/multiarch/svml_d_cosh4_core_avx2.S > create mode 100644 sysdeps/x86_64/fpu/multiarch/svml_d_cosh8_core-avx2.S > create mode 100644 sysdeps/x86_64/fpu/multiarch/svml_d_cosh8_core.c > create mode 100644 sysdeps/x86_64/fpu/multiarch/svml_d_cosh8_core_avx512.S > create mode 100644 sysdeps/x86_64/fpu/multiarch/svml_s_coshf16_core-avx2.S > create mode 100644 sysdeps/x86_64/fpu/multiarch/svml_s_coshf16_core.c > create mode 100644 sysdeps/x86_64/fpu/multiarch/svml_s_coshf16_core_avx512.S > create mode 100644 sysdeps/x86_64/fpu/multiarch/svml_s_coshf4_core-sse2.S > create mode 100644 sysdeps/x86_64/fpu/multiarch/svml_s_coshf4_core.c > create mode 100644 sysdeps/x86_64/fpu/multiarch/svml_s_coshf4_core_sse4.S > create mode 100644 sysdeps/x86_64/fpu/multiarch/svml_s_coshf8_core-sse.S > create mode 100644 sysdeps/x86_64/fpu/multiarch/svml_s_coshf8_core.c > create mode 100644 sysdeps/x86_64/fpu/multiarch/svml_s_coshf8_core_avx2.S > create mode 100644 sysdeps/x86_64/fpu/svml_d_cosh2_core.S > create mode 100644 sysdeps/x86_64/fpu/svml_d_cosh4_core.S > create mode 100644 sysdeps/x86_64/fpu/svml_d_cosh4_core_avx.S > create mode 100644 sysdeps/x86_64/fpu/svml_d_cosh8_core.S > create mode 100644 sysdeps/x86_64/fpu/svml_s_coshf16_core.S > create mode 100644 sysdeps/x86_64/fpu/svml_s_coshf4_core.S > create mode 100644 sysdeps/x86_64/fpu/svml_s_coshf8_core.S > create mode 100644 sysdeps/x86_64/fpu/svml_s_coshf8_core_avx.S > create mode 100644 sysdeps/x86_64/fpu/test-double-libmvec-cosh-avx.c > create mode 100644 sysdeps/x86_64/fpu/test-double-libmvec-cosh-avx2.c > create mode 100644 sysdeps/x86_64/fpu/test-double-libmvec-cosh-avx512f.c > create mode 100644 sysdeps/x86_64/fpu/test-double-libmvec-cosh.c > create mode 100644 sysdeps/x86_64/fpu/test-float-libmvec-coshf-avx.c > create mode 100644 sysdeps/x86_64/fpu/test-float-libmvec-coshf-avx2.c > create mode 100644 sysdeps/x86_64/fpu/test-float-libmvec-coshf-avx512f.c > create mode 100644 sysdeps/x86_64/fpu/test-float-libmvec-coshf.c > > diff --git a/bits/libm-simd-decl-stubs.h b/bits/libm-simd-decl-stubs.h > index bc18621f17..35c6ac57a8 100644 > --- a/bits/libm-simd-decl-stubs.h > +++ b/bits/libm-simd-decl-stubs.h > @@ -164,4 +164,15 @@ > #define __DECL_SIMD_exp10f32x > #define __DECL_SIMD_exp10f64x > #define __DECL_SIMD_exp10f128x > + > +#define __DECL_SIMD_cosh > +#define __DECL_SIMD_coshf > +#define __DECL_SIMD_coshl > +#define __DECL_SIMD_coshf16 > +#define __DECL_SIMD_coshf32 > +#define __DECL_SIMD_coshf64 > +#define __DECL_SIMD_coshf128 > +#define __DECL_SIMD_coshf32x > +#define __DECL_SIMD_coshf64x > +#define __DECL_SIMD_coshf128x > #endif > diff --git a/math/bits/mathcalls.h b/math/bits/mathcalls.h > index 870778457f..60a314f69e 100644 > --- a/math/bits/mathcalls.h > +++ b/math/bits/mathcalls.h > @@ -68,7 +68,7 @@ __MATHCALL (tan,, (_Mdouble_ __x)); > /* Hyperbolic functions. */ > > /* Hyperbolic cosine of X. */ > -__MATHCALL (cosh,, (_Mdouble_ __x)); > +__MATHCALL_VEC (cosh,, (_Mdouble_ __x)); > /* Hyperbolic sine of X. */ > __MATHCALL (sinh,, (_Mdouble_ __x)); > /* Hyperbolic tangent of X. */ > diff --git a/sysdeps/unix/sysv/linux/x86_64/libmvec.abilist b/sysdeps/unix/sysv/linux/x86_64/libmvec.abilist > index b3c1f59593..4907680143 100644 > --- a/sysdeps/unix/sysv/linux/x86_64/libmvec.abilist > +++ b/sysdeps/unix/sysv/linux/x86_64/libmvec.abilist > @@ -49,48 +49,56 @@ GLIBC_2.22 _ZGVeN8vvv_sincos F > GLIBC_2.35 _ZGVbN2v_acos F > GLIBC_2.35 _ZGVbN2v_asin F > GLIBC_2.35 _ZGVbN2v_atan F > +GLIBC_2.35 _ZGVbN2v_cosh F > GLIBC_2.35 _ZGVbN2v_exp10 F > GLIBC_2.35 _ZGVbN2v_exp2 F > GLIBC_2.35 _ZGVbN2vv_hypot F > GLIBC_2.35 _ZGVbN4v_acosf F > GLIBC_2.35 _ZGVbN4v_asinf F > GLIBC_2.35 _ZGVbN4v_atanf F > +GLIBC_2.35 _ZGVbN4v_coshf F > GLIBC_2.35 _ZGVbN4v_exp10f F > GLIBC_2.35 _ZGVbN4v_exp2f F > GLIBC_2.35 _ZGVbN4vv_hypotf F > GLIBC_2.35 _ZGVcN4v_acos F > GLIBC_2.35 _ZGVcN4v_asin F > GLIBC_2.35 _ZGVcN4v_atan F > +GLIBC_2.35 _ZGVcN4v_cosh F > GLIBC_2.35 _ZGVcN4v_exp10 F > GLIBC_2.35 _ZGVcN4v_exp2 F > GLIBC_2.35 _ZGVcN4vv_hypot F > GLIBC_2.35 _ZGVcN8v_acosf F > GLIBC_2.35 _ZGVcN8v_asinf F > GLIBC_2.35 _ZGVcN8v_atanf F > +GLIBC_2.35 _ZGVcN8v_coshf F > GLIBC_2.35 _ZGVcN8v_exp10f F > GLIBC_2.35 _ZGVcN8v_exp2f F > GLIBC_2.35 _ZGVcN8vv_hypotf F > GLIBC_2.35 _ZGVdN4v_acos F > GLIBC_2.35 _ZGVdN4v_asin F > GLIBC_2.35 _ZGVdN4v_atan F > +GLIBC_2.35 _ZGVdN4v_cosh F > GLIBC_2.35 _ZGVdN4v_exp10 F > GLIBC_2.35 _ZGVdN4v_exp2 F > GLIBC_2.35 _ZGVdN4vv_hypot F > GLIBC_2.35 _ZGVdN8v_acosf F > GLIBC_2.35 _ZGVdN8v_asinf F > GLIBC_2.35 _ZGVdN8v_atanf F > +GLIBC_2.35 _ZGVdN8v_coshf F > GLIBC_2.35 _ZGVdN8v_exp10f F > GLIBC_2.35 _ZGVdN8v_exp2f F > GLIBC_2.35 _ZGVdN8vv_hypotf F > GLIBC_2.35 _ZGVeN16v_acosf F > GLIBC_2.35 _ZGVeN16v_asinf F > GLIBC_2.35 _ZGVeN16v_atanf F > +GLIBC_2.35 _ZGVeN16v_coshf F > GLIBC_2.35 _ZGVeN16v_exp10f F > GLIBC_2.35 _ZGVeN16v_exp2f F > GLIBC_2.35 _ZGVeN16vv_hypotf F > GLIBC_2.35 _ZGVeN8v_acos F > GLIBC_2.35 _ZGVeN8v_asin F > GLIBC_2.35 _ZGVeN8v_atan F > +GLIBC_2.35 _ZGVeN8v_cosh F > GLIBC_2.35 _ZGVeN8v_exp10 F > GLIBC_2.35 _ZGVeN8v_exp2 F > GLIBC_2.35 _ZGVeN8vv_hypot F > diff --git a/sysdeps/x86/fpu/bits/math-vector.h b/sysdeps/x86/fpu/bits/math-vector.h > index f3f9c2e092..708e81b3d0 100644 > --- a/sysdeps/x86/fpu/bits/math-vector.h > +++ b/sysdeps/x86/fpu/bits/math-vector.h > @@ -82,6 +82,10 @@ > # define __DECL_SIMD_exp10 __DECL_SIMD_x86_64 > # undef __DECL_SIMD_exp10f > # define __DECL_SIMD_exp10f __DECL_SIMD_x86_64 > +# undef __DECL_SIMD_cosh > +# define __DECL_SIMD_cosh __DECL_SIMD_x86_64 > +# undef __DECL_SIMD_coshf > +# define __DECL_SIMD_coshf __DECL_SIMD_x86_64 > > # endif > #endif > diff --git a/sysdeps/x86/fpu/finclude/math-vector-fortran.h b/sysdeps/x86/fpu/finclude/math-vector-fortran.h > index c033abbedc..81d0238ebf 100644 > --- a/sysdeps/x86/fpu/finclude/math-vector-fortran.h > +++ b/sysdeps/x86/fpu/finclude/math-vector-fortran.h > @@ -40,6 +40,8 @@ > !GCC$ builtin (exp2f) attributes simd (notinbranch) if('x86_64') > !GCC$ builtin (exp10) attributes simd (notinbranch) if('x86_64') > !GCC$ builtin (exp10f) attributes simd (notinbranch) if('x86_64') > +!GCC$ builtin (cosh) attributes simd (notinbranch) if('x86_64') > +!GCC$ builtin (coshf) attributes simd (notinbranch) if('x86_64') > > !GCC$ builtin (cos) attributes simd (notinbranch) if('x32') > !GCC$ builtin (cosf) attributes simd (notinbranch) if('x32') > @@ -65,3 +67,5 @@ > !GCC$ builtin (exp2f) attributes simd (notinbranch) if('x32') > !GCC$ builtin (exp10) attributes simd (notinbranch) if('x32') > !GCC$ builtin (exp10f) attributes simd (notinbranch) if('x32') > +!GCC$ builtin (cosh) attributes simd (notinbranch) if('x32') > +!GCC$ builtin (coshf) attributes simd (notinbranch) if('x32') > diff --git a/sysdeps/x86_64/fpu/Makeconfig b/sysdeps/x86_64/fpu/Makeconfig > index fd0a9da439..5bc2df134f 100644 > --- a/sysdeps/x86_64/fpu/Makeconfig > +++ b/sysdeps/x86_64/fpu/Makeconfig > @@ -26,6 +26,7 @@ libmvec-funcs = \ > asin \ > atan \ > cos \ > + cosh \ > exp \ > exp10 \ > exp2 \ > diff --git a/sysdeps/x86_64/fpu/Versions b/sysdeps/x86_64/fpu/Versions > index f29cfa4cbf..53346d16a2 100644 > --- a/sysdeps/x86_64/fpu/Versions > +++ b/sysdeps/x86_64/fpu/Versions > @@ -17,12 +17,14 @@ libmvec { > _ZGVbN2v_acos; _ZGVcN4v_acos; _ZGVdN4v_acos; _ZGVeN8v_acos; > _ZGVbN2v_asin; _ZGVcN4v_asin; _ZGVdN4v_asin; _ZGVeN8v_asin; > _ZGVbN2v_atan; _ZGVcN4v_atan; _ZGVdN4v_atan; _ZGVeN8v_atan; > + _ZGVbN2v_cosh; _ZGVcN4v_cosh; _ZGVdN4v_cosh; _ZGVeN8v_cosh; > _ZGVbN2v_exp10; _ZGVcN4v_exp10; _ZGVdN4v_exp10; _ZGVeN8v_exp10; > _ZGVbN2v_exp2; _ZGVcN4v_exp2; _ZGVdN4v_exp2; _ZGVeN8v_exp2; > _ZGVbN2vv_hypot; _ZGVcN4vv_hypot; _ZGVdN4vv_hypot; _ZGVeN8vv_hypot; > _ZGVbN4v_acosf; _ZGVcN8v_acosf; _ZGVdN8v_acosf; _ZGVeN16v_acosf; > _ZGVbN4v_asinf; _ZGVcN8v_asinf; _ZGVdN8v_asinf; _ZGVeN16v_asinf; > _ZGVbN4v_atanf; _ZGVcN8v_atanf; _ZGVdN8v_atanf; _ZGVeN16v_atanf; > + _ZGVbN4v_coshf; _ZGVcN8v_coshf; _ZGVdN8v_coshf; _ZGVeN16v_coshf; > _ZGVbN4v_exp10f; _ZGVcN8v_exp10f; _ZGVdN8v_exp10f; _ZGVeN16v_exp10f; > _ZGVbN4v_exp2f; _ZGVcN8v_exp2f; _ZGVdN8v_exp2f; _ZGVeN16v_exp2f; > _ZGVbN4vv_hypotf; _ZGVcN8vv_hypotf; _ZGVdN8vv_hypotf; _ZGVeN16vv_hypotf; > diff --git a/sysdeps/x86_64/fpu/libm-test-ulps b/sysdeps/x86_64/fpu/libm-test-ulps > index 45f2e4bb53..ac70f15208 100644 > --- a/sysdeps/x86_64/fpu/libm-test-ulps > +++ b/sysdeps/x86_64/fpu/libm-test-ulps > @@ -891,6 +891,26 @@ float: 2 > float128: 3 > ldouble: 3 > > +Function: "cosh_vlen16": > +float: 2 > + > +Function: "cosh_vlen2": > +double: 2 > + > +Function: "cosh_vlen4": > +double: 2 > +float: 2 > + > +Function: "cosh_vlen4_avx2": > +double: 2 > + > +Function: "cosh_vlen8": > +double: 2 > +float: 2 > + > +Function: "cosh_vlen8_avx2": > +float: 2 > + > Function: Real part of "cpow": > double: 2 > float: 5 > diff --git a/sysdeps/x86_64/fpu/multiarch/svml_d_cosh2_core-sse2.S b/sysdeps/x86_64/fpu/multiarch/svml_d_cosh2_core-sse2.S > new file mode 100644 > index 0000000000..bfe4e3d0f0 > --- /dev/null > +++ b/sysdeps/x86_64/fpu/multiarch/svml_d_cosh2_core-sse2.S > @@ -0,0 +1,20 @@ > +/* SSE2 version of vectorized cosh, vector length is 2. > + 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 > + . */ > + > +#define _ZGVbN2v_cosh _ZGVbN2v_cosh_sse2 > +#include "../svml_d_cosh2_core.S" > diff --git a/sysdeps/x86_64/fpu/multiarch/svml_d_cosh2_core.c b/sysdeps/x86_64/fpu/multiarch/svml_d_cosh2_core.c > new file mode 100644 > index 0000000000..99561fea47 > --- /dev/null > +++ b/sysdeps/x86_64/fpu/multiarch/svml_d_cosh2_core.c > @@ -0,0 +1,27 @@ > +/* Multiple versions of vectorized cosh, vector length is 2. > + 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 > + . */ > + > +#define SYMBOL_NAME _ZGVbN2v_cosh > +#include "ifunc-mathvec-sse4_1.h" > + > +libc_ifunc_redirected (REDIRECT_NAME, SYMBOL_NAME, IFUNC_SELECTOR ()); > + > +#ifdef SHARED > +__hidden_ver1 (_ZGVbN2v_cosh, __GI__ZGVbN2v_cosh, __redirect__ZGVbN2v_cosh) > + __attribute__ ((visibility ("hidden"))); > +#endif > diff --git a/sysdeps/x86_64/fpu/multiarch/svml_d_cosh2_core_sse4.S b/sysdeps/x86_64/fpu/multiarch/svml_d_cosh2_core_sse4.S > new file mode 100644 > index 0000000000..150bfae7e1 > --- /dev/null > +++ b/sysdeps/x86_64/fpu/multiarch/svml_d_cosh2_core_sse4.S > @@ -0,0 +1,396 @@ > +/* Function cosh vectorized with SSE4. > + 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/. */ > + > +/* > + * ALGORITHM DESCRIPTION: > + * > + * Compute cosh(x) as (exp(x)+exp(-x))/2, > + * where exp is calculated as > + * exp(M*ln2 + ln2*(j/2^k) + r) = 2^M * 2^(j/2^k) * exp(r) > + * > + * Special cases: > + * > + * cosh(NaN) = quiet NaN, and raise invalid exception > + * cosh(INF) = that INF > + * cosh(0) = 1 > + * cosh(x) overflows for big x and returns MAXLOG+log(2) > + * > + */ > + > +/* Offsets for data table __svml_dcosh_data_internal > + */ > +#define _dbT 0 > +#define _dbInvLn2 2064 > +#define _dbLn2hi 2080 > +#define _dbLn2lo 2096 > +#define _dbShifter 2112 > +#define _iIndexMask 2128 > +#define _dPC2 2144 > +#define _dPC3 2160 > +#define _dPC4 2176 > +#define _iMaxIndex 2192 > +#define _lExpMask 2208 > +#define _dSign 2224 > +#define _iDomainRange 2240 > + > +#include > + > + .text > + .section .text.sse4,"ax",@progbits > +ENTRY(_ZGVbN2v_cosh_sse4) > + subq $72, %rsp > + cfi_def_cfa_offset(80) > + movaps %xmm0, %xmm4 > + movups _dSign+__svml_dcosh_data_internal(%rip), %xmm2 > + lea _dbT+__svml_dcosh_data_internal(%rip), %r8 > + > +/* Abs argument */ > + movaps %xmm2, %xmm5 > + > +/* dXSign=0x001000000000 */ > + psrlq $11, %xmm2 > + > +/* > + * Load argument > + * dM = x*2^K/log(2) + RShifter > + */ > + movups _dbInvLn2+__svml_dcosh_data_internal(%rip), %xmm3 > + andnps %xmm4, %xmm5 > + mulpd %xmm5, %xmm3 > + movups _dbShifter+__svml_dcosh_data_internal(%rip), %xmm1 > + addpd %xmm1, %xmm3 > + > +/* > + * R > + * dN = dM - RShifter > + */ > + movaps %xmm3, %xmm15 > + subpd %xmm1, %xmm15 > + > +/* dR = dX - dN*Log2_hi/2^K */ > + movups _dbLn2hi+__svml_dcosh_data_internal(%rip), %xmm14 > + mulpd %xmm15, %xmm14 > + > +/* dR = (dX - dN*Log2_hi/2^K) - dN*Log2_lo/2^K */ > + movups _dbLn2lo+__svml_dcosh_data_internal(%rip), %xmm1 > + mulpd %xmm15, %xmm1 > + > +/* > + * Check for overflow\underflow > + * > + */ > + pshufd $221, %xmm5, %xmm7 > + subpd %xmm14, %xmm5 > + movq _iIndexMask+__svml_dcosh_data_internal(%rip), %xmm8 > + > +/* Index and lookup */ > + pshufd $136, %xmm3, %xmm9 > + > +/* > + * G1,G2,G3: dTdif,dTn * 2^N,2^(-N) > + * NB: copied from sinh_la - to be optimized!!!!! > + */ > + psllq $44, %xmm3 > + > +/* > + * trick > + * 256=-iIndex > + */ > + movq _iMaxIndex+__svml_dcosh_data_internal(%rip), %xmm12 > + pand %xmm8, %xmm9 > + subpd %xmm1, %xmm5 > + psubd %xmm9, %xmm12 > + > +/* iIndex*=3 */ > + movdqa %xmm9, %xmm10 > + > +/* iDomainRange*=3 */ > + pslld $3, %xmm12 > + pslld $3, %xmm10 > + movd %xmm12, %esi > + pshufd $1, %xmm12, %xmm13 > + movq _iDomainRange+__svml_dcosh_data_internal(%rip), %xmm6 > + movd %xmm13, %edi > + pcmpgtd %xmm6, %xmm7 > + movmskps %xmm7, %eax > + > +/* dR2 = dR^2 */ > + movaps %xmm5, %xmm7 > + > +/* lM now is an EXP(2^N) */ > + pand _lExpMask+__svml_dcosh_data_internal(%rip), %xmm3 > + pshufd $1, %xmm10, %xmm11 > + movslq %esi, %rsi > + mulpd %xmm5, %xmm7 > + movd %xmm10, %edx > + movsd (%r8,%rsi), %xmm6 > + movd %xmm11, %ecx > + movslq %edi, %rdi > + movslq %edx, %rdx > + movslq %ecx, %rcx > + movhpd (%r8,%rdi), %xmm6 > + > +/* */ > + psubq %xmm3, %xmm6 > + > +/* lX- = EXP(1/2) */ > + psubq %xmm2, %xmm6 > + > +/* > + * sinh(r) = r +r*r^2*a3 .... > + * dSinh_r = r^2*a3 > + */ > + movups _dPC3+__svml_dcosh_data_internal(%rip), %xmm2 > + mulpd %xmm7, %xmm2 > + > +/* dSinh_r = r + r*r^2*a3 */ > + mulpd %xmm5, %xmm2 > + movsd (%r8,%rdx), %xmm0 > + movhpd (%r8,%rcx), %xmm0 > + paddq %xmm3, %xmm0 > + addpd %xmm2, %xmm5 > + > +/* dTn = dTn*2^N - dTn*2^-N */ > + movaps %xmm0, %xmm3 > + subpd %xmm6, %xmm3 > + > +/* dTp = dTn*2^N + dTn*2^-N */ > + addpd %xmm6, %xmm0 > + mulpd %xmm5, %xmm3 > + > +/* poly(r) = dTp + dTn*sinh(dR)+dTp*dR2*(a2 +a4*dR2) */ > + movups _dPC4+__svml_dcosh_data_internal(%rip), %xmm5 > + mulpd %xmm7, %xmm5 > + addpd _dPC2+__svml_dcosh_data_internal(%rip), %xmm5 > + mulpd %xmm5, %xmm7 > + > +/* dTn*sinh(dR)+dTp*dR2*(a2 +a4*dR2) */ > + mulpd %xmm0, %xmm7 > + addpd %xmm7, %xmm3 > + > +/* _VRES1 = dTp + dTn*sinh(dR)+dTp*dR2*(a2 +a4*dR2) */ > + addpd %xmm3, %xmm0 > + andl $3, %eax > + > +/* Ret H */ > + > +/* Go to special inputs processing branch */ > + jne L(SPECIAL_VALUES_BRANCH) > + # LOE rbx rbp r12 r13 r14 r15 eax xmm0 xmm4 > + > +/* Restore registers > + * and exit the function > + */ > + > +L(EXIT): > + addq $72, %rsp > + cfi_def_cfa_offset(8) > + ret > + cfi_def_cfa_offset(80) > + > +/* Branch to process > + * special inputs > + */ > + > +L(SPECIAL_VALUES_BRANCH): > + movups %xmm4, 32(%rsp) > + movups %xmm0, 48(%rsp) > + # LOE rbx rbp r12 r13 r14 r15 eax xmm0 > + > + xorl %edx, %edx > + movq %r12, 16(%rsp) > + cfi_offset(12, -64) > + movl %edx, %r12d > + movq %r13, 8(%rsp) > + cfi_offset(13, -72) > + movl %eax, %r13d > + movq %r14, (%rsp) > + cfi_offset(14, -80) > + # LOE rbx rbp r15 r12d r13d > + > +/* Range mask > + * bits check > + */ > + > +L(RANGEMASK_CHECK): > + btl %r12d, %r13d > + > +/* Call scalar math function */ > + jc L(SCALAR_MATH_CALL) > + # LOE rbx rbp r15 r12d r13d > + > +/* Special inputs > + * processing loop > + */ > + > +L(SPECIAL_VALUES_LOOP): > + incl %r12d > + cmpl $2, %r12d > + > +/* Check bits in range mask */ > + jl L(RANGEMASK_CHECK) > + # LOE rbx rbp r15 r12d r13d > + > + movq 16(%rsp), %r12 > + cfi_restore(12) > + movq 8(%rsp), %r13 > + cfi_restore(13) > + movq (%rsp), %r14 > + cfi_restore(14) > + movups 48(%rsp), %xmm0 > + > +/* Go to exit */ > + jmp L(EXIT) > + cfi_offset(12, -64) > + cfi_offset(13, -72) > + cfi_offset(14, -80) > + # LOE rbx rbp r12 r13 r14 r15 xmm0 > + > +/* Scalar math fucntion call > + * to process special input > + */ > + > +L(SCALAR_MATH_CALL): > + movl %r12d, %r14d > + movsd 32(%rsp,%r14,8), %xmm0 > + call cosh@PLT > + # LOE rbx rbp r14 r15 r12d r13d xmm0 > + > + movsd %xmm0, 48(%rsp,%r14,8) > + > +/* Process special inputs in loop */ > + jmp L(SPECIAL_VALUES_LOOP) > + # LOE rbx rbp r15 r12d r13d > +END(_ZGVbN2v_cosh_sse4) > + > + .section .rodata, "a" > + .align 16 > + > +#ifdef __svml_dcosh_data_internal_typedef > +typedef unsigned int VUINT32; > +typedef struct > +{ > + __declspec(align(16)) VUINT32 _dbT[(1 + (1<<8))][2]; //dTpj ONLY! > + __declspec(align(16)) VUINT32 _dbInvLn2[2][2]; > + __declspec(align(16)) VUINT32 _dbLn2hi[2][2]; > + __declspec(align(16)) VUINT32 _dbLn2lo[2][2]; > + __declspec(align(16)) VUINT32 _dbShifter[2][2]; > + __declspec(align(16)) VUINT32 _iIndexMask[4][1]; //(1< + __declspec(align(16)) VUINT32 _dPC2[2][2]; > + __declspec(align(16)) VUINT32 _dPC3[2][2]; > + __declspec(align(16)) VUINT32 _dPC4[2][2]; > + __declspec(align(16)) VUINT32 _iMaxIndex[4][1]; //(1< + __declspec(align(16)) VUINT32 _lExpMask[2][2]; > + __declspec(align(16)) VUINT32 _dSign[2][2]; //0x8000000000000000 > + __declspec(align(16)) VUINT32 _iDomainRange[4][1]; > +} __svml_dcosh_data_internal; > +#endif > +__svml_dcosh_data_internal: > + /*== _dbT ==*/ > + .quad 0x3fe0000000000000, 0x3fe00b1afa5abcbf, 0x3fe0163da9fb3335, 0x3fe02168143b0281 > + .quad 0x3fe02c9a3e778061, 0x3fe037d42e11bbcc, 0x3fe04315e86e7f85, 0x3fe04e5f72f654b1 > + .quad 0x3fe059b0d3158574, 0x3fe0650a0e3c1f89, 0x3fe0706b29ddf6de, 0x3fe07bd42b72a836 > + .quad 0x3fe0874518759bc8, 0x3fe092bdf66607e0, 0x3fe09e3ecac6f383, 0x3fe0a9c79b1f3919 > + .quad 0x3fe0b5586cf9890f, 0x3fe0c0f145e46c85, 0x3fe0cc922b7247f7, 0x3fe0d83b23395dec > + .quad 0x3fe0e3ec32d3d1a2, 0x3fe0efa55fdfa9c5, 0x3fe0fb66affed31b, 0x3fe1073028d7233e > + .quad 0x3fe11301d0125b51, 0x3fe11edbab5e2ab6, 0x3fe12abdc06c31cc, 0x3fe136a814f204ab > + .quad 0x3fe1429aaea92de0, 0x3fe14e95934f312e, 0x3fe15a98c8a58e51, 0x3fe166a45471c3c2 > + .quad 0x3fe172b83c7d517b, 0x3fe17ed48695bbc0, 0x3fe18af9388c8dea, 0x3fe1972658375d2f > + .quad 0x3fe1a35beb6fcb75, 0x3fe1af99f8138a1c, 0x3fe1bbe084045cd4, 0x3fe1c82f95281c6b > + .quad 0x3fe1d4873168b9aa, 0x3fe1e0e75eb44027, 0x3fe1ed5022fcd91d, 0x3fe1f9c18438ce4d > + .quad 0x3fe2063b88628cd6, 0x3fe212be3578a819, 0x3fe21f49917ddc96, 0x3fe22bdda27912d1 > + .quad 0x3fe2387a6e756238, 0x3fe2451ffb82140a, 0x3fe251ce4fb2a63f, 0x3fe25e85711ece75 > + .quad 0x3fe26b4565e27cdd, 0x3fe2780e341ddf29, 0x3fe284dfe1f56381, 0x3fe291ba7591bb70 > + .quad 0x3fe29e9df51fdee1, 0x3fe2ab8a66d10f13, 0x3fe2b87fd0dad990, 0x3fe2c57e39771b2f > + .quad 0x3fe2d285a6e4030b, 0x3fe2df961f641589, 0x3fe2ecafa93e2f56, 0x3fe2f9d24abd886b > + .quad 0x3fe306fe0a31b715, 0x3fe31432edeeb2fd, 0x3fe32170fc4cd831, 0x3fe32eb83ba8ea32 > + .quad 0x3fe33c08b26416ff, 0x3fe3496266e3fa2d, 0x3fe356c55f929ff1, 0x3fe36431a2de883b > + .quad 0x3fe371a7373aa9cb, 0x3fe37f26231e754a, 0x3fe38cae6d05d866, 0x3fe39a401b7140ef > + .quad 0x3fe3a7db34e59ff7, 0x3fe3b57fbfec6cf4, 0x3fe3c32dc313a8e5, 0x3fe3d0e544ede173 > + .quad 0x3fe3dea64c123422, 0x3fe3ec70df1c5175, 0x3fe3fa4504ac801c, 0x3fe40822c367a024 > + .quad 0x3fe4160a21f72e2a, 0x3fe423fb2709468a, 0x3fe431f5d950a897, 0x3fe43ffa3f84b9d4 > + .quad 0x3fe44e086061892d, 0x3fe45c2042a7d232, 0x3fe46a41ed1d0057, 0x3fe4786d668b3237 > + .quad 0x3fe486a2b5c13cd0, 0x3fe494e1e192aed2, 0x3fe4a32af0d7d3de, 0x3fe4b17dea6db7d7 > + .quad 0x3fe4bfdad5362a27, 0x3fe4ce41b817c114, 0x3fe4dcb299fddd0d, 0x3fe4eb2d81d8abff > + .quad 0x3fe4f9b2769d2ca7, 0x3fe508417f4531ee, 0x3fe516daa2cf6642, 0x3fe5257de83f4eef > + .quad 0x3fe5342b569d4f82, 0x3fe542e2f4f6ad27, 0x3fe551a4ca5d920f, 0x3fe56070dde910d2 > + .quad 0x3fe56f4736b527da, 0x3fe57e27dbe2c4cf, 0x3fe58d12d497c7fd, 0x3fe59c0827ff07cc > + .quad 0x3fe5ab07dd485429, 0x3fe5ba11fba87a03, 0x3fe5c9268a5946b7, 0x3fe5d84590998b93 > + .quad 0x3fe5e76f15ad2148, 0x3fe5f6a320dceb71, 0x3fe605e1b976dc09, 0x3fe6152ae6cdf6f4 > + .quad 0x3fe6247eb03a5585, 0x3fe633dd1d1929fd, 0x3fe6434634ccc320, 0x3fe652b9febc8fb7 > + .quad 0x3fe6623882552225, 0x3fe671c1c70833f6, 0x3fe68155d44ca973, 0x3fe690f4b19e9538 > + .quad 0x3fe6a09e667f3bcd, 0x3fe6b052fa75173e, 0x3fe6c012750bdabf, 0x3fe6cfdcddd47645 > + .quad 0x3fe6dfb23c651a2f, 0x3fe6ef9298593ae5, 0x3fe6ff7df9519484, 0x3fe70f7466f42e87 > + .quad 0x3fe71f75e8ec5f74, 0x3fe72f8286ead08a, 0x3fe73f9a48a58174, 0x3fe74fbd35d7cbfd > + .quad 0x3fe75feb564267c9, 0x3fe77024b1ab6e09, 0x3fe780694fde5d3f, 0x3fe790b938ac1cf6 > + .quad 0x3fe7a11473eb0187, 0x3fe7b17b0976cfdb, 0x3fe7c1ed0130c132, 0x3fe7d26a62ff86f0 > + .quad 0x3fe7e2f336cf4e62, 0x3fe7f3878491c491, 0x3fe80427543e1a12, 0x3fe814d2add106d9 > + .quad 0x3fe82589994cce13, 0x3fe8364c1eb941f7, 0x3fe8471a4623c7ad, 0x3fe857f4179f5b21 > + .quad 0x3fe868d99b4492ed, 0x3fe879cad931a436, 0x3fe88ac7d98a6699, 0x3fe89bd0a478580f > + .quad 0x3fe8ace5422aa0db, 0x3fe8be05bad61778, 0x3fe8cf3216b5448c, 0x3fe8e06a5e0866d9 > + .quad 0x3fe8f1ae99157736, 0x3fe902fed0282c8a, 0x3fe9145b0b91ffc6, 0x3fe925c353aa2fe2 > + .quad 0x3fe93737b0cdc5e5, 0x3fe948b82b5f98e5, 0x3fe95a44cbc8520f, 0x3fe96bdd9a7670b3 > + .quad 0x3fe97d829fde4e50, 0x3fe98f33e47a22a2, 0x3fe9a0f170ca07ba, 0x3fe9b2bb4d53fe0d > + .quad 0x3fe9c49182a3f090, 0x3fe9d674194bb8d5, 0x3fe9e86319e32323, 0x3fe9fa5e8d07f29e > + .quad 0x3fea0c667b5de565, 0x3fea1e7aed8eb8bb, 0x3fea309bec4a2d33, 0x3fea42c980460ad8 > + .quad 0x3fea5503b23e255d, 0x3fea674a8af46052, 0x3fea799e1330b358, 0x3fea8bfe53c12e59 > + .quad 0x3fea9e6b5579fdbf, 0x3feab0e521356eba, 0x3feac36bbfd3f37a, 0x3fead5ff3a3c2774 > + .quad 0x3feae89f995ad3ad, 0x3feafb4ce622f2ff, 0x3feb0e07298db666, 0x3feb20ce6c9a8952 > + .quad 0x3feb33a2b84f15fb, 0x3feb468415b749b1, 0x3feb59728de5593a, 0x3feb6c6e29f1c52a > + .quad 0x3feb7f76f2fb5e47, 0x3feb928cf22749e4, 0x3feba5b030a1064a, 0x3febb8e0b79a6f1f > + .quad 0x3febcc1e904bc1d2, 0x3febdf69c3f3a207, 0x3febf2c25bd71e09, 0x3fec06286141b33d > + .quad 0x3fec199bdd85529c, 0x3fec2d1cd9fa652c, 0x3fec40ab5fffd07a, 0x3fec544778fafb22 > + .quad 0x3fec67f12e57d14b, 0x3fec7ba88988c933, 0x3fec8f6d9406e7b5, 0x3feca3405751c4db > + .quad 0x3fecb720dcef9069, 0x3feccb0f2e6d1675, 0x3fecdf0b555dc3fa, 0x3fecf3155b5bab74 > + .quad 0x3fed072d4a07897c, 0x3fed1b532b08c968, 0x3fed2f87080d89f2, 0x3fed43c8eacaa1d6 > + .quad 0x3fed5818dcfba487, 0x3fed6c76e862e6d3, 0x3fed80e316c98398, 0x3fed955d71ff6075 > + .quad 0x3feda9e603db3285, 0x3fedbe7cd63a8315, 0x3fedd321f301b460, 0x3fede7d5641c0658 > + .quad 0x3fedfc97337b9b5f, 0x3fee11676b197d17, 0x3fee264614f5a129, 0x3fee3b333b16ee12 > + .quad 0x3fee502ee78b3ff6, 0x3fee653924676d76, 0x3fee7a51fbc74c83, 0x3fee8f7977cdb740 > + .quad 0x3feea4afa2a490da, 0x3feeb9f4867cca6e, 0x3feecf482d8e67f1, 0x3feee4aaa2188510 > + .quad 0x3feefa1bee615a27, 0x3fef0f9c1cb6412a, 0x3fef252b376bba97, 0x3fef3ac948dd7274 > + .quad 0x3fef50765b6e4540, 0x3fef6632798844f8, 0x3fef7bfdad9cbe14, 0x3fef91d802243c89 > + .quad 0x3fefa7c1819e90d8, 0x3fefbdba3692d514, 0x3fefd3c22b8f71f1, 0x3fefe9d96b2a23d9 > + .quad 0x3ff0000000000000 > + .align 16 > + .quad 0x3ff71547652b82fe, 0x3ff71547652b82fe /* _dbInvLn2 = 1/log(2) */ > + .align 16 > + .quad 0x3FE62E42FEFC0000, 0x3FE62E42FEFC0000 /* _dbLn2hi = log(2) hi*/ > + .align 16 > + .quad 0xBDAC610CA86C3899, 0xBDAC610CA86C3899 /* _dbLn2lo = log(2) lo*/ > + .align 16 > + .quad 0x42B8000000000000, 0x42B8000000000000 /* _dbShifter */ > + .align 16 > + .long 0x000000FF, 0x000000FF, 0x000000FF, 0x000000FF /* _iIndexMask */ > + .align 16 > + .quad 0x3FDFFFFFFFFFFDBD, 0x3FDFFFFFFFFFFDBD /* _dPC2 */ > + .align 16 > + .quad 0x3FC5555570813E14, 0x3FC5555570813E14 /* _dPC3 */ > + .align 16 > + .quad 0x3FA55555CF16D299, 0x3FA55555CF16D299 /* _dPC4 */ > + .align 16 > + .long 0x00000100, 0x00000100, 0x00000100, 0x00000100 /* _iMaxIndex */ > + .align 16 > + .quad 0x7ff0000000000000, 0x7ff0000000000000 /* _lExpMask */ > + .align 16 > + .quad 0x8000000000000000, 0x8000000000000000 /* _dSign*/ > + .align 16 > + .long 0x40861d99, 0x40861d99, 0x40861d99, 0x40861d99 /* _iDomainRange 0x40861d9ac12a3e85 =(1021*2^K-0.5)*log(2)/2^K -needed for quick exp*/ > + .align 16 > + .type __svml_dcosh_data_internal,@object > + .size __svml_dcosh_data_internal,.