From 04f496d6025753058bdd071fd711e9f56df149a7 Mon Sep 17 00:00:00 2001 From: Andrew Senkevich Date: Tue, 9 Jun 2015 18:29:47 +0300 Subject: Vector cosf for x86_64. Here is implementation of vectorized cosf containing SSE, AVX, AVX2 and AVX512 versions according to Vector ABI . * sysdeps/x86_64/fpu/Makefile (libmvec-support): Added new files. * sysdeps/x86_64/fpu/Versions: New versions added. * sysdeps/x86_64/fpu/svml_s_cosf4_core.S: New file. * sysdeps/x86_64/fpu/multiarch/svml_s_cosf4_core.S: New file. * sysdeps/x86_64/fpu/multiarch/svml_s_cosf4_core_sse4.S: New file. * sysdeps/x86_64/fpu/svml_s_cosf8_core_avx.S: New file. * sysdeps/x86_64/fpu/svml_s_cosf8_core.S: New file. * sysdeps/x86_64/fpu/multiarch/svml_s_cosf8_core.S: New file. * sysdeps/x86_64/fpu/multiarch/svml_s_cosf8_core_avx2.S: New file. * sysdeps/x86_64/fpu/svml_s_cosf16_core.S: New file. * sysdeps/x86_64/fpu/multiarch/svml_s_cosf16_core.S: New file. * sysdeps/x86_64/fpu/multiarch/svml_s_cosf16_core_avx512.S: New file. * sysdeps/x86_64/fpu/svml_s_wrapper_impl.h: New file. * sysdeps/x86_64/fpu/svml_s_cosf_data.S: New file. * sysdeps/x86_64/fpu/svml_s_cosf_data.h: New file. * sysdeps/x86_64/fpu/multiarch/Makefile (libmvec-sysdep_routines): Added build of SSE, AVX2 and AVX512 IFUNC versions. * sysdeps/unix/sysv/linux/x86_64/libmvec.abilist: New versions added. * sysdeps/x86/fpu/bits/math-vector.h: Added SIMD declaration for cosf. * NEWS: Mention addition of x86_64 vector cosf. --- sysdeps/x86_64/fpu/multiarch/Makefile | 3 +- sysdeps/x86_64/fpu/multiarch/svml_s_cosf16_core.S | 39 ++ .../fpu/multiarch/svml_s_cosf16_core_avx512.S | 460 +++++++++++++++++++++ sysdeps/x86_64/fpu/multiarch/svml_s_cosf4_core.S | 38 ++ .../x86_64/fpu/multiarch/svml_s_cosf4_core_sse4.S | 227 ++++++++++ sysdeps/x86_64/fpu/multiarch/svml_s_cosf8_core.S | 38 ++ .../x86_64/fpu/multiarch/svml_s_cosf8_core_avx2.S | 215 ++++++++++ 7 files changed, 1019 insertions(+), 1 deletion(-) create mode 100644 sysdeps/x86_64/fpu/multiarch/svml_s_cosf16_core.S create mode 100644 sysdeps/x86_64/fpu/multiarch/svml_s_cosf16_core_avx512.S create mode 100644 sysdeps/x86_64/fpu/multiarch/svml_s_cosf4_core.S create mode 100644 sysdeps/x86_64/fpu/multiarch/svml_s_cosf4_core_sse4.S create mode 100644 sysdeps/x86_64/fpu/multiarch/svml_s_cosf8_core.S create mode 100644 sysdeps/x86_64/fpu/multiarch/svml_s_cosf8_core_avx2.S (limited to 'sysdeps/x86_64/fpu/multiarch') diff --git a/sysdeps/x86_64/fpu/multiarch/Makefile b/sysdeps/x86_64/fpu/multiarch/Makefile index b2f3266..6b50475 100644 --- a/sysdeps/x86_64/fpu/multiarch/Makefile +++ b/sysdeps/x86_64/fpu/multiarch/Makefile @@ -54,5 +54,6 @@ endif ifeq ($(subdir),mathvec) libmvec-sysdep_routines += svml_d_cos2_core_sse4 svml_d_cos4_core_avx2 \ - svml_d_cos8_core_avx512 + svml_d_cos8_core_avx512 svml_s_cosf4_core_sse4 \ + svml_s_cosf8_core_avx2 svml_s_cosf16_core_avx512 endif diff --git a/sysdeps/x86_64/fpu/multiarch/svml_s_cosf16_core.S b/sysdeps/x86_64/fpu/multiarch/svml_s_cosf16_core.S