From c10b9b13f7471b08273effc8cd7e51b119df9348 Mon Sep 17 00:00:00 2001 From: Andrew Senkevich Date: Wed, 17 Jun 2015 16:22:26 +0300 Subject: Vector pow for x86_64 and tests. Here is implementation of vectorized pow containing SSE, AVX, AVX2 and AVX512 versions according to Vector ABI . * bits/libm-simd-decl-stubs.h: Added stubs for pow. * math/bits/mathcalls.h: Added pow declaration with __MATHCALL_VEC. * sysdeps/unix/sysv/linux/x86_64/libmvec.abilist: New versions added. * sysdeps/x86/fpu/bits/math-vector.h: Added SIMD declaration and asm redirections for pow. * sysdeps/x86_64/fpu/Makefile (libmvec-support): Added new files. * sysdeps/x86_64/fpu/Versions: New versions added. * sysdeps/x86_64/fpu/libm-test-ulps: Regenerated. * sysdeps/x86_64/fpu/multiarch/Makefile (libmvec-sysdep_routines): Added build of SSE, AVX2 and AVX512 IFUNC versions. * sysdeps/x86_64/fpu/svml_d_wrapper_impl.h: Added 2 argument wrappers. * sysdeps/x86_64/fpu/multiarch/svml_d_pow2_core.S: New file. * sysdeps/x86_64/fpu/multiarch/svml_d_pow2_core_sse4.S: New file. * sysdeps/x86_64/fpu/multiarch/svml_d_pow4_core.S: New file. * sysdeps/x86_64/fpu/multiarch/svml_d_pow4_core_avx2.S: New file. * sysdeps/x86_64/fpu/multiarch/svml_d_pow8_core.S: New file. * sysdeps/x86_64/fpu/multiarch/svml_d_pow8_core_avx512.S: New file. * sysdeps/x86_64/fpu/svml_d_pow2_core.S: New file. * sysdeps/x86_64/fpu/svml_d_pow4_core.S: New file. * sysdeps/x86_64/fpu/svml_d_pow4_core_avx.S: New file. * sysdeps/x86_64/fpu/svml_d_pow8_core.S: New file. * sysdeps/x86_64/fpu/svml_d_pow_data.S: New file. * sysdeps/x86_64/fpu/svml_d_pow_data.h: New file. * sysdeps/x86_64/fpu/test-double-vlen2-wrappers.c: Added vector pow test. * sysdeps/x86_64/fpu/test-double-vlen2.c: Likewise. * sysdeps/x86_64/fpu/test-double-vlen4-avx2-wrappers.c: Likewise. * sysdeps/x86_64/fpu/test-double-vlen4-avx2.c: Likewise. * sysdeps/x86_64/fpu/test-double-vlen4-wrappers.c: Likewise. * sysdeps/x86_64/fpu/test-double-vlen4.c: Likewise. * sysdeps/x86_64/fpu/test-double-vlen8-wrappers.c: Likewise. * sysdeps/x86_64/fpu/test-double-vlen8.c: Likewise. * NEWS: Mention addition of x86_64 vector pow. --- sysdeps/x86/fpu/bits/math-vector.h | 6 ++++++ 1 file changed, 6 insertions(+) (limited to 'sysdeps/x86/fpu/bits/math-vector.h') diff --git a/sysdeps/x86/fpu/bits/math-vector.h b/sysdeps/x86/fpu/bits/math-vector.h index 3b7158952c..a5317b9e32 100644 --- a/sysdeps/x86/fpu/bits/math-vector.h +++ b/sysdeps/x86/fpu/bits/math-vector.h @@ -44,6 +44,8 @@ # define __DECL_SIMD_exp __DECL_SIMD_x86_64 # undef __DECL_SIMD_expf # define __DECL_SIMD_expf __DECL_SIMD_x86_64 +# undef __DECL_SIMD_pow +# define __DECL_SIMD_pow __DECL_SIMD_x86_64 /* Workaround to exclude unnecessary symbol aliases in libmvec while GCC creates the vector names based on scalar asm name. @@ -65,6 +67,10 @@ __asm__ ("_ZGVbN4v___expf_finite = _ZGVbN4v_expf"); __asm__ ("_ZGVcN8v___expf_finite = _ZGVcN8v_expf"); __asm__ ("_ZGVdN8v___expf_finite = _ZGVdN8v_expf"); __asm__ ("_ZGVeN16v___expf_finite = _ZGVeN16v_expf"); +__asm__ ("_ZGVbN2vv___pow_finite = _ZGVbN2vv_pow"); +__asm__ ("_ZGVcN4vv___pow_finite = _ZGVcN4vv_pow"); +__asm__ ("_ZGVdN4vv___pow_finite = _ZGVdN4vv_pow"); +__asm__ ("_ZGVeN8vv___pow_finite = _ZGVeN8vv_pow"); # endif #endif -- cgit 1.4.1