-__svml_dcosh_data_internal > diff --git a/sysdeps/x86_64/fpu/multiarch/svml_d_cosh4_core-sse.S b/sysdeps/x86_64/fpu/multiarch/svml_d_cosh4_core-sse.S > new file mode 100644 > index 0000000000..4410d34583 > --- /dev/null > +++ b/sysdeps/x86_64/fpu/multiarch/svml_d_cosh4_core-sse.S > @@ -0,0 +1,20 @@ > +/* SSE version of vectorized cosh, vector length is 4. > + 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 > + . */ > + > +#define _ZGVdN4v_cosh _ZGVdN4v_cosh_sse_wrapper > +#include "../svml_d_cosh4_core.S" > diff --git a/sysdeps/x86_64/fpu/multiarch/svml_d_cosh4_core.c b/sysdeps/x86_64/fpu/multiarch/svml_d_cosh4_core.c > new file mode 100644 > index 0000000000..c4f59206a9 > --- /dev/null > +++ b/sysdeps/x86_64/fpu/multiarch/svml_d_cosh4_core.c > @@ -0,0 +1,27 @@ > +/* Multiple versions of vectorized cosh, vector length is 4. > + 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 > + . */ > + > +#define SYMBOL_NAME _ZGVdN4v_cosh > +#include "ifunc-mathvec-avx2.h" > + > +libc_ifunc_redirected (REDIRECT_NAME, SYMBOL_NAME, IFUNC_SELECTOR ()); > + > +#ifdef SHARED > +__hidden_ver1 (_ZGVdN4v_cosh, __GI__ZGVdN4v_cosh, __redirect__ZGVdN4v_cosh) > + __attribute__ ((visibility ("hidden"))); > +#endif > diff --git a/sysdeps/x86_64/fpu/multiarch/svml_d_cosh4_core_avx2.S b/sysdeps/x86_64/fpu/multiarch/svml_d_cosh4_core_avx2.S > new file mode 100644 > index 0000000000..2d86a02923 > --- /dev/null > +++ b/sysdeps/x86_64/fpu/multiarch/svml_d_cosh4_core_avx2.S > @@ -0,0 +1,412 @@ > +/* Function cosh vectorized with AVX2. > + 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/. */ > + > +/* > + * ALGORITHM DESCRIPTION: > + * > + * Compute cosh(x) as (exp(x)+exp(-x))/2, > + * where exp is calculated as > + * exp(M*ln2 + ln2*(j/2^k) + r) = 2^M * 2^(j/2^k) * exp(r) > + * > + * Special cases: > + * > + * cosh(NaN) = quiet NaN, and raise invalid exception > + * cosh(INF) = that INF > + * cosh(0) = 1 > + * cosh(x) overflows for big x and returns MAXLOG+log(2) > + * > + */ > + > +/* Offsets for data table __svml_dcosh_data_internal > + */ > +#define _dbT 0 > +#define _dbInvLn2 2080 > +#define _dbLn2hi 2112 > +#define _dbLn2lo 2144 > +#define _dbShifter 2176 > +#define _iIndexMask 2208 > +#define _dPC2 2240 > +#define _dPC3 2272 > +#define _dPC4 2304 > +#define _iMaxIndex 2336 > +#define _lExpMask 2368 > +#define _dSign 2400 > +#define _iDomainRange 2432 > + > +#include > + > + .text > + .section .text.avx2,"ax",@progbits > +ENTRY(_ZGVdN4v_cosh_avx2) > + pushq %rbp > + cfi_def_cfa_offset(16) > + movq %rsp, %rbp > + cfi_def_cfa(6, 16) > + cfi_offset(6, -16) > + andq $-32, %rsp > + subq $96, %rsp > + lea _dbT+__svml_dcosh_data_internal(%rip), %rax > + vmovupd _dSign+__svml_dcosh_data_internal(%rip), %ymm8 > + vmovupd _dbShifter+__svml_dcosh_data_internal(%rip), %ymm6 > + > +/* > + * Load argument > + * dM = x*2^K/log(2) + RShifter > + */ > + vmovupd _dbInvLn2+__svml_dcosh_data_internal(%rip), %ymm3 > + > +/* > + * trick > + * 256=-iIndex > + */ > + vmovups _iMaxIndex+__svml_dcosh_data_internal(%rip), %xmm14 > + > +/* dXSign=0x001000000000 */ > + vpsrlq $11, %ymm8, %ymm5 > + vmovapd %ymm0, %ymm7 > + > +/* Abs argument */ > + vandnpd %ymm7, %ymm8, %ymm4 > + vfmadd213pd %ymm6, %ymm4, %ymm3 > + > +/* Index and lookup */ > + vextractf128 $1, %ymm3, %xmm12 > + vshufps $136, %xmm12, %xmm3, %xmm13 > + vpand _iIndexMask+__svml_dcosh_data_internal(%rip), %xmm13, %xmm15 > + vpsubd %xmm15, %xmm14, %xmm0 > + > +/* iDomainRange*=3 */ > + vpslld $3, %xmm0, %xmm2 > + vmovd %xmm2, %r9d > + vpextrd $2, %xmm2, %r11d > + movslq %r9d, %r9 > + vpextrd $1, %xmm2, %r10d > + movslq %r11d, %r11 > + movslq %r10d, %r10 > + vmovsd (%rax,%r9), %xmm12 > + > +/* > + * Check for overflow\underflow > + * > + */ > + vextractf128 $1, %ymm4, %xmm9 > + vmovsd (%rax,%r11), %xmm14 > + vmovhpd (%rax,%r10), %xmm12, %xmm13 > + vshufps $221, %xmm9, %xmm4, %xmm10 > + > +/* iIndex*=3 */ > + vpslld $3, %xmm15, %xmm9 > + > +/* > + * R > + * dN = dM - RShifter > + */ > + vsubpd %ymm6, %ymm3, %ymm15 > + vmovd %xmm9, %ecx > + vpcmpgtd _iDomainRange+__svml_dcosh_data_internal(%rip), %xmm10, %xmm11 > + vmovupd _dbLn2hi+__svml_dcosh_data_internal(%rip), %ymm6 > + > +/* > + * G1,G2,G3: dTdif,dTn * 2^N,2^(-N) > + * NB: copied from sinh_la - to be optimized!!!!! > + */ > + vpsllq $44, %ymm3, %ymm3 > + vmovmskps %xmm11, %edx > + > +/* dR = dX - dN*Log2_hi/2^K */ > + vfnmadd231pd %ymm6, %ymm15, %ymm4 > + > +/* lM now is an EXP(2^N) */ > + vpand _lExpMask+__svml_dcosh_data_internal(%rip), %ymm3, %ymm3 > + > +/* dR = (dX - dN*Log2_hi/2^K) - dN*Log2_lo/2^K */ > + vfnmadd231pd _dbLn2lo+__svml_dcosh_data_internal(%rip), %ymm15, %ymm4 > + movslq %ecx, %rcx > + vpextrd $2, %xmm9, %edi > + vpextrd $1, %xmm9, %esi > + movslq %edi, %rdi > + vmovsd (%rax,%rcx), %xmm1 > + vpextrd $3, %xmm9, %r8d > + vpextrd $3, %xmm2, %ecx > + movslq %esi, %rsi > + movslq %r8d, %r8 > + movslq %ecx, %rcx > + > +/* dR2 = dR^2 */ > + vmulpd %ymm4, %ymm4, %ymm0 > + vmovsd (%rax,%rdi), %xmm10 > + vmovhpd (%rax,%rsi), %xmm1, %xmm8 > + vmovhpd (%rax,%r8), %xmm10, %xmm11 > + vmovhpd (%rax,%rcx), %xmm14, %xmm2 > + vinsertf128 $1, %xmm11, %ymm8, %ymm1 > + vinsertf128 $1, %xmm2, %ymm13, %ymm2 > + vpaddq %ymm3, %ymm1, %ymm6 > + > +/* */ > + vpsubq %ymm3, %ymm2, %ymm1 > + > +/* > + * sinh(r) = r +r*r^2*a3 .... > + * dSinh_r = r^2*a3 > + */ > + vmulpd _dPC3+__svml_dcosh_data_internal(%rip), %ymm0, %ymm2 > + > +/* lX- = EXP(1/2) */ > + vpsubq %ymm5, %ymm1, %ymm5 > + > +/* dSinh_r = r + r*r^2*a3 */ > + vfmadd213pd %ymm4, %ymm4, %ymm2 > + > +/* poly(r) = dTp + dTn*sinh(dR)+dTp*dR2*(a2 +a4*dR2) */ > + vmovupd _dPC4+__svml_dcosh_data_internal(%rip), %ymm4 > + > +/* dTn = dTn*2^N - dTn*2^-N */ > + vsubpd %ymm5, %ymm6, %ymm1 > + > +/* dTp = dTn*2^N + dTn*2^-N */ > + vaddpd %ymm5, %ymm6, %ymm3 > + vfmadd213pd _dPC2+__svml_dcosh_data_internal(%rip), %ymm0, %ymm4 > + vmulpd %ymm2, %ymm1, %ymm1 > + vmulpd %ymm4, %ymm0, %ymm0 > + > +/* dTn*sinh(dR)+dTp*dR2*(a2 +a4*dR2) */ > + vfmadd213pd %ymm1, %ymm3, %ymm0 > + > +/* _VRES1 = dTp + dTn*sinh(dR)+dTp*dR2*(a2 +a4*dR2) */ > + vaddpd %ymm0, %ymm3, %ymm0 > + > +/* Ret H */ > + testl %edx, %edx > + > +/* Go to special inputs processing branch */ > + jne L(SPECIAL_VALUES_BRANCH) > + # LOE rbx r12 r13 r14 r15 edx ymm0 ymm7 > + > +/* Restore registers > + * and exit the function > + */ > + > +L(EXIT): > + movq %rbp, %rsp > + popq %rbp > + cfi_def_cfa(7, 8) > + cfi_restore(6) > + ret > + cfi_def_cfa(6, 16) > + cfi_offset(6, -16) > + > +/* Branch to process > + * special inputs > + */ > + > +L(SPECIAL_VALUES_BRANCH): > + vmovupd %ymm7, 32(%rsp) > + vmovupd %ymm0, 64(%rsp) > + # LOE rbx r12 r13 r14 r15 edx ymm0 > + > + xorl %eax, %eax > + # LOE rbx r12 r13 r14 r15 eax edx > + > + vzeroupper > + movq %r12, 16(%rsp) > + /* DW_CFA_expression: r12 (r12) (DW_OP_lit8; DW_OP_minus; DW_OP_const4s: -32; DW_OP_and; DW_OP_const4s: -80; DW_OP_plus) */ > + .cfi_escape 0x10, 0x0c, 0x0e, 0x38, 0x1c, 0x0d, 0xe0, 0xff, 0xff, 0xff, 0x1a, 0x0d, 0xb0, 0xff, 0xff, 0xff, 0x22 > + movl %eax, %r12d > + movq %r13, 8(%rsp) > + /* DW_CFA_expression: r13 (r13) (DW_OP_lit8; DW_OP_minus; DW_OP_const4s: -32; DW_OP_and; DW_OP_const4s: -88; DW_OP_plus) */ > + .cfi_escape 0x10, 0x0d, 0x0e, 0x38, 0x1c, 0x0d, 0xe0, 0xff, 0xff, 0xff, 0x1a, 0x0d, 0xa8, 0xff, 0xff, 0xff, 0x22 > + movl %edx, %r13d > + movq %r14, (%rsp) > + /* DW_CFA_expression: r14 (r14) (DW_OP_lit8; DW_OP_minus; DW_OP_const4s: -32; DW_OP_and; DW_OP_const4s: -96; DW_OP_plus) */ > + .cfi_escape 0x10, 0x0e, 0x0e, 0x38, 0x1c, 0x0d, 0xe0, 0xff, 0xff, 0xff, 0x1a, 0x0d, 0xa0, 0xff, 0xff, 0xff, 0x22 > + # LOE rbx r15 r12d r13d > + > +/* Range mask > + * bits check > + */ > + > +L(RANGEMASK_CHECK): > + btl %r12d, %r13d > + > +/* Call scalar math function */ > + jc L(SCALAR_MATH_CALL) > + # LOE rbx r15 r12d r13d > + > +/* Special inputs > + * processing loop > + */ > + > +L(SPECIAL_VALUES_LOOP): > + incl %r12d > + cmpl $4, %r12d > + > +/* Check bits in range mask */ > + jl L(RANGEMASK_CHECK) > + # LOE rbx r15 r12d r13d > + > + movq 16(%rsp), %r12 > + cfi_restore(12) > + movq 8(%rsp), %r13 > + cfi_restore(13) > + movq (%rsp), %r14 > + cfi_restore(14) > + vmovupd 64(%rsp), %ymm0 > + > +/* Go to exit */ > + jmp L(EXIT) > + /* DW_CFA_expression: r12 (r12) (DW_OP_lit8; DW_OP_minus; DW_OP_const4s: -32; DW_OP_and; DW_OP_const4s: -80; DW_OP_plus) */ > + .cfi_escape 0x10, 0x0c, 0x0e, 0x38, 0x1c, 0x0d, 0xe0, 0xff, 0xff, 0xff, 0x1a, 0x0d, 0xb0, 0xff, 0xff, 0xff, 0x22 > + /* DW_CFA_expression: r13 (r13) (DW_OP_lit8; DW_OP_minus; DW_OP_const4s: -32; DW_OP_and; DW_OP_const4s: -88; DW_OP_plus) */ > + .cfi_escape 0x10, 0x0d, 0x0e, 0x38, 0x1c, 0x0d, 0xe0, 0xff, 0xff, 0xff, 0x1a, 0x0d, 0xa8, 0xff, 0xff, 0xff, 0x22 > + /* DW_CFA_expression: r14 (r14) (DW_OP_lit8; DW_OP_minus; DW_OP_const4s: -32; DW_OP_and; DW_OP_const4s: -96; DW_OP_plus) */ > + .cfi_escape 0x10, 0x0e, 0x0e, 0x38, 0x1c, 0x0d, 0xe0, 0xff, 0xff, 0xff, 0x1a, 0x0d, 0xa0, 0xff, 0xff, 0xff, 0x22 > + # LOE rbx r12 r13 r14 r15 ymm0 > + > +/* Scalar math fucntion call > + * to process special input > + */ > + > +L(SCALAR_MATH_CALL): > + movl %r12d, %r14d > + movsd 32(%rsp,%r14,8), %xmm0 > + call cosh@PLT > + # LOE rbx r14 r15 r12d r13d xmm0 > + > + movsd %xmm0, 64(%rsp,%r14,8) > + > +/* Process special inputs in loop */ > + jmp L(SPECIAL_VALUES_LOOP) > + # LOE rbx r15 r12d r13d > +END(_ZGVdN4v_cosh_avx2) > + > + .section .rodata, "a" > + .align 32 > + > +#ifdef __svml_dcosh_data_internal_typedef > +typedef unsigned int VUINT32; > +typedef struct > +{ > + __declspec(align(32)) VUINT32 _dbT[(1 + (1<<8))][2]; //dTpj ONLY! > + __declspec(align(32)) VUINT32 _dbInvLn2[4][2]; > + __declspec(align(32)) VUINT32 _dbLn2hi[4][2]; > + __declspec(align(32)) VUINT32 _dbLn2lo[4][2]; > + __declspec(align(32)) VUINT32 _dbShifter[4][2]; > + __declspec(align(32)) VUINT32 _iIndexMask[8][1]; //(1< + __declspec(align(32)) VUINT32 _dPC2[4][2]; > + __declspec(align(32)) VUINT32 _dPC3[4][2]; > + __declspec(align(32)) VUINT32 _dPC4[4][2]; > + __declspec(align(32)) VUINT32 _iMaxIndex[8][1]; //(1< + __declspec(align(32)) VUINT32 _lExpMask[4][2]; > + __declspec(align(32)) VUINT32 _dSign[4][2]; //0x8000000000000000 > + __declspec(align(32)) VUINT32 _iDomainRange[8][1]; > +} __svml_dcosh_data_internal; > +#endif > +__svml_dcosh_data_internal: > + /*== _dbT ==*/ > + .quad 0x3fe0000000000000, 0x3fe00b1afa5abcbf, 0x3fe0163da9fb3335, 0x3fe02168143b0281 > + .quad 0x3fe02c9a3e778061, 0x3fe037d42e11bbcc, 0x3fe04315e86e7f85, 0x3fe04e5f72f654b1 > + .quad 0x3fe059b0d3158574, 0x3fe0650a0e3c1f89, 0x3fe0706b29ddf6de, 0x3fe07bd42b72a836 > + .quad 0x3fe0874518759bc8, 0x3fe092bdf66607e0, 0x3fe09e3ecac6f383, 0x3fe0a9c79b1f3919 > + .quad 0x3fe0b5586cf9890f, 0x3fe0c0f145e46c85, 0x3fe0cc922b7247f7, 0x3fe0d83b23395dec > + .quad 0x3fe0e3ec32d3d1a2, 0x3fe0efa55fdfa9c5, 0x3fe0fb66affed31b, 0x3fe1073028d7233e > + .quad 0x3fe11301d0125b51, 0x3fe11edbab5e2ab6, 0x3fe12abdc06c31cc, 0x3fe136a814f204ab > + .quad 0x3fe1429aaea92de0, 0x3fe14e95934f312e, 0x3fe15a98c8a58e51, 0x3fe166a45471c3c2 > + .quad 0x3fe172b83c7d517b, 0x3fe17ed48695bbc0, 0x3fe18af9388c8dea, 0x3fe1972658375d2f > + .quad 0x3fe1a35beb6fcb75, 0x3fe1af99f8138a1c, 0x3fe1bbe084045cd4, 0x3fe1c82f95281c6b > + .quad 0x3fe1d4873168b9aa, 0x3fe1e0e75eb44027, 0x3fe1ed5022fcd91d, 0x3fe1f9c18438ce4d > + .quad 0x3fe2063b88628cd6, 0x3fe212be3578a819, 0x3fe21f49917ddc96, 0x3fe22bdda27912d1 > + .quad 0x3fe2387a6e756238, 0x3fe2451ffb82140a, 0x3fe251ce4fb2a63f, 0x3fe25e85711ece75 > + .quad 0x3fe26b4565e27cdd, 0x3fe2780e341ddf29, 0x3fe284dfe1f56381, 0x3fe291ba7591bb70 > + .quad 0x3fe29e9df51fdee1, 0x3fe2ab8a66d10f13, 0x3fe2b87fd0dad990, 0x3fe2c57e39771b2f > + .quad 0x3fe2d285a6e4030b, 0x3fe2df961f641589, 0x3fe2ecafa93e2f56, 0x3fe2f9d24abd886b > + .quad 0x3fe306fe0a31b715, 0x3fe31432edeeb2fd, 0x3fe32170fc4cd831, 0x3fe32eb83ba8ea32 > + .quad 0x3fe33c08b26416ff, 0x3fe3496266e3fa2d, 0x3fe356c55f929ff1, 0x3fe36431a2de883b > + .quad 0x3fe371a7373aa9cb, 0x3fe37f26231e754a, 0x3fe38cae6d05d866, 0x3fe39a401b7140ef > + .quad 0x3fe3a7db34e59ff7, 0x3fe3b57fbfec6cf4, 0x3fe3c32dc313a8e5, 0x3fe3d0e544ede173 > + .quad 0x3fe3dea64c123422, 0x3fe3ec70df1c5175, 0x3fe3fa4504ac801c, 0x3fe40822c367a024 > + .quad 0x3fe4160a21f72e2a, 0x3fe423fb2709468a, 0x3fe431f5d950a897, 0x3fe43ffa3f84b9d4 > + .quad 0x3fe44e086061892d, 0x3fe45c2042a7d232, 0x3fe46a41ed1d0057, 0x3fe4786d668b3237 > + .quad 0x3fe486a2b5c13cd0, 0x3fe494e1e192aed2, 0x3fe4a32af0d7d3de, 0x3fe4b17dea6db7d7 > + .quad 0x3fe4bfdad5362a27, 0x3fe4ce41b817c114, 0x3fe4dcb299fddd0d, 0x3fe4eb2d81d8abff > + .quad 0x3fe4f9b2769d2ca7, 0x3fe508417f4531ee, 0x3fe516daa2cf6642, 0x3fe5257de83f4eef > + .quad 0x3fe5342b569d4f82, 0x3fe542e2f4f6ad27, 0x3fe551a4ca5d920f, 0x3fe56070dde910d2 > + .quad 0x3fe56f4736b527da, 0x3fe57e27dbe2c4cf, 0x3fe58d12d497c7fd, 0x3fe59c0827ff07cc > + .quad 0x3fe5ab07dd485429, 0x3fe5ba11fba87a03, 0x3fe5c9268a5946b7, 0x3fe5d84590998b93 > + .quad 0x3fe5e76f15ad2148, 0x3fe5f6a320dceb71, 0x3fe605e1b976dc09, 0x3fe6152ae6cdf6f4 > + .quad 0x3fe6247eb03a5585, 0x3fe633dd1d1929fd, 0x3fe6434634ccc320, 0x3fe652b9febc8fb7 > + .quad 0x3fe6623882552225, 0x3fe671c1c70833f6, 0x3fe68155d44ca973, 0x3fe690f4b19e9538 > + .quad 0x3fe6a09e667f3bcd, 0x3fe6b052fa75173e, 0x3fe6c012750bdabf, 0x3fe6cfdcddd47645 > + .quad 0x3fe6dfb23c651a2f, 0x3fe6ef9298593ae5, 0x3fe6ff7df9519484, 0x3fe70f7466f42e87 > + .quad 0x3fe71f75e8ec5f74, 0x3fe72f8286ead08a, 0x3fe73f9a48a58174, 0x3fe74fbd35d7cbfd > + .quad 0x3fe75feb564267c9, 0x3fe77024b1ab6e09, 0x3fe780694fde5d3f, 0x3fe790b938ac1cf6 > + .quad 0x3fe7a11473eb0187, 0x3fe7b17b0976cfdb, 0x3fe7c1ed0130c132, 0x3fe7d26a62ff86f0 > + .quad 0x3fe7e2f336cf4e62, 0x3fe7f3878491c491, 0x3fe80427543e1a12, 0x3fe814d2add106d9 > + .quad 0x3fe82589994cce13, 0x3fe8364c1eb941f7, 0x3fe8471a4623c7ad, 0x3fe857f4179f5b21 > + .quad 0x3fe868d99b4492ed, 0x3fe879cad931a436, 0x3fe88ac7d98a6699, 0x3fe89bd0a478580f > + .quad 0x3fe8ace5422aa0db, 0x3fe8be05bad61778, 0x3fe8cf3216b5448c, 0x3fe8e06a5e0866d9 > + .quad 0x3fe8f1ae99157736, 0x3fe902fed0282c8a, 0x3fe9145b0b91ffc6, 0x3fe925c353aa2fe2 > + .quad 0x3fe93737b0cdc5e5, 0x3fe948b82b5f98e5, 0x3fe95a44cbc8520f, 0x3fe96bdd9a7670b3 > + .quad 0x3fe97d829fde4e50, 0x3fe98f33e47a22a2, 0x3fe9a0f170ca07ba, 0x3fe9b2bb4d53fe0d > + .quad 0x3fe9c49182a3f090, 0x3fe9d674194bb8d5, 0x3fe9e86319e32323, 