new file mode 100644 index 0000000..91564de --- /dev/null +++ b/sysdeps/x86_64/fpu/multiarch/svml_s_cosf16_core.S @@ -0,0 +1,39 @@ +/* Multiple versions of vectorized cosf. + Copyright (C) 2014-2015 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 + + .text +ENTRY (_ZGVeN16v_cosf) + .type _ZGVeN16v_cosf, @gnu_indirect_function + cmpl $0, KIND_OFFSET+__cpu_features(%rip) + jne 1 + call __init_cpu_features +1: leaq _ZGVeN16v_cosf_skx(%rip), %rax + testl $bit_AVX512DQ_Usable, __cpu_features+FEATURE_OFFSET+index_AVX512DQ_Usable(%rip) + jnz 3 +2: leaq _ZGVeN16v_cosf_knl(%rip), %rax + testl $bit_AVX512F_Usable, __cpu_features+FEATURE_OFFSET+index_AVX512F_Usable(%rip) + jnz 3 + leaq _ZGVeN16v_cosf_avx2_wrapper(%rip), %rax +3: ret +END (_ZGVeN16v_cosf) + +#define _ZGVeN16v_cosf _ZGVeN16v_cosf_avx2_wrapper +#include "../svml_s_cosf16_core.S" diff --git a/sysdeps/x86_64/fpu/multiarch/svml_s_cosf16_core_avx512.S b/sysdeps/x86_64/fpu/multiarch/svml_s_cosf16_core_avx512.S new file mode 100644 index 0000000..a78ae2e --- /dev/null +++ b/sysdeps/x86_64/fpu/multiarch/svml_s_cosf16_core_avx512.S @@ -0,0 +1,460 @@ +/* Function cosf vectorized with AVX-512. KNL and SKX versions. + Copyright (C) 2014-2015 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_cosf_data.h" +#include "svml_s_wrapper_impl.h" + + .text +ENTRY (_ZGVeN16v_cosf_knl) +#ifndef HAVE_AVX512_ASM_SUPPORT +WRAPPER_IMPL_AVX512 _ZGVdN8v_cosf +#else +/* + ALGORITHM DESCRIPTION: + + 1) Range reduction to [-Pi/2; +Pi/2] interval + a) We remove sign using AND operation + b) Add Pi/2 value to argument X for Cos to Sin transformation + c) Getting octant Y by 1/Pi multiplication + d) Add "Right Shifter" value + e) Treat obtained value as integer for destination sign setting. + Shift first bit of this value to the last (sign) position + f) Subtract "Right Shifter" value + g) Subtract 0.5 from result for octant correction + h) Subtract Y*PI from X argument, where PI divided to 4 parts: + X = X - Y*PI1 - Y*PI2 - Y*PI3 - Y*PI4; + 2) Polynomial (minimax for sin within [-Pi/2; +Pi/2] interval) + a) Calculate X^2 = X * X + b) Calculate polynomial: + R = X + X * X^2 * (A3 + x^2 * (A5 + ..... + 3) Destination sign setting + a) Set shifted destination sign using XOR operation: + R = XOR( R, S ); + */ + pushq %rbp + cfi_adjust_cfa_offset (8) + cfi_rel_offset (%rbp, 0) + movq %rsp, %rbp + cfi_def_cfa_register (%rbp) + andq $-64, %rsp + subq $1280, %rsp + movq __svml_scos_data@GOTPCREL(%rip), %rdx + +/* + h) Subtract Y*PI from X argument, where PI divided to 4 parts: + X = X - Y*PI1 - Y*PI2 - Y*PI3 + */ + vmovaps %zmm0, %zmm6 + movl $-1, %eax + +/* b) Add Pi/2 value to argument X for Cos to Sin transformation */ + vaddps __sHalfPI(%rdx), %zmm0, %zmm2 + vmovups __sRShifter(%rdx), %zmm3 + +/* + 1) Range reduction to [-Pi/2; +Pi/2] interval + c) Getting octant Y by 1/Pi multiplication + d) Add "Right Shifter" (0x4B000000) value + */ + vfmadd132ps __sInvPI(%rdx), %zmm3, %zmm2 + vmovups __sPI1_FMA(%rdx), %zmm5 + +/* f) Subtract "Right Shifter" (0x4B000000) value */ + vsubps %zmm3, %zmm2, %zmm4 + vmovups __sA9_FMA(%rdx), %zmm9 + +/* Check for large and special arguments */ + vpandd __sAbsMask(%rdx), %zmm0, %zmm1 + +/* + e) Treat obtained value as integer for destination sign setting. + Shift first bit of this value to the last (sign) position (S << 31) + */ + vpslld $31, %zmm2, %zmm8 + vcmpps $22, __sRangeReductionVal(%rdx), %zmm1, %k1 + vpbroadcastd %eax, %zmm12{%k1}{z} + +/* g) Subtract 0.5 from result for octant correction */ + vsubps __sOneHalf(%rdx), %zmm4, %zmm7 + vptestmd %zmm12, %zmm12, %k0 + vfnmadd231ps %zmm7, %zmm5, %zmm6 + kmovw %k0, %ecx + vfnmadd231ps __sPI2_FMA(%rdx), %zmm7, %zmm6 + vfnmadd132ps __sPI3_FMA(%rdx), %zmm6, %zmm7 + +/* a) Calculate X^2 = X * X */ + vmulps %zmm7, %zmm7, %zmm10 + +/* + 3) Destination sign setting + a) Set shifted destination sign using XOR operation: + R = XOR( R, S ); + */ + vpxord %zmm8, %zmm7, %zmm11 + +/* + b) Calculate polynomial: + R = X + X * X^2 * (A3 + x^2 * (A5 + x^2 * (A7 + x^2 * (A9)))); + */ + vfmadd213ps __sA7_FMA(%rdx), %zmm10, %zmm9 + vfmadd213ps __sA5_FMA(%rdx), %zmm10, %zmm9 + vfmadd213ps __sA3(%rdx), %zmm10, %zmm9 + vmulps %zmm10, %zmm9, %zmm1 + vfmadd213ps %zmm11, %zmm11, %zmm1 + testl %ecx, %ecx + jne .LBL_1_3 + +.LBL_1_2: + cfi_remember_state + vmovaps %zmm1, %zmm0 + movq %rbp, %rsp + cfi_def_cfa_register (%rsp) + popq %rbp + cfi_adjust_cfa_offset (-8) + cfi_restore (%rbp) + ret + +.LBL_1_3: + cfi_restore_state + vmovups %zmm0, 1152(%rsp) + vmovups %zmm1, 1216(%rsp) + je .LBL_1_2 + + xorb %dl, %dl + kmovw %k4, 1048(%rsp) + xorl %eax, %eax + kmovw %k5, 1040(%rsp) + kmovw %k6, 1032(%rsp) + kmovw %k7, 1024(%rsp) + vmovups %zmm16, 960(%rsp) + vmovups %zmm17, 896(%rsp) + vmovups %zmm18, 832(%rsp) + vmovups %zmm19, 768(%rsp) + vmovups %zmm20, 704(%rsp) + vmovups %zmm21, 640(%rsp) + vmovups %zmm22, 576(%rsp) + vmovups %zmm23, 512(%rsp) + vmovups %zmm24, 448(%rsp) + vmovups %zmm25, 384(%rsp) + vmovups %zmm26, 320(%rsp) + vmovups %zmm27, 256(%rsp) + vmovups %zmm28, 192(%rsp) + vmovups %zmm29, 128(%rsp) + vmovups %zmm30, 64(%rsp) + vmovups %zmm31, (%rsp) + movq %rsi, 1064(%rsp) + movq %rdi, 1056(%rsp) + movq %r12, 1096(%rsp) + cfi_offset_rel_rsp (12, 1096) + movb %dl, %r12b + movq %r13, 1088(%rsp) + cfi_offset_rel_rsp (13, 1088) + movl %ecx, %r13d + movq %r14, 1080(%rsp) + cfi_offset_rel_rsp (14, 1080) + movl %eax, %r14d + movq %r15, 1072(%rsp) + cfi_offset_rel_rsp (15, 1072) + cfi_remember_state + +.LBL_1_6: + btl %r14d, %r13d + jc .LBL_1_12 + +.LBL_1_7: + lea 1(%r14), %esi + btl %esi, %r13d + jc .LBL_1_10 + +.LBL_1_8: + addb $1, %r12b + addl $2, %r14d + cmpb $16, %r12b + jb .LBL_1_6 + + kmovw 1048(%rsp), %k4 + movq 1064(%rsp), %rsi + kmovw 1040(%rsp), %k5 + movq 1056(%rsp), %rdi + kmovw 1032(%rsp), %k6 + movq 1096(%rsp), %r12 + cfi_restore (%r12) + movq 1088(%rsp), %r13 + cfi_restore (%r13) + kmovw 1024(%rsp), %k7 + vmovups 960(%rsp), %zmm16 + vmovups 