0x3fe9fa5e8d07f29e > + .quad 0x3fea0c667b5de565, 0x3fea1e7aed8eb8bb, 0x3fea309bec4a2d33, 0x3fea42c980460ad8 > + .quad 0x3fea5503b23e255d, 0x3fea674a8af46052, 0x3fea799e1330b358, 0x3fea8bfe53c12e59 > + .quad 0x3fea9e6b5579fdbf, 0x3feab0e521356eba, 0x3feac36bbfd3f37a, 0x3fead5ff3a3c2774 > + .quad 0x3feae89f995ad3ad, 0x3feafb4ce622f2ff, 0x3feb0e07298db666, 0x3feb20ce6c9a8952 > + .quad 0x3feb33a2b84f15fb, 0x3feb468415b749b1, 0x3feb59728de5593a, 0x3feb6c6e29f1c52a > + .quad 0x3feb7f76f2fb5e47, 0x3feb928cf22749e4, 0x3feba5b030a1064a, 0x3febb8e0b79a6f1f > + .quad 0x3febcc1e904bc1d2, 0x3febdf69c3f3a207, 0x3febf2c25bd71e09, 0x3fec06286141b33d > + .quad 0x3fec199bdd85529c, 0x3fec2d1cd9fa652c, 0x3fec40ab5fffd07a, 0x3fec544778fafb22 > + .quad 0x3fec67f12e57d14b, 0x3fec7ba88988c933, 0x3fec8f6d9406e7b5, 0x3feca3405751c4db > + .quad 0x3fecb720dcef9069, 0x3feccb0f2e6d1675, 0x3fecdf0b555dc3fa, 0x3fecf3155b5bab74 > + .quad 0x3fed072d4a07897c, 0x3fed1b532b08c968, 0x3fed2f87080d89f2, 0x3fed43c8eacaa1d6 > + .quad 0x3fed5818dcfba487, 0x3fed6c76e862e6d3, 0x3fed80e316c98398, 0x3fed955d71ff6075 > + .quad 0x3feda9e603db3285, 0x3fedbe7cd63a8315, 0x3fedd321f301b460, 0x3fede7d5641c0658 > + .quad 0x3fedfc97337b9b5f, 0x3fee11676b197d17, 0x3fee264614f5a129, 0x3fee3b333b16ee12 > + .quad 0x3fee502ee78b3ff6, 0x3fee653924676d76, 0x3fee7a51fbc74c83, 0x3fee8f7977cdb740 > + .quad 0x3feea4afa2a490da, 0x3feeb9f4867cca6e, 0x3feecf482d8e67f1, 0x3feee4aaa2188510 > + .quad 0x3feefa1bee615a27, 0x3fef0f9c1cb6412a, 0x3fef252b376bba97, 0x3fef3ac948dd7274 > + .quad 0x3fef50765b6e4540, 0x3fef6632798844f8, 0x3fef7bfdad9cbe14, 0x3fef91d802243c89 > + .quad 0x3fefa7c1819e90d8, 0x3fefbdba3692d514, 0x3fefd3c22b8f71f1, 0x3fefe9d96b2a23d9 > + .quad 0x3ff0000000000000 > + .align 32 > + .quad 0x3ff71547652b82fe, 0x3ff71547652b82fe, 0x3ff71547652b82fe, 0x3ff71547652b82fe /* _dbInvLn2 = 1/log(2) */ > + .align 32 > + .quad 0x3FE62E42FEFC0000, 0x3FE62E42FEFC0000, 0x3FE62E42FEFC0000, 0x3FE62E42FEFC0000 /* _dbLn2hi = log(2) hi*/ > + .align 32 > + .quad 0xBDAC610CA86C3899, 0xBDAC610CA86C3899, 0xBDAC610CA86C3899, 0xBDAC610CA86C3899 /* _dbLn2lo = log(2) lo*/ > + .align 32 > + .quad 0x42B8000000000000, 0x42B8000000000000, 0x42B8000000000000, 0x42B8000000000000 /* _dbShifter */ > + .align 32 > + .long 0x000000FF, 0x000000FF, 0x000000FF, 0x000000FF, 0x000000FF, 0x000000FF, 0x000000FF, 0x000000FF /* _iIndexMask */ > + .align 32 > + .quad 0x3FDFFFFFFFFFFDBD, 0x3FDFFFFFFFFFFDBD, 0x3FDFFFFFFFFFFDBD, 0x3FDFFFFFFFFFFDBD /* _dPC2 */ > + .align 32 > + .quad 0x3FC5555570813E14, 0x3FC5555570813E14, 0x3FC5555570813E14, 0x3FC5555570813E14 /* _dPC3 */ > + .align 32 > + .quad 0x3FA55555CF16D299, 0x3FA55555CF16D299, 0x3FA55555CF16D299, 0x3FA55555CF16D299 /* _dPC4 */ > + .align 32 > + .long 0x00000100, 0x00000100, 0x00000100, 0x00000100, 0x00000100, 0x00000100, 0x00000100, 0x00000100 /* _iMaxIndex */ > + .align 32 > + .quad 0x7ff0000000000000, 0x7ff0000000000000, 0x7ff0000000000000, 0x7ff0000000000000 /* _lExpMask */ > + .align 32 > + .quad 0x8000000000000000, 0x8000000000000000, 0x8000000000000000, 0x8000000000000000 /* _dSign*/ > + .align 32 > + .long 0x40861d99, 0x40861d99, 0x40861d99, 0x40861d99, 0x40861d99, 0x40861d99, 0x40861d99, 0x40861d99 /* _iDomainRange 0x40861d9ac12a3e85 =(1021*2^K-0.5)*log(2)/2^K -needed for quick exp*/ > + .align 32 > + .type __svml_dcosh_data_internal,@object > + .size __svml_dcosh_data_internal,.-__svml_dcosh_data_internal > diff --git a/sysdeps/x86_64/fpu/multiarch/svml_d_cosh8_core-avx2.S b/sysdeps/x86_64/fpu/multiarch/svml_d_cosh8_core-avx2.S > new file mode 100644 > index 0000000000..8b385cc297 > --- /dev/null > +++ b/sysdeps/x86_64/fpu/multiarch/svml_d_cosh8_core-avx2.S > @@ -0,0 +1,20 @@ > +/* AVX2 version of vectorized cosh, vector length is 8. > + 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 > + . */ > + > +#define _ZGVeN8v_cosh _ZGVeN8v_cosh_avx2_wrapper > +#include "../svml_d_cosh8_core.S" > diff --git a/sysdeps/x86_64/fpu/multiarch/svml_d_cosh8_core.c b/sysdeps/x86_64/fpu/multiarch/svml_d_cosh8_core.c > new file mode 100644 > index 0000000000..576b3186d5 > --- /dev/null > +++ b/sysdeps/x86_64/fpu/multiarch/svml_d_cosh8_core.c > @@ -0,0 +1,27 @@ > +/* Multiple versions of vectorized cosh, vector length is 8. > + 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 > + . */ > + > +#define SYMBOL_NAME _ZGVeN8v_cosh > +#include "ifunc-mathvec-avx512-skx.h" > + > +libc_ifunc_redirected (REDIRECT_NAME, SYMBOL_NAME, IFUNC_SELECTOR ()); > + > +#ifdef SHARED > +__hidden_ver1 (_ZGVeN8v_cosh, __GI__ZGVeN8v_cosh, __redirect__ZGVeN8v_cosh) > + __attribute__ ((visibility ("hidden"))); > +#endif > diff --git a/sysdeps/x86_64/fpu/multiarch/svml_d_cosh8_core_avx512.S b/sysdeps/x86_64/fpu/multiarch/svml_d_cosh8_core_avx512.S > new file mode 100644 > index 0000000000..53040cef9a > --- /dev/null > +++ b/sysdeps/x86_64/fpu/multiarch/svml_d_cosh8_core_avx512.S > @@ -0,0 +1,323 @@ > +/* Function cosh vectorized with AVX-512. > + 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/. */ > + > +/* > + * ALGORITHM DESCRIPTION: > + * > + * Compute cosh(x) as (exp(x)+exp(-x))/2, > + * where exp is calculated as > + * exp(M*ln2 + ln2*(j/2^k) + r) = 2^M * 2^(j/2^k) * exp(r) > + * > + * Special cases: > + * > + * cosh(NaN) = quiet NaN, and raise invalid exception > + * cosh(INF) = that INF > + * cosh(0) = 1 > + * cosh(x) overflows for big x and returns MAXLOG+log(2) > + * > + */ > + > +/* Offsets for data table __svml_dcosh_data_internal > + */ > +#define _dTp_h 0 > +#define _dTn_h 128 > +#define _dbShifter_UISA 256 > +#define _dPC2_UISA 320 > +#define _dPC3_UISA 384 > +#define _dPC4_UISA 448 > +#define _dPC5_UISA 512 > +#define _dPC6_UISA 576 > +#define _dPC7_UISA 640 > +#define _dbInvLn2 704 > +#define _dbLn2hi 768 > +#define _dbLn2lo 832 > +#define _dbShifter 896 > +#define _dPC2 960 > +#define _dPC3 1024 > +#define _dPC4 1088 > +#define _lExpMask 1152 > +#define _dSign 1216 > +#define _iDomainRange 1280 > + > +#include > + > + .text > + .section .text.evex512,"ax",@progbits > +ENTRY(_ZGVeN8v_cosh_skx) > + pushq %rbp > + cfi_def_cfa_offset(16) > + movq %rsp, %rbp > + cfi_def_cfa(6, 16) > + cfi_offset(6, -16) > + andq $-64, %rsp > + subq $192, %rsp > + vmovups _dSign+__svml_dcosh_data_internal(%rip), %zmm11 > + vmovups _dbShifter_UISA+__svml_dcosh_data_internal(%rip), %zmm15 > + > +/* > + * Load argument > + * dM = x*2^K/log(2) + RShifter > + */ > + vmovups _dbInvLn2+__svml_dcosh_data_internal(%rip), %zmm4 > + vmovups _dbLn2hi+__svml_dcosh_data_internal(%rip), %zmm2 > + vmovups _dbLn2lo+__svml_dcosh_data_internal(%rip), %zmm3 > + vmovups _dPC7_UISA+__svml_dcosh_data_internal(%rip), %zmm8 > + vmovups _dPC6_UISA+__svml_dcosh_data_internal(%rip), %zmm9 > + vmovups _dPC2_UISA+__svml_dcosh_data_internal(%rip), %zmm7 > + vmovups _dPC3_UISA+__svml_dcosh_data_internal(%rip), %zmm6 > + vmovaps %zmm0, %zmm10 > + > +/* Abs argument */ > + vandnpd %zmm10, %zmm11, %zmm5 > + > +/* Index and lookup */ > + vmovups __svml_dcosh_data_internal(%rip), %zmm11 > + vmovups _dTn_h+__svml_dcosh_data_internal(%rip), %zmm0 > + vfmadd213pd {rn-sae}, %zmm15, %zmm5, %zmm4 > + > +/* > + * Check for overflow\underflow > + * > + */ > + vpsrlq $32, %zmm5, %zmm12 > + > +/* dN = dM - RShifter */ > + vsubpd {rn-sae}, %zmm15, %zmm4, %zmm1 > + vpmovqd %zmm12, %ymm13 > + vpermt2pd _dTn_h+64+__svml_dcosh_data_internal(%rip), %zmm4, %zmm0 > + vpermt2pd _dTp_h+64+__svml_dcosh_data_internal(%rip), %zmm4, %zmm11 > + > +/* dR = dX - dN*Log2_hi/2^K */ > + vfnmadd231pd {rn-sae}, %zmm2, %zmm1, %zmm5 > + > +/* > + * poly(r) = Gmjp(1 + a2*r^2 + a4*r^4) + Gmjn*(r+ a3*r^3 +a5*r^5) = > + * = Gmjp_h +Gmjp_l+ Gmjp*r^2*(a2 + a4*r^2) + Gmjn*(r+ r^3*(a3 +a5*r^2) > + */ > + vmovups _dPC5_UISA+__svml_dcosh_data_internal(%rip), %zmm12 > + vpsllq $48, %zmm4, %zmm2 > + > +/* dR = dX - dN*Log2_hi/2^K */ > + vfnmadd231pd {rn-sae}, %zmm3, %zmm1, %zmm5 > + vmulpd {rn-sae}, %zmm5, %zmm5, %zmm1 > + vfmadd231pd {rn-sae}, %zmm1, %zmm8, %zmm12 > + vmovups _dPC4_UISA+__svml_dcosh_data_internal(%rip), %zmm8 > + vfmadd213pd {rn-sae}, %zmm6, %zmm1, %zmm12 > + vfmadd231pd {rn-sae}, %zmm1, %zmm9, %zmm8 > + vfmadd213pd {rn-sae}, %zmm7, %zmm1, %zmm8 > + vpcmpgtd _iDomainRange+__svml_dcosh_data_internal(%rip), %ymm13, %ymm14 > + vmovmskps %ymm14, %edx > + > +/* dOut=r^2*(a2 + a4*r^2) */ > + vmulpd {rn-sae}, %zmm1, %zmm8, %zmm6 > + > +/* lM now is an EXP(2^N) */ > + vpandq _lExpMask+__svml_dcosh_data_internal(%rip), %zmm2, %zmm3 > + vpaddq %zmm3, %zmm11, %zmm4 > + vpsubq %zmm3, %zmm0, %zmm0 > + vsubpd {rn-sae}, %zmm0, %zmm4, %zmm14 > + vaddpd {rn-sae}, %zmm0, %zmm4, %zmm13 > + > +/* dM=r^2*(a3 +a5*r^2) */ > + vmulpd {rn-sae}, %zmm1, %zmm12, %zmm0 > + vfmadd213pd {rn-sae}, %zmm13, %zmm13, %zmm6 > + > +/* dM= r + r^3*(a3 +a5*r^2) */ > + vfmadd213pd {rn-sae}, %zmm5, %zmm5, %zmm0 > + vfmadd213pd {rn-sae}, %zmm6, %zmm14, %zmm0 > + testl %edx, %edx > + > +/* Go to special inputs processing branch */ > + jne L(SPECIAL_VALUES_BRANCH) > + # LOE rbx r12 r13 r14 r15 edx zmm0 zmm10 > + > +/* Restore registers > + * and exit the function > + */ > + > +L(EXIT): > + movq %rbp, %rsp > + popq %rbp > + cfi_def_cfa(7, 8) > + cfi_restore(6) > + ret > + cfi_def_cfa(6, 16) > + cfi_offset(6, -16) > + > +/* Branch to process > + * special inputs > + */ > + > +L(SPECIAL_VALUES_BRANCH): > + vmovups %zmm10, 64(%rsp) > + vmovups %zmm0, 128(%rsp) > + # LOE rbx r12 r13 r14 r15 edx zmm0 > + > + xorl %eax, %eax > + # LOE rbx r12 r13 r14 r15 eax edx > + > + vzeroupper > + movq %r12, 16(%rsp) > + /* DW_CFA_expression: r12 (r12) (DW_OP_lit8; DW_OP_minus; DW_OP_const4s: -64; DW_OP_and; DW_OP_const4s: -176; DW_OP_plus) */ > + .cfi_escape 0x10, 0x0c, 0x0e, 0x38, 0x1c, 0x0d, 0xc0, 0xff, 0xff, 0xff, 0x1a, 0x0d, 0x50, 0xff, 0xff, 0xff, 0x22 > + movl %eax, %r12d > + movq %r13, 8(%rsp) > + /* DW_CFA_expression: r13 (r13) (DW_OP_lit8; DW_OP_minus; DW_OP_const4s: -64; DW_OP_and; DW_OP_const4s: -184; DW_OP_plus) */ > + .cfi_escape 0x10, 0x0d, 0x0e, 0x38, 0x1c, 0x0d, 0xc0, 0xff, 0xff, 0xff, 0x1a, 0x0d, 0x48, 0xff, 0xff, 0xff, 0x22 > + movl %edx, %r13d > + movq %r14, (%rsp) > + /* DW_CFA_expression: r14 (r14) (DW_OP_lit8; DW_OP_minus; DW_OP_const4s: -64; DW_OP_and; DW_OP_const4s: -192; DW_OP_plus) */ > + .cfi_escape 0x10, 0x0e, 0x0e, 0x38, 0x1c, 0x0d, 0xc0, 0xff, 0xff, 0xff, 0x1a, 0x0d, 0x40, 0xff, 0xff, 0xff, 0x22 > + # LOE rbx r15 r12d r13d > + > +/* Range mask > + * bits check > + */ > + > +L(RANGEMASK_CHECK): > + btl %r12d, %r13d > + > +/* Call scalar math function */ > + jc L(SCALAR_MATH_CALL) > + # LOE rbx r15 r12d r13d > + > +/* Special inputs > + * processing loop > + */ > + > +L(SPECIAL_VALUES_LOOP): > + incl %r12d > + cmpl $8, %r12d > + > +/* Check bits in range mask */ > + jl L(RANGEMASK_CHECK) > + # LOE rbx r15 r12d r13d > + > + movq 16(%rsp), %r12 > + cfi_restore(12) > + movq 8(%rsp), %r13 > + cfi_restore(13) > + movq (%rsp), %r14 > + cfi_restore(14) > + vmovups 128(%rsp), %zmm0 > + > +/* Go to exit */ > + jmp L(EXIT) > + /* DW_CFA_expression: r12 (r12) (DW_OP_lit8; DW_OP_minus; DW_OP_const4s: -64; DW_OP_and; DW_OP_const4s: -176; DW_OP_plus) */ > + .cfi_escape 0x10, 0x0c, 0x0e, 0x38, 0x1c, 0x0d, 0xc0, 0xff, 0xff, 0xff, 0x1a, 0x0d, 0x50, 0xff, 0xff, 0xff, 0x22 > + /* DW_CFA_expression: r13 (r13) (DW_OP_lit8; DW_OP_minus; DW_OP_const4s: -64; DW_OP_and; DW_OP_const4s: -184; DW_OP_plus) */ > + .cfi_escape 0x10, 0x0d, 0x0e, 0x38, 0x1c, 0x0d, 0xc0, 0xff, 0xff, 0xff, 0x1a, 0x0d, 0x48, 0xff, 0xff, 0xff, 0x22 > + /* DW_CFA_expression: r14 (r14) (DW_OP_lit8; DW_OP_minus; DW_OP_const4s: -64; DW_OP_and; DW_OP_const4s: -192; DW_OP_plus) */ > + .cfi_escape 0x10, 0x0e, 0x0e, 0x38, 0x1c, 0x0d, 0xc0, 0xff, 0xff, 0xff, 0x1a, 0x0d, 0x40, 0xff, 0xff, 0xff, 0x22 > + # LOE rbx r12 r13 r14 r15 zmm0 > + > +/* Scalar math fucntion call > + * to process special input > + */ > + > +L(SCALAR_MATH_CALL): > + movl %r12d, %r14d > + movsd 64(%rsp,%r14,8), %xmm0 > + call cosh@PLT > + # LOE rbx r14 r15 r12d r13d xmm0 > + > + movsd %xmm0, 128(%rsp,%r14,8) > + > +/* Process special inputs in loop */ > + jmp L(SPECIAL_VALUES_LOOP) > + # LOE rbx r15 r12d r13d > +END(_ZGVeN8v_cosh_skx) > + > + .section .rodata, "a" > + .align 64 > + > +#ifdef __svml_dcosh_data_internal_typedef > +typedef unsigned int VUINT32; > +typedef struct > +{ > + __declspec(align(64)) VUINT32 _dTp_h[(1<<4)][2]; > + __declspec(align(64)) VUINT32 _dTn_h[(1<<4)][2]; > + __declspec(align(64)) VUINT32 _dbShifter_UISA[8][2]; > + __declspec(align(64)) VUINT32 _dPC2_UISA[8][2]; > + __declspec(align(64)) VUINT32 _dPC3_UISA[8][2]; > + __declspec(align(64)) VUINT32 _dPC4_UISA[8][2]; > + __declspec(align(64)) VUINT32 _dPC5_UISA[8][2]; > + __declspec(align(64)) VUINT32 _dPC6_UISA[8][2]; > + __declspec(align(64)) VUINT32 _dPC7_UISA[8][2]; > + __declspec(align(64)) VUINT32 _dbInvLn2[8][2]; > + __declspec(align(64)) VUINT32 _dbLn2hi[8][2]; > + __declspec(align(64)) VUINT32 _dbLn2lo[8][2]; > + __declspec(align(64)) VUINT32 _dbShifter[8][2]; > + __declspec(align(64)) VUINT32 _dPC2[8][2]; > + __declspec(align(64)) VUINT32 _dPC3[8][2]; > + __declspec(align(64)) VUINT32 _dPC4[8][2]; > + __declspec(align(64)) VUINT32 _lExpMask[8][2]; > + __declspec(align(64)) VUINT32 _dSign[8][2]; //0x8000000000000000 > + __declspec(align(64)) VUINT32 _iDomainRange[16][1]; > +} __svml_dcosh_data_internal; > +#endif > +__svml_dcosh_data_internal: > + /*== _dTp_h ==*/ > + .quad 0x3fe0000000000000, 0x3fe0b5586cf9890f, 0x3fe172b83c7d517b, 0x3fe2387a6e756238 > + .quad 0x3fe306fe0a31b715, 0x3fe3dea64c123422, 0x3fe4bfdad5362a27, 0x3fe5ab07dd485429 > + .quad 0x3fe6a09e667f3bcd, 0x3fe7a11473eb0187, 0x3fe8ace5422aa0db, 0x3fe9c49182a3f090 > + .quad 0x3feae89f995ad3ad, 0x3fec199bdd85529c, 0x3fed5818dcfba487, 0x3feea4afa2a490da > + /*== dTn_h ==*/ > + .align 64 > + .quad 0x3fe0000000000000, 0x3fdea4afa2a490da, 0x3fdd5818dcfba487, 0x3fdc199bdd85529c > + .quad 0x3fdae89f995ad3ad, 0x3fd9c49182a3f090, 0x3fd8ace5422aa0db, 0x3fd7a11473eb0187 > + .quad 0x3fd6a09e667f3bcd, 0x3fd5ab07dd485429, 0x3fd4bfdad5362a27, 0x3fd3dea64c123422 > + .quad 0x3fd306fe0a31b715, 0x3fd2387a6e756238, 0x3fd172b83c7d517b, 0x3fd0b5586cf9890f > + .align 64 > + .quad 0x42F8000000000000, 0x42F8000000000000, 0x42F8000000000000, 0x42F8000000000000, 0x42F8000000000000, 0x42F8000000000000, 0x42F8000000000000, 0x42F8000000000000 /* _dbShifter_UISA */ > + .align 64 > + .quad 0x3fe0000000000004, 0x3fe0000000000004, 0x3fe0000000000004, 0x3fe0000000000004, 0x3fe0000000000004, 0x3fe0000000000004, 0x3fe0000000000004, 0x3fe0000000000004 /* _dPC2_UISA */ > + .align 64 > + .quad 0x3fc5555555555543, 0x3fc5555555555543, 0x3fc5555555555543, 0x3fc5555555555543, 0x3fc5555555555543, 0x3fc5555555555543, 0x3fc5555555555543, 0x3fc5555555555543 /* _dPC3_UISA */ > + .align 64 > + .quad 0x3fa5555555484f37, 0x3fa5555555484f37, 0x3fa5555555484f37, 0x3fa5555555484f37, 0x3fa5555555484f37, 0x3fa5555555484f37, 0x3fa5555555484f37, 0x3fa5555555484f37 /* _dPC4_UISA */ > + .align 64 > + .quad 0x3f81111111286a0c, 0x3f81111111286a0c, 0x3f81111111286a0c, 0x3f81111111286a0c, 0x3f81111111286a0c, 0x3f81111111286a0c, 0x3f81111111286a0c, 0x3f81111111286a0c /* _dPC5_UISA */ > + .align 64 > + .quad 0x3f56c183da08f116, 0x3f56c183da08f116, 0x3f56c183da08f116, 0x3f56c183da08f116, 0x3f56c183da08f116, 0x3f56c183da08f116, 0x3f56c183da08f116, 0x3f56c183da08f116 /* _dPC6_UISA */ > + .align 64 > + .quad 0x3f2a018d76da03da, 0x3f2a018d76da03da, 0x3f2a018d76da03da, 0x3f2a018d76da03da, 0x3f2a018d76da03da, 0x3f2a018d76da03da, 0x3f2a018d76da03da, 0x3f2a018d76da03da /* _dPC7_UISA */ > + /*== _dbT ==*/ > + .align 64 > + .quad 0x3ff71547652b82fe, 0x3ff71547652b82fe, 0x3ff71547652b82fe, 0x3ff71547652b82fe, 0x3ff71547652b82fe, 0x3ff71547652b82fe, 0x3ff71547652b82fe, 0x3ff71547652b82fe /* _dbInvLn2 = 1/log(2) */ > + .align 64 > + .quad 0x3FE62E42FEFC0000, 0x3FE62E42FEFC0000, 0x3FE62E42FEFC0000, 0x3FE62E42FEFC0000, 0x3FE62E42FEFC0000, 0x3FE62E42FEFC0000, 0x3FE62E42FEFC0000, 0x3FE62E42FEFC0000 /* _dbLn2hi = log(2) hi*/ > + .align 64 > + .quad 0xBDAC610CA86C3899, 0xBDAC610CA86C3899, 0xBDAC610CA86C3899, 0xBDAC610CA86C3899, 0xBDAC610CA86C3899, 0xBDAC610CA86C3899, 0xBDAC610CA86C3899, 0xBDAC610CA86C3899 /* _dbLn2lo = log(2) lo*/ > + .align 64 > + .quad 0x42B8000000000000, 0x42B8000000000000, 0x42B8000000000000, 0x42B8000000000000, 0x42B8000000000000, 0x42B8000000000000, 0x42B8000000000000, 0x42B8000000000000 /* _dbShifter */ > + .align 64 > + .quad 0x3FDFFFFFFFFFFDBD, 0x3FDFFFFFFFFFFDBD, 0x3FDFFFFFFFFFFDBD, 0x3FDFFFFFFFFFFDBD, 0x3FDFFFFFFFFFFDBD, 0x3FDFFFFFFFFFFDBD, 0x3FDFFFFFFFFFFDBD, 0x3FDFFFFFFFFFFDBD /* _dPC2 */ > + .align 64 > + .quad 0x3FC5555570813E14, 0x3FC5555570813E14, 0x3FC5555570813E14, 0x3FC5555570813E14, 0x3FC5555570813E14, 0x3FC5555570813E14, 0x3FC5555570813E14, 0x3FC5555570813E14 /* _dPC3 */ > + .align 64 > + .quad 0x3FA55555CF16D299, 0x3FA55555CF16D299, 0x3FA55555CF16D299, 0x3FA55555CF16D299, 0x3FA55555CF16D299, 0x3FA55555CF16D299, 0x3FA55555CF16D299, 0x3FA55555CF16D299 /* _dPC4 */ > + .align 64 > + .quad 0x7ff0000000000000, 0x7ff0000000000000, 0x7ff0000000000000, 0x7ff0000000000000, 0x7ff0000000000000, 0x7ff0000000000000, 0x7ff0000000000000, 0x7ff0000000000000 /* _lExpMask */ > + .align 64 > + .quad 0x8000000000000000, 0x8000000000000000, 0x8000000000000000, 0x8000000000000000, 0x8000000000000000, 0x8000000000000000, 0x8000000000000000, 0x8000000000000000 /* _dSign*/ > + .align 64 > + .long 0x40861d99, 0x40861d99, 0x40861d99, 0x40861d99, 0x40861d99, 0x40861d99, 0x40861d99, 0x40861d99, 0x40861d99, 0x40861d99, 0x40861d99, 0x40861d99, 0x40861d99, 0x40861d99, 0x40861d99, 0x40861d99 /* _iDomainRange 0x40861d9ac12a3e85 =(1021*2^K-0.5)*log(2)/2^K -needed for quick exp*/ > + .align 64 > + .type __svml_dcosh_data_internal,@object > + .size __svml_dcosh_data_internal,.-__svml_dcosh_data_internal > diff --git a/sysdeps/x86_64/fpu/multiarch/svml_s_coshf16_core-avx2.S b/sysdeps/x86_64/fpu/multiarch/svml_s_coshf16_core-avx2.S > new file mode 100644 > index 0000000000..456d8a129f > --- /dev/null > +++ b/sysdeps/x86_64/fpu/multiarch/svml_s_coshf16_core-avx2.S > @@ -0,0 +1,20 @@ > +/* AVX2 version of vectorized coshf. > + 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 > + . */ > + > +#define _ZGVeN16v_coshf _ZGVeN16v_coshf_avx2_wrapper > +#include "../svml_s_coshf16_core.S" > diff --git a/sysdeps/x86_64/fpu/multiarch/svml_s_coshf16_core.c b/sysdeps/x86_64/fpu/multiarch/svml_s_coshf16_core.c > new file mode 100644 > index 0000000000..34c008871a > --- /dev/null > +++ b/sysdeps/x86_64/fpu/multiarch/svml_s_coshf16_core.c > @@ -0,0 +1,28 @@ > +/* Multiple versions of vectorized coshf, vector length is 16. > + 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 > + . */ > + > +#define SYMBOL_NAME _ZGVeN16v_coshf > +#include "ifunc-mathvec-avx512-skx.h" > + > +libc_ifunc_redirected (REDIRECT_NAME, SYMBOL_NAME, IFUNC_SELECTOR ()); > + > +#ifdef SHARED > +__hidden_ver1 (_ZGVeN16v_coshf, __GI__ZGVeN16v_coshf, > + __redirect__ZGVeN16v_coshf) > + __attribute__ ((visibility ("hidden"))); > +#endif > diff --git a/sysdeps/x86_64/fpu/multiarch/svml_s_coshf16_core_avx512.S b/sysdeps/x86_64/fpu/multiarch/svml_s_coshf16_core_avx512.S > new file mode 100644 > index 0000000000..276e3cfe4d > --- /dev/null > +++ b/sysdeps/x86_64/fpu/multiarch/svml_s_coshf16_core_avx512.S > @@ -0,0 +1,321 @@ > +/* Function coshf vectorized with AVX-512. > + 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/. */ > + > +/* > + * ALGORITHM DESCRIPTION: > + * > + * Compute cosh(x) as (exp(x)+exp(-x))/2, > + * where exp is calculated as > + * exp(M*ln2 + ln2*(j/2^k) + r) = 2^M * 2^(j/2^k) * exp(r) > + * > + * Special cases: > + * > + * cosh(NaN) = quiet NaN, and raise invalid exception > + * cosh(INF) = that INF > + * cosh(0) = 1 > + * cosh(x) overflows for big x and returns MAXLOG+log(2) > + * > + */ > + > +/* Offsets for data table __svml_scosh_data_internal > + */ > +#define _sExp_tbl_PH 0 > +#define _sExp_tbl_NH 128 > +#define _sShifter_UISA 256 > +#define _iDomainRange_UISA 320 > +#define _sPC1_UISA 384 > +#define _sPC2_UISA 448 > +#define _sPC3_UISA 512 > +#define _sInvLn2 576 > +#define _sLn2hi 640 > +#define _sLn2lo 704 > +#define _sSign 768 > +#define _iExpMask 832 > +#define _sShifter 896 > +#define _iDomainRange 960 > +#define _sPC1 1024 > +#define _sPC2 1088 > +#define _sPC3 1152 > + > +#include > + > + .text > + .section .text.exex512,"ax",@progbits > +ENTRY(_ZGVeN16v_coshf_skx) > + pushq %rbp > + cfi_def_cfa_offset(16) > + movq %rsp, %rbp > + cfi_def_cfa(6, 16) > + cfi_offset(6, -16) > + andq $-64, %rsp > + subq $192, %rsp > + vmovups _sSign+__svml_scosh_data_internal(%rip), %zmm4 > + vmovups _sShifter_UISA+__svml_scosh_data_internal(%rip), %zmm6 > + > +/* > + * Load argument > + * dM = x/log(2) + RShifter > + */ > + vmovups _sInvLn2+__svml_scosh_data_internal(%rip), %zmm10 > + vmovups _sLn2hi+__svml_scosh_data_internal(%rip), %zmm7 > + vmovups _sLn2lo+__svml_scosh_data_internal(%rip), %zmm9 > + > +/* */ > + vmovups _sPC3_UISA+__svml_scosh_data_internal(%rip), %zmm2 > + > +/* x^2 */ > + vmovups _sPC2_UISA+__svml_scosh_data_internal(%rip), %zmm3 > + > +/* G1,G2 2^N,2^(-N) */ > + vmovups __svml_scosh_data_internal(%rip), %zmm12 > + vmovups _sExp_tbl_NH+__svml_scosh_data_internal(%rip), %zmm13 > + > +/* > + * Implementation > + * Abs argument > + */ > + vandnps %zmm0, %zmm4, %zmm1 > + > +/* Check for overflow\underflow */ > + vpternlogd $255, %zmm5, %zmm5, %zmm5 > + vfmadd213ps {rn-sae}, %zmm6, %zmm1, %zmm10 > + vpcmpd $1, _iDomainRange_UISA+__svml_scosh_data_internal(%rip), %zmm1, %k1 > + > +/* iM now is an EXP(2^N) */ > + vpslld $18, %zmm10, %zmm11 > + > +/* > + * R > + * sN = sM - RShifter > + */ > + vsubps {rn-sae}, %zmm6, %zmm10, %zmm8 > + vpermt2ps _sExp_tbl_PH+64+__svml_scosh_data_internal(%rip), %zmm10, %zmm12 > + vpermt2ps _sExp_tbl_NH+64+__svml_scosh_data_internal(%rip), %zmm10, %zmm13 > + vpandnd %zmm1, %zmm1, %zmm5{%k1} > + > +/* sR = sX - sN*Log2_hi */ > + vfnmadd231ps {rn-sae}, %zmm7, %zmm8, %zmm1 > + vptestmd %zmm5, %zmm5, %k0 > + > +/* sR = (sX - sN*Log2_hi) - sN*Log2_lo */ > + vfnmadd231ps {rn-sae}, %zmm9, %zmm8, %zmm1 > + kmovw %k0, %edx > + vmulps {rn-sae}, %zmm1, %zmm1, %zmm4 > + vmulps {rn-sae}, %zmm4, %zmm2, %zmm2 > + > +/* sSinh_r = r + r*(r^2*(a3)) */ > + vfmadd213ps {rn-sae}, %zmm1, %zmm1, %zmm2 > + > +/* sOut = r^2*(a2) */ > + vmulps {rn-sae}, %zmm4, %zmm3, %zmm1 > + vpandd _iExpMask+__svml_scosh_data_internal(%rip), %zmm11, %zmm14 > + vpaddd %zmm14, %zmm12, %zmm15 > + vpsubd %zmm14, %zmm13, %zmm10 > + > +/* sG2 = 2^N*Th + 2^(-N)*T_h */ > + vaddps {rn-sae}, %zmm10, %zmm15, %zmm5 > + > +/* sG1 = 2^N*Th - 2^(-N)*T_h */ > + vsubps {rn-sae}, %zmm10, %zmm15, %zmm6 > + > +/* res = sG1*(r + r*(r^2*(a3))) + sG2*(1+r^2*(a2)) */ > + vfmadd213ps {rn-sae}, %zmm5, %zmm5, %zmm1 > + vfmadd213ps {rn-sae}, %zmm1, %zmm2, %zmm6 > + testl %edx, %edx > + > +/* Go to special inputs processing branch */ > + jne L(SPECIAL_VALUES_BRANCH) > + # LOE rbx r12 r13 r14 r15 edx zmm0 zmm6 > + > +/* Restore registers > + * and exit the function > + */ > + > +L(EXIT): > + vmovaps %zmm6, %zmm0 > + movq %rbp, %rsp > + popq %rbp > + cfi_def_cfa(7, 8) > + cfi_restore(6) > + ret > + cfi_def_cfa(6, 16) > + cfi_offset(6, -16) > + > +/* Branch to process > + * special inputs > + */ > + > +L(SPECIAL_VALUES_BRANCH): > + vmovups %zmm0, 64(%rsp) > + vmovups %zmm6, 128(%rsp) > + # LOE rbx r12 r13 r14 r15 edx zmm6 > + > + xorl %eax, %eax > + # LOE rbx r12 r13 r14 r15 eax edx > + > + vzeroupper > + movq %r12, 16(%rsp) > + /* DW_CFA_expression: r12 (r12) (DW_OP_lit8; DW_OP_minus; DW_OP_const4s: -64; DW_OP_and; DW_OP_const4s: -176; DW_OP_plus) */ > + .cfi_escape 0x10, 0x0c, 0x0e, 0x38, 0x1c, 0x0d, 0xc0, 0xff, 0xff, 0xff, 0x1a, 0x0d, 0x50, 0xff, 0xff, 0xff, 0x22 > + movl %eax, %r12d > + movq %r13, 8(%rsp) > + /* DW_CFA_expression: r13 (r13) (DW_OP_lit8; DW_OP_minus; DW_OP_const4s: -64; DW_OP_and; DW_OP_const4s: -184; DW_OP_plus) */ > + .cfi_escape 0x10, 0x0d, 0x0e, 0x38, 0x1c, 0x0d, 0xc0, 0xff, 0xff, 0xff, 0x1a, 0x0d, 0x48, 0xff, 0xff, 0xff, 0x22 > + movl %edx, %r13d > + movq %r14, (%rsp) > + /* DW_CFA_expression: r14 (r14) (DW_OP_lit8; DW_OP_minus; DW_OP_const4s: -64; DW_OP_and; DW_OP_const4s: -192; DW_OP_plus) */ > + .cfi_escape 0x10, 0x0e, 0x0e, 0x38, 0x1c, 0x0d, 0xc0, 0xff, 0xff, 0xff, 0x1a, 0x0d, 0x40, 0xff, 0xff, 0xff, 0x22 > + # LOE rbx r15 r12d r13d > + > +/* Range mask > + * bits check > + */ > + > +L(RANGEMASK_CHECK): > + btl %r12d, %r13d > + > +/* Call scalar math function */ > + jc L(SCALAR_MATH_CALL) > + # LOE rbx r15 r12d r13d > + > +/* Special inputs > + * processing loop > + */ > + > +L(SPECIAL_VALUES_LOOP): > + incl %r12d > + cmpl $16, %r12d > + > +/* Check bits in range mask */ > + jl L(RANGEMASK_CHECK) > + # LOE rbx r15 r12d r13d > + > + movq 16(%rsp), %r12 > + cfi_restore(12) > + movq 8(%rsp), %r13 > + cfi_restore(13) > + movq (%rsp), %r14 > + cfi_restore(14) > + vmovups 128(%rsp), %zmm6 > + > +/* Go to exit */ > + jmp L(EXIT) > + /* DW_CFA_expression: r12 (r12) (DW_OP_lit8; DW_OP_minus; DW_OP_const4s: -64; DW_OP_and; DW_OP_const4s: -176; DW_OP_plus) */ > + .cfi_escape 0x10, 0x0c, 0x0e, 0x38, 0x1c, 0x0d, 0xc0, 0xff, 0xff, 0xff, 0x1a, 0x0d, 0x50, 0xff, 0xff, 0xff, 0x22 > + /* DW_CFA_expression: r13 (r13) (DW_OP_lit8; DW_OP_minus; DW_OP_const4s: -64; DW_OP_and; DW_OP_const4s: -184; DW_OP_plus) */ > + .cfi_escape 0x10, 0x0d, 0x0e, 0x38, 0x1c, 0x0d, 0xc0, 0xff, 0xff, 0xff, 0x1a, 0x0d, 0x48, 0xff, 0xff, 0xff, 0x22 > + /* DW_CFA_expression: r14 (r14) (DW_OP_lit8; DW_OP_minus; DW_OP_const4s: -64; DW_OP_and; DW_OP_const4s: -192; DW_OP_plus) */ > + .cfi_escape 0x10, 0x0e, 0x0e, 0x38, 0x1c, 0x0d, 0xc0, 0xff, 0xff, 0xff, 0x1a, 0x0d, 0x40, 0xff, 0xff, 0xff, 0x22 > + # LOE rbx r12 r13 r14 r15 zmm6 > + > +/* Scalar math fucntion call > + * to process special input > + */ > + > +L(SCALAR_MATH_CALL): > + movl %r12d, %r14d > + movss 64(%rsp,%r14,4), %xmm0 > + call coshf@PLT > + # LOE rbx r14 r15 r12d r13d xmm0 > + > + movss %xmm0, 128(%rsp,%r14,4) > + > +/* Process special inputs in loop */ > + jmp L(SPECIAL_VALUES_LOOP) > + # LOE rbx r15 r12d r13d > +END(_ZGVeN16v_coshf_skx) > + > + .section .rodata, "a" > + .align 64 > + > +#ifdef __svml_scosh_data_internal_typedef > +typedef unsigned int VUINT32; > +typedef struct > +{ > + __declspec(align(64)) VUINT32 _sExp_tbl_PH[32][1]; > + __declspec(align(64)) VUINT32 _sExp_tbl_NH[32][1]; > + __declspec(align(64)) VUINT32 _sShifter_UISA[16][1]; > + __declspec(align(64)) VUINT32 _iDomainRange_UISA[16][1]; > + __declspec(align(64)) VUINT32 _sPC1_UISA[16][1]; > + __declspec(align(64)) VUINT32 _sPC2_UISA[16][1]; > + __declspec(align(64)) VUINT32 _sPC3_UISA[16][1]; > + __declspec(align(64)) VUINT32 _sInvLn2[16][1]; > + __declspec(align(64)) VUINT32 _sLn2hi[16][1]; > + __declspec(align(64)) VUINT32 _sLn2lo[16][1]; > + __declspec(align(64)) VUINT32 _sSign[16][1]; > + __declspec(align(64)) VUINT32 _iExpMask[16][1]; > + __declspec(align(64)) VUINT32 _sShifter[16][1]; > + __declspec(align(64)) VUINT32 _iDomainRange[16][1]; > + __declspec(align(64)) VUINT32 _sPC1[16][1]; > + __declspec(align(64)) VUINT32 _sPC2[16][1]; > + __declspec(align(64)) VUINT32 _sPC3[16][1]; > +} __svml_scosh_data_internal; > +#endif > +__svml_scosh_data_internal: > + /* _sExp_tbl_PH 2^(i/32-1), i=0..31 */ > + .long 0x3f000000, 0x3f02cd87, 0x3f05aac3, 0x3f08980f > + .long 0x3f0b95c2, 0x3f0ea43a, 0x3f11c3d3, 0x3f14f4f0 > + .long 0x3f1837f0, 0x3f1b8d3a, 0x3f1ef532, 0x3f227043 > + .long 0x3f25fed7, 0x3f29a15b, 0x3f2d583f, 0x3f3123f6 > + .long 0x3f3504f3, 0x3f38fbaf, 0x3f3d08a4, 0x3f412c4d > + .long 0x3f45672a, 0x3f49b9be, 0x3f4e248c, 0x3f52a81e > + .long 0x3f5744fd, 0x3f5bfbb8, 0x3f60ccdf, 0x3f65b907 > + .long 0x3f6ac0c7, 0x3f6fe4ba, 0x3f75257d, 0x3f7a83b3 > + /* _sExp_tbl_NH 2^(-i/32-1), i=0..31 */ > + .align 64 > + .long 0x3f000000, 0x3efa83b3, 0x3ef5257d, 0x3eefe4ba > + .long 0x3eeac0c7, 0x3ee5b907, 0x3ee0ccdf, 0x3edbfbb8 > + .long 0x3ed744fd, 0x3ed2a81e, 0x3ece248c, 0x3ec9b9be > + .long 0x3ec5672a, 0x3ec12c4d, 0x3ebd08a4, 0x3eb8fbaf > + .long 0x3eb504f3, 0x3eb123f6, 0x3ead583f, 0x3ea9a15b > + .long 0x3ea5fed7, 0x3ea27043, 0x3e9ef532, 0x3e9b8d3a > + .long 0x3e9837f0, 0x3e94f4f0, 0x3e91c3d3, 0x3e8ea43a > + .long 0x3e8b95c2, 0x3e88980f, 0x3e85aac3, 0x3e82cd87 > + .align 64 > + .long 0x48c00000, 0x48c00000, 0x48c00000, 0x48c00000, 0x48c00000, 0x48c00000, 0x48c00000, 0x48c00000, 0x48c00000, 0x48c00000, 0x48c00000, 0x48c00000, 0x48c00000, 0x48c00000, 0x48c00000, 0x48c00000 /* 1.5*2^18 _sShifter_UISA */ > + .align 64 > + .long 0x42AEAC4E, 0x42AEAC4E, 0x42AEAC4E, 0x42AEAC4E, 0x42AEAC4E, 0x42AEAC4E, 0x42AEAC4E, 0x42AEAC4E, 0x42AEAC4E, 0x42AEAC4E, 0x42AEAC4E, 0x42AEAC4E, 0x42AEAC4E, 0x42AEAC4E, 0x42AEAC4E, 0x42AEAC4E /* _iDomainRange_UISA */ > + .align 64 > + .long 0x3F800000, 0x3F800000, 0x3F800000, 0x3F800000, 0x3F800000, 0x3F800000, 0x3F800000, 0x3F800000, 0x3F800000, 0x3F800000, 0x3F800000, 0x3F800000, 0x3F800000, 0x3F800000, 0x3F800000, 0x3F800000 /* _sPC1_UISA=1 */ > + .align 64 > + .long 0x3f00010f, 0x3f00010f, 0x3f00010f, 0x3f00010f, 0x3f00010f, 0x3f00010f, 0x3f00010f, 0x3f00010f, 0x3f00010f, 0x3f00010f, 0x3f00010f, 0x3f00010f, 0x3f00010f, 0x3f00010f, 0x3f00010f, 0x3f00010f /* _sPC2_UISA */ > + .align 64 > + .long 0x3e2aaacd, 0x3e2aaacd, 0x3e2aaacd, 0x3e2aaacd, 0x3e2aaacd, 0x3e2aaacd, 0x3e2aaacd, 0x3e2aaacd, 0x3e2aaacd, 0x3e2aaacd, 0x3e2aaacd, 0x3e2aaacd, 0x3e2aaacd, 0x3e2aaacd, 0x3e2aaacd, 0x3e2aaacd /* _sPC3_UISA */ > + .align 64 > + .long 0x3FB8AA3B, 0x3FB8AA3B, 0x3FB8AA3B, 0x3FB8AA3B, 0x3FB8AA3B, 0x3FB8AA3B, 0x3FB8AA3B, 0x3FB8AA3B, 0x3FB8AA3B, 0x3FB8AA3B, 0x3FB8AA3B, 0x3FB8AA3B, 0x3FB8AA3B, 0x3FB8AA3B, 0x3FB8AA3B, 0x3FB8AA3B /* _sInvLn2 */ //k=0 > + .align 64 > + .long 0x3F317000, 0x3F317000, 0x3F317000, 0x3F317000, 0x3F317000, 0x3F317000, 0x3F317000, 0x3F317000, 0x3F317000, 0x3F317000, 0x3F317000, 0x3F317000, 0x3F317000, 0x3F317000, 0x3F317000, 0x3F317000 /* _sLn2hi */ > + .align 64 > + .long 0x3805fdf4, 0x3805fdf4, 0x3805fdf4, 0x3805fdf4, 0x3805fdf4, 0x3805fdf4, 0x3805fdf4, 0x3805fdf4, 0x3805fdf4, 0x3805fdf4, 0x3805fdf4, 0x3805fdf4, 0x3805fdf4, 0x3805fdf4, 0x3805fdf4, 0x3805fdf4 /* _sLn2lo */ > + .align 64 > + .long 0x80000000, 0x80000000, 0x80000000, 0x80000000, 0x80000000, 0x80000000, 0x80000000, 0x80000000, 0x80000000, 0x80000000, 0x80000000, 0x80000000, 0x80000000, 0x80000000, 0x80000000, 0x80000000 /* _sSign */ > + .align 64 > + .long 0x7f800000, 0x7f800000, 0x7f800000, 0x7f800000, 0x7f800000, 0x7f800000, 0x7f800000, 0x7f800000, 0x7f800000, 0x7f800000, 0x7f800000, 0x7f800000, 0x7f800000, 0x7f800000, 0x7f800000, 0x7f800000 /* _iExpMask */ > + .align 64 > + .long 0x4b400000, 0x4b400000, 0x4b400000, 0x4b400000, 0x4b400000, 0x4b400000, 0x4b400000, 0x4b400000, 0x4b400000, 0x4b400000, 0x4b400000, 0x4b400000, 0x4b400000, 0x4b400000, 0x4b400000, 0x4b400000 /* _sShifter */ > + .align 64 > + .long 0x42AEAC4E, 0x42AEAC4E, 0x42AEAC4E, 0x42AEAC4E, 0x42AEAC4E, 0x42AEAC4E, 0x42AEAC4E, 0x42AEAC4E, 0x42AEAC4E, 0x42AEAC4E, 0x42AEAC4E, 0x42AEAC4E, 0x42AEAC4E, 0x42AEAC4E, 0x42AEAC4E, 0x42AEAC4E /* _iDomainRange */ > + .align 64 > + .long 0x3F800000, 0x3F800000, 0x3F800000, 0x3F800000, 0x3F800000, 0x3F800000, 0x3F800000, 0x3F800000, 0x3F800000, 0x3F800000, 0x3F800000, 0x3F800000, 0x3F800000, 0x3F800000, 0x3F800000, 0x3F800000 /* _sPC1=1 */ > + .align 64 > + .long 0x3f000000, 0x3f000000, 0x3f000000, 0x3f000000, 0x3f000000, 0x3f000000, 0x3f000000, 0x3f000000, 0x3f000000, 0x3f000000, 0x3f000000, 0x3f000000, 0x3f000000, 0x3f000000, 0x3f000000, 0x3f000000 /* _sPC2 */ > + .align 64 > + .long 0x3e2aaa57, 0x3e2aaa57, 0x3e2aaa57, 0x3e2aaa57, 0x3e2aaa57, 0x3e2aaa57, 0x3e2aaa57, 0x3e2aaa57, 0x3e2aaa57, 0x3e2aaa57, 0x3e2aaa57, 0x3e2aaa57, 0x3e2aaa57, 0x3e2aaa57, 0x3e2aaa57, 0x3e2aaa57 /* _sPC3 */ > + .align 64 > + .type __svml_scosh_data_internal,@object > + .size __svml_scosh_data_internal,.-__svml_scosh_data_internal > diff --git a/sysdeps/x86_64/fpu/multiarch/svml_s_coshf4_core-sse2.S b/sysdeps/x86_64/fpu/multiarch/svml_s_coshf4_core-sse2.S > new file mode 100644 > index 0000000000..c719dc7d6a > --- /dev/null > +++ b/sysdeps/x86_64/fpu/multiarch/svml_s_coshf4_core-sse2.S > @@ -0,0 +1,20 @@ > +/* SSE2 version of vectorized coshf, vector length is 4. > + 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 > + . */ > + > +#define _ZGVbN4v_coshf _ZGVbN4v_coshf_sse2 > +#include "../svml_s_coshf4_core.S" > diff --git a/sysdeps/x86_64/fpu/multiarch/svml_s_coshf4_core.c b/sysdeps/x86_64/fpu/multiarch/svml_s_coshf4_core.c > new file mode 100644 > index 0000000000..c2dfcd44f8 > --- /dev/null > +++ b/sysdeps/x86_64/fpu/multiarch/svml_s_coshf4_core.c > @@ -0,0 +1,28 @@ > +/* Multiple versions of vectorized coshf, vector length is 4. > + 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 > + . */ > + > +#define SYMBOL_NAME _ZGVbN4v_coshf > +#include "ifunc-mathvec-sse4_1.h" > + > +libc_ifunc_redirected (REDIRECT_NAME, SYMBOL_NAME, IFUNC_SELECTOR ()); > + > +#ifdef SHARED > +__hidden_ver1 (_ZGVbN4v_coshf, __GI__ZGVbN4v_coshf, > + __redirect__ZGVbN4v_coshf) > + __attribute__ ((visibility ("hidden"))); > +#endif > diff --git a/sysdeps/x86_64/fpu/multiarch/svml_s_coshf4_core_sse4.S b/sysdeps/x86_64/fpu/multiarch/svml_s_coshf4_core_sse4.S > new file mode 100644 > index 0000000000..506f6a4bd9 > --- /dev/null > +++ b/sysdeps/x86_64/fpu/multiarch/svml_s_coshf4_core_sse4.S > @@ -0,0 +1,305 @@ > +/* Function coshf vectorized with SSE4. > + 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/. */ > + > +/* > + * ALGORITHM DESCRIPTION: > + * > + * Compute cosh(x) as (exp(x)+exp(-x))/2, > + * where exp is calculated as > + * exp(M*ln2 + ln2*(j/2^k) + r) = 2^M * 2^(j/2^k) * exp(r) > + * > + * Special cases: > + * > + * cosh(NaN) = quiet NaN, and raise invalid exception > + * cosh(INF) = that INF > + * cosh(0) = 1 > + * cosh(x) overflows for big x and returns MAXLOG+log(2) > + * > + */ > + > +/* Offsets for data table __svml_scosh_data_internal > + */ > +#define _sInvLn2 0 > +#define _sLn2hi 16 > +#define _sLn2lo 32 > +#define _sSign 48 > +#define _sShifter 64 > +#define _iDomainRange 80 > +#define _sPC1 96 > +#define _sPC2 112 > +#define _sPC3 128 > +#define _sPC4 144 > +#define _sPC5 160 > +#define _sPC6 176 > +#define _iHalf 192 > + > +#include > + > + .text > + .section .text.sse4,"ax",@progbits > +ENTRY(_ZGVbN4v_coshf_sse4) > + subq $72, %rsp > + cfi_def_cfa_offset(80) > + > +/* > + * Implementation > + * Abs argument > + */ > + movups _sSign+__svml_scosh_data_internal(%rip), %xmm1 > + > +/* > + * Load argument > + * dM = x/log(2) + RShifter > + */ > + movups _sInvLn2+__svml_scosh_data_internal(%rip), %xmm9 > + andnps %xmm0, %xmm1 > + mulps %xmm1, %xmm9 > + > +/* Check for overflow\underflow */ > + movaps %xmm1, %xmm3 > + movups _sShifter+__svml_scosh_data_internal(%rip), %xmm4 > + movups _sLn2hi+__svml_scosh_data_internal(%rip), %xmm5 > + addps %xmm4, %xmm9 > + > +/* > + * R > + * sN = sM - RShifter > + */ > + movaps %xmm9, %xmm6 > + > +/* > + * G1,G2 2^N,2^(-N) > + * iM now is an EXP(2^N) > + */ > + pslld $23, %xmm9 > + movups _sLn2lo+__svml_scosh_data_internal(%rip), %xmm7 > + subps %xmm4, %xmm6 > + > +/* sR = sX - sN*Log2_hi */ > + mulps %xmm6, %xmm5 > + > +/* sR = (sX - sN*Log2_hi) - sN*Log2_lo */ > + mulps %xmm6, %xmm7 > + movdqu _iDomainRange+__svml_scosh_data_internal(%rip), %xmm2 > + pcmpgtd %xmm2, %xmm3 > + pcmpeqd %xmm1, %xmm2 > + > +/* > + * sinh(r) = r*((a1=1)+r^2*(a3+r^2*(a5+{v1 r^2*a7})))) = r + r*(r^2*(a3+r^2*(a5+r^2*a7))) .... > + * sSinh_r = (a3+r^2*a5) > + */ > + movups _sPC5+__svml_scosh_data_internal(%rip), %xmm10 > + por %xmm2, %xmm3 > + > +/* > + * sinh(X) = sG2 + sG1*sinh(dR) + sG2*sR2*(a2+sR2*(a4+a6*sR2) > + * sOut = (a4 +a6*sR2) > + */ > + movups _sPC6+__svml_scosh_data_internal(%rip), %xmm11 > + subps %xmm5, %xmm1 > + movmskps %xmm3, %edx > + movdqu _iHalf+__svml_scosh_data_internal(%rip), %xmm8 > + subps %xmm7, %xmm1 > + > +/* sR2 = sR^2,shaffled */ > + movaps %xmm1, %xmm13 > + movdqa %xmm8, %xmm2 > + mulps %xmm1, %xmm13 > + paddd %xmm9, %xmm2 > + mulps %xmm13, %xmm10 > + psubd %xmm9, %xmm8 > + mulps %xmm13, %xmm11 > + addps _sPC3+__svml_scosh_data_internal(%rip), %xmm10 > + addps _sPC4+__svml_scosh_data_internal(%rip), %xmm11 > + > +/* sSinh_r = r^2*(a3+r^2*a5) */ > + mulps %xmm13, %xmm10 > + > +/* sOut = a2+sR2*(a4+a6*sR2) */ > + mulps %xmm13, %xmm11 > + > +/* sSinh_r = r + r*(r^2*(a3+r^2*a5)) */ > + mulps %xmm1, %xmm10 > + addps _sPC2+__svml_scosh_data_internal(%rip), %xmm11 > + addps %xmm10, %xmm1 > + > +/* sOut = sR2*(a2+sR2*(a4+a6*sR2) */ > + mulps %xmm11, %xmm13 > + > +/* sG1 = 2^(N-1)-2^(-N-1) */ > + movdqa %xmm2, %xmm12 > + > +/* sG2 = 2^(N-1)+2^(-N-1) */ > + addps %xmm8, %xmm2 > + subps %xmm8, %xmm12 > + > +/* sOut = sG2*sR2*(a2+sR2*(a4+a6*sR2) */ > + mulps %xmm2, %xmm13 > + > +/* sOut = sG1*sinh(dR)+sG2*sR2*(a2+sR2*(a4+a6*sR2) */ > + mulps %xmm1, %xmm12 > + addps %xmm12, %xmm13 > + > +/* sOut = sG2 + sG1*sinh(dR) + sG2*sR2*(a2+sR2*(a4+a6*sR2) */ > + addps %xmm13, %xmm2 > + > +/* Ret H */ > + testl %edx, %edx > + > +/* Go to special inputs processing branch */ > + jne L(SPECIAL_VALUES_BRANCH) > + # LOE rbx rbp r12 r13 r14 r15 edx xmm0 xmm2 > + > +/* Restore registers > + * and exit the function > + */ > + > +L(EXIT): > + movaps %xmm2, %xmm0 > + addq $72, %rsp > + cfi_def_cfa_offset(8) > + ret > + cfi_def_cfa_offset(80) > + > +/* Branch to process > + * special inputs > + */ > + > +L(SPECIAL_VALUES_BRANCH): > + movups %xmm0, 32(%rsp) > + movups %xmm2, 48(%rsp) > + # LOE rbx rbp r12 r13 r14 r15 edx > + > + xorl %eax, %eax > + movq %r12, 16(%rsp) > + cfi_offset(12, -64) > + movl %eax, %r12d > + movq %r13, 8(%rsp) > + cfi_offset(13, -72) > + movl %edx, %r13d > + movq %r14, (%rsp) > + cfi_offset(14, -80) > + # LOE rbx rbp r15 r12d r13d > + > +/* Range mask > + * bits check > + */ > + > +L(RANGEMASK_CHECK): > + btl %r12d, %r13d > + > +/* Call scalar math function */ > + jc L(SCALAR_MATH_CALL) > + # LOE rbx rbp r15 r12d r13d > + > +/* Special inputs > + * processing loop > + */ > + > +L(SPECIAL_VALUES_LOOP): > + incl %r12d > + cmpl $4, %r12d > + > +/* Check bits in range mask */ > + jl L(RANGEMASK_CHECK) > + # LOE rbx rbp r15 r12d r13d > + > + movq 16(%rsp), %r12 > + cfi_restore(12) > + movq 8(%rsp), %r13 > + cfi_restore(13) > + movq (%rsp), %r14 > + cfi_restore(14) > + movups 48(%rsp), %xmm2 > + > +/* Go to exit */ > + jmp L(EXIT) > + cfi_offset(12, -64) > + cfi_offset(13, -72) > + cfi_offset(14, -80) > + # LOE rbx rbp r12 r13 r14 r15 xmm2 > + > +/* Scalar math fucntion call > + * to process special input > + */ > + > +L(SCALAR_MATH_CALL): > + movl %r12d, %r14d > + movss 32(%rsp,%r14,4), %xmm0 > + call coshf@PLT > + # LOE rbx rbp r14 r15 r12d r13d xmm0 > + > + movss %xmm0, 48(%rsp,%r14,4) > + > +/* Process special inputs in loop */ > + jmp L(SPECIAL_VALUES_LOOP) > + # LOE rbx rbp r15 r12d r13d > +END(_ZGVbN4v_coshf_sse4) > + > + .section .rodata, "a" > + .align 16 > + > +#ifdef __svml_scosh_data_internal_typedef > +typedef unsigned int VUINT32; > +typedef struct > +{ > + __declspec(align(16)) VUINT32 _sInvLn2[4][1]; > + __declspec(align(16)) VUINT32 _sLn2hi[4][1]; > + __declspec(align(16)) VUINT32 _sLn2lo[4][1]; > + __declspec(align(16)) VUINT32 _sSign[4][1]; > + __declspec(align(16)) VUINT32 _sShifter[4][1]; > + __declspec(align(16)) VUINT32 _iDomainRange[4][1]; > + __declspec(align(16)) VUINT32 _sPC1[4][1]; > + __declspec(align(16)) VUINT32 _sPC2[4][1]; > + __declspec(align(16)) VUINT32 _sPC3[4][1]; > + __declspec(align(16)) VUINT32 _sPC4[4][1]; > + __declspec(align(16)) VUINT32 _sPC5[4][1]; > + __declspec(align(16)) VUINT32 _sPC6[4][1]; > + __declspec(align(16)) VUINT32 _iHalf[4][1]; > +} __svml_scosh_data_internal; > +#endif > +__svml_scosh_data_internal: > + .long 0x3FB8AA3B, 0x3FB8AA3B, 0x3FB8AA3B, 0x3FB8AA3B /* _sInvLn2 */ //k=0 > + .align 16 > + .long 0x3F317000, 0x3F317000, 0x3F317000, 0x3F317000 /* _sLn2hi */ > + .align 16 > + .long 0x3805fdf4, 0x3805fdf4, 0x3805fdf4, 0x3805fdf4 /* _sLn2lo */ > + .align 16 > + .long 0x80000000, 0x80000000, 0x80000000, 0x80000000 /* _sSign */ > + .align 16 > + .long 0x4b400000, 0x4b400000, 0x4b400000, 0x4b400000 /* _sShifter */ > + .align 16 > + .long 0x42AEAC4E, 0x42AEAC4E, 0x42AEAC4E, 0x42AEAC4E /* _iDomainRange */ > + .align 16 > + .long 0x3F800000, 0x3F800000, 0x3F800000, 0x3F800000 /* _sPC1=1 */ > + .align 16 > + .long 0x3f000000, 0x3f000000, 0x3f000000, 0x3f000000 /* _sPC2 */ > + .align 16 > + .long 0x3e2aaa57, 0x3e2aaa57, 0x3e2aaa57, 0x3e2aaa57 /* _sPC3 */ > + .align 16 > + .long 0x3d2aaa72, 0x3d2aaa72, 0x3d2aaa72, 0x3d2aaa72 /* _sPC4 */ > + .align 16 > + .long 0x3c091461, 0x3c091461, 0x3c091461, 0x3c091461 /* _sPC5 */ > + .align 16 > + .long 0x3ab6a8a3, 0x3ab6a8a3, 0x3ab6a8a3, 0x3ab6a8a3 /* _sPC6 */ > + // Integer constants > + .align 16 > + .long 0x3f000000, 0x3f000000, 0x3f000000, 0x3f000000 /* _iHalf*/ > + .align 16 > + .type __svml_scosh_data_internal,@object > + .size __svml_scosh_data_internal,.-__svml_scosh_data_internal > diff --git a/sysdeps/x86_64/fpu/multiarch/svml_s_coshf8_core-sse.S b/sysdeps/x86_64/fpu/multiarch/svml_s_coshf8_core-sse.S > new file mode 100644 > index 0000000000..c27229e1fa > --- /dev/null > +++ b/sysdeps/x86_64/fpu/multiarch/svml_s_coshf8_core-sse.S > @@ -0,0 +1,20 @@ > +/* SSE version of vectorized coshf, vector length is 8. > + 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 > + . */ > + > +#define _ZGVdN8v_coshf _ZGVdN8v_coshf_sse_wrapper > +#include "../svml_s_coshf8_core.S" > diff --git a/sysdeps/x86_64/fpu/multiarch/svml_s_coshf8_core.c b/sysdeps/x86_64/fpu/multiarch/svml_s_coshf8_core.c > new file mode 100644 > index 0000000000..e82818b2c9 > --- /dev/null > +++ b/sysdeps/x86_64/fpu/multiarch/svml_s_coshf8_core.c > @@ -0,0 +1,28 @@ > +/* Multiple versions of vectorized coshf, vector length is 8. > + 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 > + . */ > + > +#define SYMBOL_NAME _ZGVdN8v_coshf > +#include "ifunc-mathvec-avx2.h" > + > +libc_ifunc_redirected (REDIRECT_NAME, SYMBOL_NAME, IFUNC_SELECTOR ()); > + > +#ifdef SHARED > +__hidden_ver1 (_ZGVdN8v_coshf, __GI__ZGVdN8v_coshf, > + __redirect__ZGVdN8v_coshf) > + __attribute__ ((visibility ("hidden"))); > +#endif > diff --git a/sysdeps/x86_64/fpu/multiarch/svml_s_coshf8_core_avx2.S b/sysdeps/x86_64/fpu/multiarch/svml_s_coshf8_core_avx2.S > new file mode 100644 > index 0000000000..9149061e7e > --- /dev/null > +++ b/sysdeps/x86_64/fpu/multiarch/svml_s_coshf8_core_avx2.S > @@ -0,0 +1,308 @@ > +/* Function coshf vectorized with AVX2. > + 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/. */ > + > +/* > + * ALGORITHM DESCRIPTION: > + * > + * Compute cosh(x) as (exp(x)+exp(-x))/2, > + * where exp is calculated as > + * exp(M*ln2 + ln2*(j/2^k) + r) = 2^M * 2^(j/2^k) * exp(r) > + * > + * Special cases: > + * > + * cosh(NaN) = quiet NaN, and raise invalid exception > + * cosh(INF) = that INF > + * cosh(0) = 1 > + * cosh(x) overflows for big x and returns MAXLOG+log(2) > + * > + */ > + > +/* Offsets for data table __svml_scosh_data_internal > + */ > +#define _sInvLn2 0 > +#define _sLn2hi 32 > +#define _sLn2lo 64 > +#define _sSign 96 > +#define _sShifter 128 > +#define _iDomainRange 160 > +#define _sPC1 192 > +#define _sPC2 224 > +#define _sPC3 256 > +#define _sPC4 288 > +#define _sPC5 320 > +#define _sPC6 352 > +#define _iHalf 384 > + > +#include > + > + .text > + .section .text.avx2,"ax",@progbits > +ENTRY(_ZGVdN8v_coshf_avx2) > + pushq %rbp > + cfi_def_cfa_offset(16) > + movq %rsp, %rbp > + cfi_def_cfa(6, 16) > + cfi_offset(6, -16) > + andq $-32, %rsp > + subq $96, %rsp > + vmovups _sSign+__svml_scosh_data_internal(%rip), %ymm2 > + vmovups _sShifter+__svml_scosh_data_internal(%rip), %ymm7 > + > +/* > + * Load argument > + * dM = x/log(2) + RShifter > + */ > + vmovups _sInvLn2+__svml_scosh_data_internal(%rip), %ymm10 > + vmovups _sLn2hi+__svml_scosh_data_internal(%rip), %ymm8 > + vmovups _iDomainRange+__svml_scosh_data_internal(%rip), %ymm3 > + > +/* > + * sinh(r) = r*((a1=1)+r^2*(a3+r^2*(a5+{v1 r^2*a7})))) = r + r*(r^2*(a3+r^2*(a5+r^2*a7))) .... > + * sSinh_r = (a3+r^2*a5) > + */ > + vmovups _sPC5+__svml_scosh_data_internal(%rip), %ymm15 > + vmovups _iHalf+__svml_scosh_data_internal(%rip), %ymm11 > + vmovaps %ymm0, %ymm1 > + > +/* > + * Implementation > + * Abs argument > + */ > + vandnps %ymm1, %ymm2, %ymm0 > + vfmadd213ps %ymm7, %ymm0, %ymm10 > + > +/* > + * R > + * sN = sM - RShifter > + */ > + vsubps %ymm7, %ymm10, %ymm9 > + > +/* > + * G1,G2 2^N,2^(-N) > + * iM now is an EXP(2^N) > + */ > + vpslld $23, %ymm10, %ymm12 > + > +/* Check for overflow\underflow */ > + vpcmpgtd %ymm3, %ymm0, %ymm4 > + vpcmpeqd %ymm3, %ymm0, %ymm5 > + > +/* sR = sX - sN*Log2_hi */ > + vfnmadd231ps %ymm8, %ymm9, %ymm0 > + vpaddd %ymm12, %ymm11, %ymm13 > + vpsubd %ymm12, %ymm11, %ymm14 > + vpor %ymm5, %ymm4, %ymm6 > + > +/* sR = (sX - sN*Log2_hi) - sN*Log2_lo */ > + vfnmadd231ps _sLn2lo+__svml_scosh_data_internal(%rip), %ymm9, %ymm0 > + > +/* sG1 = 2^(N-1)-2^(-N-1) */ > + vsubps %ymm14, %ymm13, %ymm4 > + > +/* sG2 = 2^(N-1)+2^(-N-1) */ > + vaddps %ymm14, %ymm13, %ymm3 > + > +/* sR2 = sR^2,shaffled */ > + vmulps %ymm0, %ymm0, %ymm2 > + vfmadd213ps _sPC3+__svml_scosh_data_internal(%rip), %ymm2, %ymm15 > + > +/* sSinh_r = r^2*(a3+r^2*a5) */ > + vmulps %ymm15, %ymm2, %ymm13 > + > +/* sSinh_r = r + r*(r^2*(a3+r^2*a5)) */ > + vfmadd213ps %ymm0, %ymm0, %ymm13 > + > +/* > + * sinh(X) = sG2 + sG1*sinh(dR) + sG2*sR2*(a2+sR2*(a4+a6*sR2) > + * sOut = (a4 +a6*sR2) > + */ > + vmovups _sPC6+__svml_scosh_data_internal(%rip), %ymm0 > + vfmadd213ps _sPC4+__svml_scosh_data_internal(%rip), %ymm2, %ymm0 > + > +/* sOut = a2+sR2*(a4+a6*sR2) */ > + vfmadd213ps _sPC2+__svml_scosh_data_internal(%rip), %ymm2, %ymm0 > + > +/* sOut = sR2*(a2+sR2*(a4+a6*sR2) */ > + vmulps %ymm0, %ymm2, %ymm15 > + > +/* sOut = sG2*sR2*(a2+sR2*(a4+a6*sR2) */ > + vmulps %ymm15, %ymm3, %ymm14 > + > +/* sOut = sG1*sinh(dR)+sG2*sR2*(a2+sR2*(a4+a6*sR2) */ > + vfmadd213ps %ymm14, %ymm13, %ymm4 > + vmovmskps %ymm6, %edx > + > +/* sOut = sG2 + sG1*sinh(dR) + sG2*sR2*(a2+sR2*(a4+a6*sR2) */ > + vaddps %ymm4, %ymm3, %ymm0 > + > +/* Ret H */ > + testl %edx, %edx > + > +/* Go to special inputs processing branch */ > + jne L(SPECIAL_VALUES_BRANCH) > + # LOE rbx r12 r13 r14 r15 edx ymm0 ymm1 > + > +/* Restore registers > + * and exit the function > + */ > + > +L(EXIT): > + movq %rbp, %rsp > + popq %rbp > + cfi_def_cfa(7, 8) > + cfi_restore(6) > + ret > + cfi_def_cfa(6, 16) > + cfi_offset(6, -16) > + > +/* Branch to process > + * special inputs > + */ > + > +L(SPECIAL_VALUES_BRANCH): > + vmovups %ymm1, 32(%rsp) > + vmovups %ymm0, 64(%rsp) > + # LOE rbx r12 r13 r14 r15 edx ymm0 > + > + xorl %eax, %eax > + # LOE rbx r12 r13 r14 r15 eax edx > + > + vzeroupper > + movq %r12, 16(%rsp) > + /* DW_CFA_expression: r12 (r12) (DW_OP_lit8; DW_OP_minus; DW_OP_const4s: -32; DW_OP_and; DW_OP_const4s: -80; DW_OP_plus) */ > + .cfi_escape 0x10, 0x0c, 0x0e, 0x38, 0x1c, 0x0d, 0xe0, 0xff, 0xff, 0xff, 0x1a, 0x0d, 0xb0, 0xff, 0xff, 0xff, 0x22 > + movl %eax, %r12d > + movq %r13, 8(%rsp) > + /* DW_CFA_expression: r13 (r13) (DW_OP_lit8; DW_OP_minus; DW_OP_const4s: -32; DW_OP_and; DW_OP_const4s: -88; DW_OP_plus) */ > + .cfi_escape 0x10, 0x0d, 0x0e, 0x38, 0x1c, 0x0d, 0xe0, 0xff, 0xff, 0xff, 0x1a, 0x0d, 0xa8, 0xff, 0xff, 0xff, 0x22 > + movl %edx, %r13d > + movq %r14, (%rsp) > + /* DW_CFA_expression: r14 (r14) (DW_OP_lit8; DW_OP_minus; DW_OP_const4s: -32; DW_OP_and; DW_OP_const4s: -96; DW_OP_plus) */ > + .cfi_escape 0x10, 0x0e, 0x0e, 0x38, 0x1c, 0x0d, 0xe0, 0xff, 0xff, 0xff, 0x1a, 0x0d, 0xa0, 0xff, 0xff, 0xff, 0x22 > + # LOE rbx r15 r12d r13d > + > +/* Range mask > + * bits check > + */ > + > +L(RANGEMASK_CHECK): > + btl %r12d, %r13d > + > +/* Call scalar math function */ > + jc L(SCALAR_MATH_CALL) > + # LOE rbx r15 r12d r13d > + > +/* Special inputs > + * processing loop > + */ > + > +L(SPECIAL_VALUES_LOOP): > + incl %r12d > + cmpl $8, %r12d > + > +/* Check bits in range mask */ > + jl L(RANGEMASK_CHECK) > + # LOE rbx r15 r12d r13d > + > + movq 16(%rsp), %r12 > + cfi_restore(12) > + movq 8(%rsp), %r13 > + cfi_restore(13) > + movq (%rsp), %r14 > + cfi_restore(14) > + vmovups 64(%rsp), %ymm0 > + > +/* Go to exit */ > + jmp L(EXIT) > + /* DW_CFA_expression: r12 (r12) (DW_OP_lit8; DW_OP_minus; DW_OP_const4s: -32; DW_OP_and; DW_OP_const4s: -80; DW_OP_plus) */ > + .cfi_escape 0x10, 0x0c, 0x0e, 0x38, 0x1c, 0x0d, 0xe0, 0xff, 0xff, 0xff, 0x1a, 0x0d, 0xb0, 0xff, 0xff, 0xff, 0x22 > + /* DW_CFA_expression: r13 (r13) (DW_OP_lit8; DW_OP_minus; DW_OP_const4s: -32; DW_OP_and; DW_OP_const4s: -88; DW_OP_plus) */ > + .cfi_escape 0x10, 0x0d, 0x0e, 0x38, 0x1c, 0x0d, 0xe0, 0xff, 0xff, 0xff, 0x1a, 0x0d, 0xa8, 0xff, 0xff, 0xff, 0x22 > + /* DW_CFA_expression: r14 (r14) (DW_OP_lit8; DW_OP_minus; DW_OP_const4s: -32; DW_OP_and; DW_OP_const4s: -96; DW_OP_plus) */ > + .cfi_escape 0x10, 0x0e, 0x0e, 0x38, 0x1c, 0x0d, 0xe0, 0xff, 0xff, 0xff, 0x1a, 0x0d, 0xa0, 0xff, 0xff, 0xff, 0x22 > + # LOE rbx r12 r13 r14 r15 ymm0 > + > +/* Scalar math fucntion call > + * to process special input > + */ > + > +L(SCALAR_MATH_CALL): > + movl %r12d, %r14d > + movss 32(%rsp,%r14,4), %xmm0 > + call coshf@PLT > + # LOE rbx r14 r15 r12d r13d xmm0 > + > + movss %xmm0, 64(%rsp,%r14,4) > + > +/* Process special inputs in loop */ > + jmp L(SPECIAL_VALUES_LOOP) > + # LOE rbx r15 r12d r13d > +END(_ZGVdN8v_coshf_avx2) > + > + .section .rodata, "a" > + .align 32 > + > +#ifdef __svml_scosh_data_internal_typedef > +typedef unsigned int VUINT32; > +typedef struct > +{ > + __declspec(align(32)) VUINT32 _sInvLn2[8][1]; > + __declspec(align(32)) VUINT32 _sLn2hi[8][1]; > + __declspec(align(32)) VUINT32 _sLn2lo[8][1]; > + __declspec(align(32)) VUINT32 _sSign[8][1]; > + __declspec(align(32)) VUINT32 _sShifter[8][1]; > + __declspec(align(32)) VUINT32 _iDomainRange[8][1]; > + __declspec(align(32)) VUINT32 _sPC1[8][1]; > + __declspec(align(32)) VUINT32 _sPC2[8][1]; > + __declspec(align(32)) VUINT32 _sPC3[8][1]; > + __declspec(align(32)) VUINT32 _sPC4[8][1]; > + __declspec(align(32)) VUINT32 _sPC5[8][1]; > + __declspec(align(32)) VUINT32 _sPC6[8][1]; > + __declspec(align(32)) VUINT32 _iHalf[8][1]; > +} __svml_scosh_data_internal; > +#endif > +__svml_scosh_data_internal: > + .long 0x3FB8AA3B, 0x3FB8AA3B, 0x3FB8AA3B, 0x3FB8AA3B, 0x3FB8AA3B, 0x3FB8AA3B, 0x3FB8AA3B, 0x3FB8AA3B /* _sInvLn2 */ //k=0 > + .align 32 > + .long 0x3F317000, 0x3F317000, 0x3F317000, 0x3F317000, 0x3F317000, 0x3F317000, 0x3F317000, 0x3F317000 /* _sLn2hi */ > + .align 32 > + .long 0x3805fdf4, 0x3805fdf4, 0x3805fdf4, 0x3805fdf4, 0x3805fdf4, 0x3805fdf4, 0x3805fdf4, 0x3805fdf4 /* _sLn2lo */ > + .align 32 > + .long 0x80000000, 0x80000000, 0x80000000, 0x80000000, 0x80000000, 0x80000000, 0x80000000, 0x80000000 /* _sSign */ > + .align 32 > + .long 0x4b400000, 0x4b400000, 0x4b400000, 0x4b400000, 0x4b400000, 0x4b400000, 0x4b400000, 0x4b400000 /* _sShifter */ > + .align 32 > + .long 0x42AEAC4E, 0x42AEAC4E, 0x42AEAC4E, 0x42AEAC4E, 0x42AEAC4E, 0x42AEAC4E, 0x42AEAC4E, 0x42AEAC4E /* _iDomainRange */ > + .align 32 > + .long 0x3F800000, 0x3F800000, 0x3F800000, 0x3F800000, 0x3F800000, 0x3F800000, 0x3F800000, 0x3F800000 /* _sPC1=1 */ > + .align 32 > + .long 0x3f000000, 0x3f000000, 0x3f000000, 0x3f000000, 0x3f000000, 0x3f000000, 0x3f000000, 0x3f000000 /* _sPC2 */ > + .align 32 > + .long 0x3e2aaa57, 0x3e2aaa57, 0x3e2aaa57, 0x3e2aaa57, 0x3e2aaa57, 0x3e2aaa57, 0x3e2aaa57, 0x3e2aaa57 /* _sPC3 */ > + .align 32 > + .long 0x3d2aaa72, 0x3d2aaa72, 0x3d2aaa72, 0x3d2aaa72, 0x3d2aaa72, 0x3d2aaa72, 0x3d2aaa72, 0x3d2aaa72 /* _sPC4 */ > + .align 32 > + .long 0x3c091461, 0x3c091461, 0x3c091461, 0x3c091461, 0x3c091461, 0x3c091461, 0x3c091461, 0x3c091461 /* _sPC5 */ > + .align 32 > + .long 0x3ab6a8a3, 0x3ab6a8a3, 0x3ab6a8a3, 0x3ab6a8a3, 0x3ab6a8a3, 0x3ab6a8a3, 0x3ab6a8a3, 0x3ab6a8a3 /* _sPC6 */ > + // Integer constants > + .align 32 > + .long 0x3f000000, 0x3f000000, 0x3f000000, 0x3f000000, 0x3f000000, 0x3f000000, 0x3f000000, 0x3f000000 /* _iHalf*/ > + .align 32 > + .type __svml_scosh_data_internal,@object > + .size __svml_scosh_data_internal,.-__svml_scosh_data_internal > diff --git a/sysdeps/x86_64/fpu/svml_d_cosh2_core.S b/sysdeps/x86_64/fpu/svml_d_cosh2_core.S > new file mode 100644 > index 0000000000..f95952cfe5 > --- /dev/null > +++ b/sysdeps/x86_64/fpu/svml_d_cosh2_core.S > @@ -0,0 +1,29 @@ > +/* Function cosh vectorized with SSE2. > + 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 > + . */ > + > +#include > +#include "svml_d_wrapper_impl.h" > + > + .text > +ENTRY (_ZGVbN2v_cosh) > +WRAPPER_IMPL_SSE2 cosh > +END (_ZGVbN2v_cosh) > + > +#ifndef USE_MULTIARCH > + libmvec_hidden_def (_ZGVbN2v_cosh) > +#endif > diff --git a/sysdeps/x86_64/fpu/svml_d_cosh4_core.S b/sysdeps/x86_64/fpu/svml_d_cosh4_core.S > new file mode 100644 > index 0000000000..cc24d0fb6b > --- /dev/null > +++ b/sysdeps/x86_64/fpu/svml_d_cosh4_core.S > @@ -0,0 +1,29 @@ > +/* Function cosh vectorized with AVX2, wrapper 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 > + . */ > + > +#include > +#include "svml_d_wrapper_impl.h" > + > + .text > +ENTRY (_ZGVdN4v_cosh) > +WRAPPER_IMPL_AVX _ZGVbN2v_cosh > +END (_ZGVdN4v_cosh) > + > +#ifndef USE_MULTIARCH > + libmvec_hidden_def (_ZGVdN4v_cosh) > +#endif > diff --git a/sysdeps/x86_64/fpu/svml_d_cosh4_core_avx.S b/sysdeps/x86_64/fpu/svml_d_cosh4_core_avx.S > new file mode 100644 > index 0000000000..4323f5e308 > --- /dev/null > +++ b/sysdeps/x86_64/fpu/svml_d_cosh4_core_avx.S > @@ -0,0 +1,25 @@ > +/* Function cosh vectorized in AVX ISA as wrapper to SSE4 ISA 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 > + . */ > + > +#include > +#include "svml_d_wrapper_impl.h" > + > + .text > +ENTRY (_ZGVcN4v_cosh) > +WRAPPER_IMPL_AVX _ZGVbN2v_cosh > +END (_ZGVcN4v_cosh) > diff --git a/sysdeps/x86_64/fpu/svml_d_cosh8_core.S b/sysdeps/x86_64/fpu/svml_d_cosh8_core.S > new file mode 100644 > index 0000000000..90ee1ca125 > --- /dev/null > +++ b/sysdeps/x86_64/fpu/svml_d_cosh8_core.S > @@ -0,0 +1,25 @@ > +/* Function cosh vectorized with AVX-512, wrapper to AVX2. > + 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 > + . */ > + > +#include > +#include "svml_d_wrapper_impl.h" > + > + .text > +ENTRY (_ZGVeN8v_cosh) > +WRAPPER_IMPL_AVX512 _ZGVdN4v_cosh > +END (_ZGVeN8v_cosh) > diff --git a/sysdeps/x86_64/fpu/svml_s_coshf16_core.S b/sysdeps/x86_64/fpu/svml_s_coshf16_core.S > new file mode 100644 > index 0000000000..fe243b8b94 > --- /dev/null > +++ b/sysdeps/x86_64/fpu/svml_s_coshf16_core.S > @@ -0,0 +1,25 @@ > +/* Function coshf vectorized with AVX-512. Wrapper to AVX2 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 > + . */ > + > +#include > +#include "svml_s_wrapper_impl.h" > + > + .text > +ENTRY (_ZGVeN16v_coshf) > +WRAPPER_IMPL_AVX512 _ZGVdN8v_coshf > +END (_ZGVeN16v_coshf) > diff --git a/sysdeps/x86_64/fpu/svml_s_coshf4_core.S b/sysdeps/x86_64/fpu/svml_s_coshf4_core.S > new file mode 100644 > index 0000000000..b55ede6e38 > --- /dev/null > +++ b/sysdeps/x86_64/fpu/svml_s_coshf4_core.S > @@ -0,0 +1,29 @@ > +/* Function coshf vectorized with SSE2, wrapper 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 > + . */ > + > +#include > +#include "svml_s_wrapper_impl.h" > + > + .text > +ENTRY (_ZGVbN4v_coshf) > +WRAPPER_IMPL_SSE2 coshf > +END (_ZGVbN4v_coshf) > + > +#ifndef USE_MULTIARCH > + libmvec_hidden_def (_ZGVbN4v_coshf) > +#endif > diff --git a/sysdeps/x86_64/fpu/svml_s_coshf8_core.S b/sysdeps/x86_64/fpu/svml_s_coshf8_core.S > new file mode 100644 > index 0000000000..3ea02d0f19 > --- /dev/null > +++ b/sysdeps/x86_64/fpu/svml_s_coshf8_core.S > @@ -0,0 +1,29 @@ > +/* Function coshf vectorized with AVX2, wrapper 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 > + . */ > + > +#include > +#include "svml_s_wrapper_impl.h" > + > + .text > +ENTRY (_ZGVdN8v_coshf) > +WRAPPER_IMPL_AVX _ZGVbN4v_coshf > +END (_ZGVdN8v_coshf) > + > +#ifndef USE_MULTIARCH > + libmvec_hidden_def (_ZGVdN8v_coshf) > +#endif > diff --git a/sysdeps/x86_64/fpu/svml_s_coshf8_core_avx.S b/sysdeps/x86_64/fpu/svml_s_coshf8_core_avx.S > new file mode 100644 > index 0000000000..9b3002f7c9 > --- /dev/null > +++ b/sysdeps/x86_64/fpu/svml_s_coshf8_core_avx.S > @@ -0,0 +1,25 @@ > +/* Function coshf vectorized in AVX ISA as wrapper to SSE4 ISA 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 > + . */ > + > +#include > +#include "svml_s_wrapper_impl.h" > + > + .text > +ENTRY (_ZGVcN8v_coshf) > +WRAPPER_IMPL_AVX _ZGVbN4v_coshf > +END (_ZGVcN8v_coshf) > diff --git a/sysdeps/x86_64/fpu/test-double-libmvec-cosh-avx.c b/sysdeps/x86_64/fpu/test-double-libmvec-cosh-avx.c > new file mode 100644 > index 0000000000..1dd311a562 > --- /dev/null > +++ b/sysdeps/x86_64/fpu/test-double-libmvec-cosh-avx.c > @@ -0,0 +1 @@ > +#include "test-double-libmvec-cosh.c" > diff --git a/sysdeps/x86_64/fpu/test-double-libmvec-cosh-avx2.c b/sysdeps/x86_64/fpu/test-double-libmvec-cosh-avx2.c > new file mode 100644 > index 0000000000..1dd311a562 > --- /dev/null > +++ b/sysdeps/x86_64/fpu/test-double-libmvec-cosh-avx2.c > @@ -0,0 +1 @@ > +#include "test-double-libmvec-cosh.c" > diff --git a/sysdeps/x86_64/fpu/test-double-libmvec-cosh-avx512f.c b/sysdeps/x86_64/fpu/test-double-libmvec-cosh-avx512f.c > new file mode 100644 > index 0000000000..1dd311a562 > --- /dev/null > +++ b/sysdeps/x86_64/fpu/test-double-libmvec-cosh-avx512f.c > @@ -0,0 +1 @@ > +#include "test-double-libmvec-cosh.c" > diff --git a/sysdeps/x86_64/fpu/test-double-libmvec-cosh.c b/sysdeps/x86_64/fpu/test-double-libmvec-cosh.c > new file mode 100644 > index 0000000000..cf49ec5d87 > --- /dev/null > +++ b/sysdeps/x86_64/fpu/test-double-libmvec-cosh.c > @@ -0,0 +1,3 @@ > +#define LIBMVEC_TYPE double > +#define LIBMVEC_FUNC cosh > +#include "test-vector-abi-arg1.h" > diff --git a/sysdeps/x86_64/fpu/test-double-vlen2-wrappers.c b/sysdeps/x86_64/fpu/test-double-vlen2-wrappers.c > index 256e8f07c9..68c449e04a 100644 > --- a/sysdeps/x86_64/fpu/test-double-vlen2-wrappers.c > +++ b/sysdeps/x86_64/fpu/test-double-vlen2-wrappers.c > @@ -33,6 +33,7 @@ VECTOR_WRAPPER (WRAPPER_NAME (asin), _ZGVbN2v_asin) > VECTOR_WRAPPER_ff (WRAPPER_NAME (hypot), _ZGVbN2vv_hypot) > VECTOR_WRAPPER (WRAPPER_NAME (exp2), _ZGVbN2v_exp2) > VECTOR_WRAPPER (WRAPPER_NAME (exp10), _ZGVbN2v_exp10) > +VECTOR_WRAPPER (WRAPPER_NAME (cosh), _ZGVbN2v_cosh) > > #define VEC_INT_TYPE __m128i > > diff --git a/sysdeps/x86_64/fpu/test-double-vlen4-avx2-wrappers.c b/sysdeps/x86_64/fpu/test-double-vlen4-avx2-wrappers.c > index 9de1dab2c2..df67306373 100644 > --- a/sysdeps/x86_64/fpu/test-double-vlen4-avx2-wrappers.c > +++ b/sysdeps/x86_64/fpu/test-double-vlen4-avx2-wrappers.c > @@ -36,6 +36,7 @@ VECTOR_WRAPPER (WRAPPER_NAME (asin), _ZGVdN4v_asin) > VECTOR_WRAPPER_ff (WRAPPER_NAME (hypot), _ZGVdN4vv_hypot) > VECTOR_WRAPPER (WRAPPER_NAME (exp2), _ZGVdN4v_exp2) > VECTOR_WRAPPER (WRAPPER_NAME (exp10), _ZGVdN4v_exp10) > +VECTOR_WRAPPER (WRAPPER_NAME (cosh), _ZGVdN4v_cosh) > > #ifndef __ILP32__ > # define VEC_INT_TYPE __m256i > diff --git a/sysdeps/x86_64/fpu/test-double-vlen4-wrappers.c b/sysdeps/x86_64/fpu/test-double-vlen4-wrappers.c > index 43865ab099..1a6731098f 100644 > --- a/sysdeps/x86_64/fpu/test-double-vlen4-wrappers.c > +++ b/sysdeps/x86_64/fpu/test-double-vlen4-wrappers.c > @@ -33,6 +33,7 @@ VECTOR_WRAPPER (WRAPPER_NAME (asin), _ZGVcN4v_asin) > VECTOR_WRAPPER_ff (WRAPPER_NAME (hypot), _ZGVcN4vv_hypot) > VECTOR_WRAPPER (WRAPPER_NAME (exp2), _ZGVcN4v_exp2) > VECTOR_WRAPPER (WRAPPER_NAME (exp10), _ZGVcN4v_exp10) > +VECTOR_WRAPPER (WRAPPER_NAME (cosh), _ZGVcN4v_cosh) > > #define VEC_INT_TYPE __m128i > > diff --git a/sysdeps/x86_64/fpu/test-double-vlen8-wrappers.c b/sysdeps/x86_64/fpu/test-double-vlen8-wrappers.c > index 5dbdacf617..4cdfa918e8 100644 > --- a/sysdeps/x86_64/fpu/test-double-vlen8-wrappers.c > +++ b/sysdeps/x86_64/fpu/test-double-vlen8-wrappers.c > @@ -33,6 +33,7 @@ VECTOR_WRAPPER (WRAPPER_NAME (asin), _ZGVeN8v_asin) > VECTOR_WRAPPER_ff (WRAPPER_NAME (hypot), _ZGVeN8vv_hypot) > VECTOR_WRAPPER (WRAPPER_NAME (exp2), _ZGVeN8v_exp2) > VECTOR_WRAPPER (WRAPPER_NAME (exp10), _ZGVeN8v_exp10) > +VECTOR_WRAPPER (WRAPPER_NAME (cosh), _ZGVeN8v_cosh) > > #ifndef __ILP32__ > # define VEC_INT_TYPE __m512i > diff --git a/sysdeps/x86_64/fpu/test-float-libmvec-coshf-avx.c b/sysdeps/x86_64/fpu/test-float-libmvec-coshf-avx.c > new file mode 100644 > index 0000000000..905dc3ca4a > --- /dev/null > +++ b/sysdeps/x86_64/fpu/test-float-libmvec-coshf-avx.c > @@ -0,0 +1 @@ > +#include "test-float-libmvec-coshf.c" > diff --git a/sysdeps/x86_64/fpu/test-float-libmvec-coshf-avx2.c b/sysdeps/x86_64/fpu/test-float-libmvec-coshf-avx2.c > new file mode 100644 > index 0000000000..905dc3ca4a > --- /dev/null > +++ b/sysdeps/x86_64/fpu/test-float-libmvec-coshf-avx2.c > @@ -0,0 +1 @@ > +#include "test-float-libmvec-coshf.c" > diff --git a/sysdeps/x86_64/fpu/test-float-libmvec-coshf-avx512f.c b/sysdeps/x86_64/fpu/test-float-libmvec-coshf-avx512f.c > new file mode 100644 > index 0000000000..905dc3ca4a > --- /dev/null > +++ b/sysdeps/x86_64/fpu/test-float-libmvec-coshf-avx512f.c > @@ -0,0 +1 @@ > +#include "test-float-libmvec-coshf.c" > diff --git a/sysdeps/x86_64/fpu/test-float-libmvec-coshf.c b/sysdeps/x86_64/fpu/test-float-libmvec-coshf.c > new file mode 100644 > index 0000000000..94b899076b > --- /dev/null > +++ b/sysdeps/x86_64/fpu/test-float-libmvec-coshf.c > @@ -0,0 +1,3 @@ > +#define LIBMVEC_TYPE float > +#define LIBMVEC_FUNC coshf > +#include "test-vector-abi-arg1.h" > diff --git a/sysdeps/x86_64/fpu/test-float-vlen16-wrappers.c b/sysdeps/x86_64/fpu/test-float-vlen16-wrappers.c > index c159c8f583..47a9862233 100644 > --- a/sysdeps/x86_64/fpu/test-float-vlen16-wrappers.c > +++ b/sysdeps/x86_64/fpu/test-float-vlen16-wrappers.c > @@ -33,6 +33,7 @@ VECTOR_WRAPPER (WRAPPER_NAME (asinf), _ZGVeN16v_asinf) > VECTOR_WRAPPER_ff (WRAPPER_NAME (hypotf), _ZGVeN16vv_hypotf) > VECTOR_WRAPPER (WRAPPER_NAME (exp2f), _ZGVeN16v_exp2f) > VECTOR_WRAPPER (WRAPPER_NAME (exp10f), _ZGVeN16v_exp10f) > +VECTOR_WRAPPER (WRAPPER_NAME (coshf), _ZGVeN16v_coshf) > > #define VEC_INT_TYPE __m512i > > diff --git a/sysdeps/x86_64/fpu/test-float-vlen4-wrappers.c b/sysdeps/x86_64/fpu/test-float-vlen4-wrappers.c > index c745ef744a..e7c5410e7b 100644 > --- a/sysdeps/x86_64/fpu/test-float-vlen4-wrappers.c > +++ b/sysdeps/x86_64/fpu/test-float-vlen4-wrappers.c > @@ -33,6 +33,7 @@ VECTOR_WRAPPER (WRAPPER_NAME (asinf), _ZGVbN4v_asinf) > VECTOR_WRAPPER_ff (WRAPPER_NAME (hypotf), _ZGVbN4vv_hypotf) > VECTOR_WRAPPER (WRAPPER_NAME (exp2f), _ZGVbN4v_exp2f) > VECTOR_WRAPPER (WRAPPER_NAME (exp10f), _ZGVbN4v_exp10f) > +VECTOR_WRAPPER (WRAPPER_NAME (coshf), _ZGVbN4v_coshf) > > #define VEC_INT_TYPE __m128i > > diff --git a/sysdeps/x86_64/fpu/test-float-vlen8-avx2-wrappers.c b/sysdeps/x86_64/fpu/test-float-vlen8-avx2-wrappers.c > index c9226cf4dc..b8e9d48cd6 100644 > --- a/sysdeps/x86_64/fpu/test-float-vlen8-avx2-wrappers.c > +++ b/sysdeps/x86_64/fpu/test-float-vlen8-avx2-wrappers.c > @@ -36,6 +36,7 @@ VECTOR_WRAPPER (WRAPPER_NAME (asinf), _ZGVdN8v_asinf) > VECTOR_WRAPPER_ff (WRAPPER_NAME (hypotf), _ZGVdN8vv_hypotf) > VECTOR_WRAPPER (WRAPPER_NAME (exp2f), _ZGVdN8v_exp2f) > VECTOR_WRAPPER (WRAPPER_NAME (exp10f), _ZGVdN8v_exp10f) > +VECTOR_WRAPPER (WRAPPER_NAME (coshf), _ZGVdN8v_coshf) > > /* Redefinition of wrapper to be compatible with _ZGVdN8vvv_sincosf. */ > #undef VECTOR_WRAPPER_fFF > diff --git a/sysdeps/x86_64/fpu/test-float-vlen8-wrappers.c b/sysdeps/x86_64/fpu/test-float-vlen8-wrappers.c > index 92970c5ace..328c827b27 100644 > --- a/sysdeps/x86_64/fpu/test-float-vlen8-wrappers.c > +++ b/sysdeps/x86_64/fpu/test-float-vlen8-wrappers.c > @@ -33,6 +33,7 @@ VECTOR_WRAPPER (WRAPPER_NAME (asinf), _ZGVcN8v_asinf) > VECTOR_WRAPPER_ff (WRAPPER_NAME (hypotf), _ZGVcN8vv_hypotf) > VECTOR_WRAPPER (WRAPPER_NAME (exp2f), _ZGVcN8v_exp2f) > VECTOR_WRAPPER (WRAPPER_NAME (exp10f), _ZGVcN8v_exp10f) > +VECTOR_WRAPPER (WRAPPER_NAME (coshf), _ZGVcN8v_coshf) > > #define VEC_INT_TYPE __m128i > > -- > 2.31.1 > LGTM. Reviewed-by: H.J. Lu Thanks. H.J.