896(%rsp), %zmm17 + vmovups 832(%rsp), %zmm18 + vmovups 768(%rsp), %zmm19 + vmovups 704(%rsp), %zmm20 + vmovups 640(%rsp), %zmm21 + vmovups 576(%rsp), %zmm22 + vmovups 512(%rsp), %zmm23 + vmovups 448(%rsp), %zmm24 + vmovups 384(%rsp), %zmm25 + vmovups 320(%rsp), %zmm26 + vmovups 256(%rsp), %zmm27 + vmovups 192(%rsp), %zmm28 + vmovups 128(%rsp), %zmm29 + vmovups 64(%rsp), %zmm30 + vmovups (%rsp), %zmm31 + movq 1080(%rsp), %r14 + cfi_restore (%r14) + movq 1072(%rsp), %r15 + cfi_restore (%r15) + vmovups 1216(%rsp), %zmm1 + jmp .LBL_1_2 + +.LBL_1_10: + cfi_restore_state + movzbl %r12b, %r15d + vmovss 1156(%rsp,%r15,8), %xmm0 + call cosf@PLT + vmovss %xmm0, 1220(%rsp,%r15,8) + jmp .LBL_1_8 + +.LBL_1_12: + movzbl %r12b, %r15d + vmovss 1152(%rsp,%r15,8), %xmm0 + call cosf@PLT + vmovss %xmm0, 1216(%rsp,%r15,8) + jmp .LBL_1_7 +#endif +END (_ZGVeN16v_cosf_knl) + +ENTRY (_ZGVeN16v_cosf_skx) +#ifndef HAVE_AVX512_ASM_SUPPORT +WRAPPER_IMPL_AVX512 _ZGVdN8v_cosf +#else +/* + ALGORITHM DESCRIPTION: + + 1) Range reduction to [-Pi/2; +Pi/2] interval + a) We remove sign using AND operation + b) Add Pi/2 value to argument X for Cos to Sin transformation + c) Getting octant Y by 1/Pi multiplication + d) Add "Right Shifter" value + e) Treat obtained value as integer for destination sign setting. + Shift first bit of this value to the last (sign) position + f) Subtract "Right Shifter" value + g) Subtract 0.5 from result for octant correction + h) Subtract Y*PI from X argument, where PI divided to 4 parts: + X = X - Y*PI1 - Y*PI2 - Y*PI3 - Y*PI4; + 2) Polynomial (minimax for sin within [-Pi/2; +Pi/2] interval) + a) Calculate X^2 = X * X + b) Calculate polynomial: + R = X + X * X^2 * (A3 + x^2 * (A5 + ..... + 3) Destination sign setting + a) Set shifted destination sign using XOR operation: + R = XOR( R, S ); + */ + pushq %rbp + cfi_adjust_cfa_offset (8) + cfi_rel_offset (%rbp, 0) + movq %rsp, %rbp + cfi_def_cfa_register (%rbp) + andq $-64, %rsp + subq $1280, %rsp + movq __svml_scos_data@GOTPCREL(%rip), %rax + +/* + h) Subtract Y*PI from X argument, where PI divided to 4 parts: + X = X - Y*PI1 - Y*PI2 - Y*PI3 + */ + vmovaps %zmm0, %zmm6 + vmovups .L_2il0floatpacket.13(%rip), %zmm12 + vmovups __sRShifter(%rax), %zmm3 + vmovups __sPI1_FMA(%rax), %zmm5 + vmovups __sA9_FMA(%rax), %zmm9 + +/* b) Add Pi/2 value to argument X for Cos to Sin transformation */ + vaddps __sHalfPI(%rax), %zmm0, %zmm2 + +/* Check for large and special arguments */ + vandps __sAbsMask(%rax), %zmm0, %zmm1 + +/* + 1) Range reduction to [-Pi/2; +Pi/2] interval + c) Getting octant Y by 1/Pi multiplication + d) Add "Right Shifter" (0x4B000000) value + */ + vfmadd132ps __sInvPI(%rax), %zmm3, %zmm2 + vcmpps $18, __sRangeReductionVal(%rax), %zmm1, %k1 + +/* + e) Treat obtained value as integer for destination sign setting. + Shift first bit of this value to the last (sign) position (S << 31) + */ + vpslld $31, %zmm2, %zmm8 + +/* f) Subtract "Right Shifter" (0x4B000000) value */ + vsubps %zmm3, %zmm2, %zmm4 + +/* g) Subtract 0.5 from result for octant correction */ + vsubps __sOneHalf(%rax), %zmm4, %zmm7 + vfnmadd231ps %zmm7, %zmm5, %zmm6 + vfnmadd231ps __sPI2_FMA(%rax), %zmm7, %zmm6 + vfnmadd132ps __sPI3_FMA(%rax), %zmm6, %zmm7 + +/* a) Calculate X^2 = X * X */ + vmulps %zmm7, %zmm7, %zmm10 + +/* + 3) Destination sign setting + a) Set shifted destination sign using XOR operation: + R = XOR( R, S ); + */ + vxorps %zmm8, %zmm7, %zmm11 + +/* + b) Calculate polynomial: + R = X + X * X^2 * (A3 + x^2 * (A5 + x^2 * (A7 + x^2 * (A9)))); + */ + vfmadd213ps __sA7_FMA(%rax), %zmm10, %zmm9 + vfmadd213ps __sA5_FMA(%rax), %zmm10, %zmm9 + vfmadd213ps __sA3(%rax), %zmm10, %zmm9 + vpandnd %zmm1, %zmm1, %zmm12{%k1} + vmulps %zmm10, %zmm9, %zmm1 + vptestmd %zmm12, %zmm12, %k0 + vfmadd213ps %zmm11, %zmm11, %zmm1 + kmovw %k0, %ecx + testl %ecx, %ecx + jne .LBL_2_3 +.LBL_2_2: + cfi_remember_state + vmovaps %zmm1, %zmm0 + movq %rbp, %rsp + cfi_def_cfa_register (%rsp) + popq %rbp + cfi_adjust_cfa_offset (-8) + cfi_restore (%rbp) + ret + +.LBL_2_3: + cfi_restore_state + vmovups %zmm0, 1152(%rsp) + vmovups %zmm1, 1216(%rsp) + je .LBL_2_2 + + xorb %dl, %dl + xorl %eax, %eax + kmovw %k4, 1048(%rsp) + kmovw %k5, 1040(%rsp) + kmovw %k6, 1032(%rsp) + kmovw %k7, 1024(%rsp) + vmovups %zmm16, 960(%rsp) + vmovups %zmm17, 896(%rsp) + vmovups %zmm18, 832(%rsp) + vmovups %zmm19, 768(%rsp) + vmovups %zmm20, 704(%rsp) + vmovups %zmm21, 640(%rsp) + vmovups %zmm22, 576(%rsp) + vmovups %zmm23, 512(%rsp) + vmovups %zmm24, 448(%rsp) + vmovups %zmm25, 384(%rsp) + vmovups %zmm26, 320(%rsp) + vmovups %zmm27, 256(%rsp) + vmovups %zmm28, 192(%rsp) + vmovups %zmm29, 128(%rsp) + vmovups %zmm30, 64(%rsp) + vmovups %zmm31, (%rsp) + movq %rsi, 1064(%rsp) + movq %rdi, 1056(%rsp) + movq %r12, 1096(%rsp) + cfi_offset_rel_rsp (12, 1096) + movb %dl, %r12b + movq %r13, 1088(%rsp) + cfi_offset_rel_rsp (13, 1088) + movl %ecx, %r13d + movq %r14, 1080(%rsp) + cfi_offset_rel_rsp (14, 1080) + movl %eax, %r14d + movq %r15, 1072(%rsp) + cfi_offset_rel_rsp (15, 1072) + cfi_remember_state + +.LBL_2_6: + btl %r14d, %r13d + jc .LBL_2_12 +.LBL_2_7: + lea 1(%r14), %esi + btl %esi, %r13d + jc .LBL_2_10 +.LBL_2_8: + incb %r12b + addl $2, %r14d + cmpb $16, %r12b + jb .LBL_2_6 + kmovw 1048(%rsp), %k4 + kmovw 1040(%rsp), %k5 + kmovw 1032(%rsp), %k6 + kmovw 1024(%rsp), %k7 + vmovups 960(%rsp), %zmm16 + vmovups 896(%rsp), %zmm17 + vmovups 832(%rsp), %zmm18 + vmovups 768(%rsp), %zmm19 + vmovups 704(%rsp), %zmm20 + vmovups 640(%rsp), %zmm21 + vmovups 576(%rsp), %zmm22 + vmovups 512(%rsp), %zmm23 + vmovups 448(%rsp), %zmm24 + vmovups 384(%rsp), %zmm25 + vmovups 320(%rsp), %zmm26 + vmovups 256(%rsp), %zmm27 + vmovups 192(%rsp), %zmm28 + vmovups 128(%rsp), %zmm29 + vmovups 64(%rsp), %zmm30 + vmovups (%rsp), %zmm31 + vmovups 1216(%rsp), %zmm1 + movq 1064(%rsp), %rsi + movq 1056(%rsp), %rdi + movq 1096(%rsp), %r12 + cfi_restore (%r12) + movq 1088(%rsp), %r13 + cfi_restore (%r13) + movq 1080(%rsp), %r14 + cfi_restore (%r14) + movq 1072(%rsp), %r15 + cfi_restore (%r15) + jmp .LBL_2_2 + +.LBL_2_10: + cfi_restore_state + movzbl %r12b, %r15d + vmovss 1156(%rsp,%r15,8), %xmm0 + vzeroupper + vmovss 1156(%rsp,%r15,8), %xmm0 + call cosf@PLT + vmovss %xmm0, 1220(%rsp,%r15,8) + jmp .LBL_2_8 +.LBL_2_12: + movzbl %r12b, %r15d + vmovss 1152(%rsp,%r15,8), %xmm0 + vzeroupper + vmovss 1152(%rsp,%r15,8), %xmm0 + call cosf@PLT + vmovss %xmm0, 1216(%rsp,%r15,8) + jmp .LBL_2_7 +#endif +END (_ZGVeN16v_cosf_skx) + + .section .rodata, "a" +.L_2il0floatpacket.13: + .long 0xffffffff,0xffffffff,0xffffffff,0xffffffff,0xffffffff,0xffffffff,0xffffffff,0xffffffff,0xffffffff,0xffffffff,0xffffffff,0xffffffff,0xffffffff,0xffffffff,0xffffffff,0xffffffff + .type .L_2il0floatpacket.13,@object diff --git a/sysdeps/x86_64/fpu/multiarch/svml_s_cosf4_core.S b/sysdeps/x86_64/fpu/multiarch/svml_s_cosf4_core.S new file mode 100644 index 0000000..fa2363b --- /dev/null +++ b/sysdeps/x86_64/fpu/multiarch/svml_s_cosf4_core.S @@ -0,0 +1,38 @@ +/* Multiple versions of vectorized cosf, vector length is 4. + Copyright (C) 2014-2015 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 + + .text +ENTRY (_ZGVbN4v_cosf) + .type _ZGVbN4v_cosf, @gnu_indirect_function + cmpl $0, KIND_OFFSET+__cpu_features(%rip) + jne 1f + call __init_cpu_features +1: leaq _ZGVbN4v_cosf_sse4(%rip), %rax + testl $bit_SSE4_1, __cpu_features+CPUID_OFFSET+index_SSE4_1(%rip) + jz 2f + ret +2: leaq _ZGVbN4v_cosf_sse2(%rip), %rax + ret +END (_ZGVbN4v_cosf) +libmvec_hidden_def (_ZGVbN4v_cosf) + +#define _ZGVbN4v_cosf _ZGVbN4v_cosf_sse2 +#include "../svml_s_cosf4_core.S" diff --git a/sysdeps/x86_64/fpu/multiarch/svml_s_cosf4_core_sse4.S b/sysdeps/x86_64/fpu/multiarch/svml_s_cosf4_core_sse4.S new file mode 100644 index 0000000..f231ba2 --- /dev/null +++ b/sysdeps/x86_64/fpu/multiarch/svml_s_cosf4_core_sse4.S @@ -0,0 +1,227 @@ +/* Function cosf vectorized with SSE4. + Copyright (C) 2014-2015 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_cosf_data.h" + + .text +ENTRY (_ZGVbN4v_cosf_sse4) +/* + ALGORITHM DESCRIPTION: + + 1) Range reduction to [-Pi/2; +Pi/2] interval + a) We remove sign using AND operation + b) Add Pi/2 value to argument X for Cos to Sin transformation + c) Getting octant Y by 1/Pi multiplication + d) Add "Right Shifter" value + e) Treat obtained value as integer for destination sign setting. + Shift first bit of this value to the last (sign) position + f) Subtract "Right Shifter" value + g) Subtract 0.5 from result for octant correction + h) Subtract Y*PI from X argument, where PI divided to 4 parts: + X = X - Y*PI1 - Y*PI2 - Y*PI3 - Y*PI4; + 2) Polynomial (minimax for sin within [-Pi/2; +Pi/2] interval) + a) Calculate X^2 = X * X + b) Calculate polynomial: + R = X + X * X^2 * (A3 + x^2 * (A5 + ..... + 3) Destination sign setting + a) Set shifted destination sign using XOR operation: + R = XOR( R, S ); + */ + pushq %rbp + cfi_adjust_cfa_offset (8) + cfi_rel_offset (%rbp, 0) + movq %rsp, %rbp + cfi_def_cfa_register (%rbp) + andq $-64, %rsp + subq $320, %rsp + movaps %xmm0, %xmm4 + movq __svml_scos_data@GOTPCREL(%rip), %rax + movups __sHalfPI(%rax), %xmm1 + movups __sRShifter(%rax), %xmm5 + +/* b) Add Pi/2 value to argument X for Cos to Sin transformation */ + addps %xmm4, %xmm1 + +/* + 1) Range reduction to [-Pi/2; +Pi/2] interval + c) Getting octant Y by 1/Pi multiplication + d) Add "Right Shifter" (0x4B000000) value + */ + mulps __sInvPI(%rax), %xmm1 + movups __sPI1(%rax), %xmm6 + addps %xmm5, %xmm1 + +/* + e) Treat obtained value as integer for destination sign setting. + Shift first bit of this value to the last (sign) position (S << 31) + */ + movaps %xmm1, %xmm2 + +/* f) Subtract "Right Shifter" (0x4B000000) value */ + subps %xmm5, %xmm1 + movups __sPI2(%rax), %xmm7 + pslld $31, %xmm2 + movups __sPI3(%rax), %xmm5 + movups __sAbsMask(%rax), %xmm3 + +/* Check for large and special arguments */ + andps %xmm4, %xmm3 + +/* g) Subtract 0.5 from result for octant correction */ + subps __sOneHalf(%rax), %xmm1 + cmpnleps __sRangeReductionVal(%rax), %xmm3 + +/* + h) Subtract Y*PI from X argument, where PI divided to 4 parts: + X = X - Y*PI1 - Y*PI2 - Y*PI3 - Y*PI4; + */ + mulps %xmm1, %xmm6 + mulps %xmm1, %xmm7 + mulps %xmm1, %xmm5 + subps %xmm6, %xmm0 + movmskps %xmm3, %ecx + movups __sPI4(%rax), %xmm6 + subps %xmm7, %xmm0 + mulps %xmm6, %xmm1 + subps %xmm5, %xmm0 + subps %xmm1, %xmm0 + +/* a) Calculate X^2 = X * X */ + movaps %xmm0, %xmm1 + mulps %xmm0, %xmm1 + +/* + 3) Destination sign setting + a) Set shifted destination sign using XOR operation: + R = XOR( R, S ); + */ + xorps %xmm2, %xmm0 + movups __sA9(%rax), %xmm2 + +/* + b) Calculate polynomial: + R = X + X * X^2 * (A3 + x^2 * (A5 + x^2 * (A7 + x^2 * (A9)))); + */ + mulps %xmm1, %xmm2 + addps __sA7(%rax), %xmm2 + mulps %xmm1, %xmm2 + addps __sA5(%rax), %xmm2 + mulps %xmm1, %xmm2 + addps __sA3(%rax), %xmm2 + mulps %xmm2, %xmm1 + mulps %xmm0, %xmm1 + addps %xmm1, %xmm0 + testl %ecx, %ecx + jne .LBL_1_3 + +.LBL_1_2: + cfi_remember_state + movq %rbp, %rsp + cfi_def_cfa_register (%rsp) + popq %rbp + cfi_adjust_cfa_offset (-8) + cfi_restore (%rbp) + ret + +.LBL_1_3: + cfi_restore_state + movups %xmm4, 192(%rsp) + movups %xmm0, 256(%rsp) + je .LBL_1_2 + + xorb %dl, %dl + xorl %eax, %eax + movups %xmm8, 112(%rsp) + movups %xmm9, 96(%rsp) + movups %xmm10, 80(%rsp) + movups %xmm11, 64(%rsp) + movups %xmm12, 48(%rsp) + movups %xmm13, 32(%rsp) + movups %xmm14, 16(%rsp) + movups %xmm15, (%rsp) + movq %rsi, 136(%rsp) + movq %rdi, 128(%rsp) + movq %r12, 168(%rsp) + cfi_offset_rel_rsp (12, 168) + movb %dl, %r12b + movq %r13, 160(%rsp) + cfi_offset_rel_rsp (13, 160) + movl %ecx, %r13d + movq %r14, 152(%rsp) + cfi_offset_rel_rsp (14, 152) + movl %eax, %r14d + movq %r15, 144(%rsp) + cfi_offset_rel_rsp (15, 144) + cfi_remember_state + +.LBL_1_6: + btl %r14d, %r13d + jc .LBL_1_12 + +.LBL_1_7: + lea 1(%r14), %esi + btl %esi, %r13d + jc .LBL_1_10 + +.LBL_1_8: + incb %r12b + addl $2, %r14d + cmpb $16, %r12b + jb .LBL_1_6 + + movups 112(%rsp), %xmm8 + movups 96(%rsp), %xmm9 + movups 80(%rsp), %xmm10 + movups 64(%rsp), %xmm11 + movups 48(%rsp), %xmm12 + movups 32(%rsp), %xmm13 + movups 16(%rsp), %xmm14 + movups (%rsp), %xmm15 + movq 136(%rsp), %rsi + movq 128(%rsp), %rdi + movq 168(%rsp), %r12 + cfi_restore (%r12) + movq 160(%rsp), %r13 + cfi_restore (%r13) + movq 152(%rsp), %r14 + cfi_restore (%r14) + movq 144(%rsp), %r15 + cfi_restore (%r15) + movups 256(%rsp), %xmm0 + jmp .LBL_1_2 + +.LBL_1_10: + cfi_restore_state + movzbl %r12b, %r15d + movss 196(%rsp,%r15,8), %xmm0 + + call cosf@PLT + + movss %xmm0, 260(%rsp,%r15,8) + jmp .LBL_1_8 + +.LBL_1_12: + movzbl %r12b, %r15d + movss 192(%rsp,%r15,8), %xmm0 + + call cosf@PLT + + movss %xmm0, 256(%rsp,%r15,8) + jmp .LBL_1_7 +END (_ZGVbN4v_cosf_sse4) diff --git a/sysdeps/x86_64/fpu/multiarch/svml_s_cosf8_core.S b/sysdeps/x86_64/fpu/multiarch/svml_s_cosf8_core.S new file mode 100644 index 0000000..e14bba4 --- /dev/null +++ b/sysdeps/x86_64/fpu/multiarch/svml_s_cosf8_core.S @@ -0,0 +1,38 @@ +/* Multiple versions of vectorized cosf, vector length is 8. + Copyright (C) 2014-2015 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 + + .text +ENTRY (_ZGVdN8v_cosf) + .type _ZGVdN8v_cosf, @gnu_indirect_function + cmpl $0, KIND_OFFSET+__cpu_features(%rip) + jne 1f + call __init_cpu_features +1: leaq _ZGVdN8v_cosf_avx2(%rip), %rax + testl $bit_AVX2_Usable, __cpu_features+FEATURE_OFFSET+index_AVX2_Usable(%rip) + jz 2f + ret +2: leaq _ZGVdN8v_cosf_sse_wrapper(%rip), %rax + ret +END (_ZGVdN8v_cosf) +libmvec_hidden_def (_ZGVdN8v_cosf) + +#define _ZGVdN8v_cosf _ZGVdN8v_cosf_sse_wrapper +#include "../svml_s_cosf8_core.S" diff --git a/sysdeps/x86_64/fpu/multiarch/svml_s_cosf8_core_avx2.S b/sysdeps/x86_64/fpu/multiarch/svml_s_cosf8_core_avx2.S new file mode 100644 index 0000000..6c25e14 --- /dev/null +++ b/sysdeps/x86_64/fpu/multiarch/svml_s_cosf8_core_avx2.S @@ -0,0 +1,215 @@ +/* Function cosf vectorized with AVX2. + Copyright (C) 2014-2015 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_cosf_data.h" + + .text +ENTRY (_ZGVdN8v_cosf_avx2) +/* + ALGORITHM DESCRIPTION: + + 1) Range reduction to [-Pi/2; +Pi/2] interval + a) We remove sign using AND operation + b) Add Pi/2 value to argument X for Cos to Sin transformation + c) Getting octant Y by 1/Pi multiplication + d) Add "Right Shifter" value + e) Treat obtained value as integer for destination sign setting. + Shift first bit of this value to the last (sign) position + f) Subtract "Right Shifter" value + g) Subtract 0.5 from result for octant correction + h) Subtract Y*PI from X argument, where PI divided to 4 parts: + X = X - Y*PI1 - Y*PI2 - Y*PI3 - Y*PI4; + 2) Polynomial (minimax for sin within [-Pi/2; +Pi/2] interval) + a) Calculate X^2 = X * X + b) Calculate polynomial: + R = X + X * X^2 * (A3 + x^2 * (A5 + ..... + 3) Destination sign setting + a) Set shifted destination sign using XOR operation: + R = XOR( R, S ); + */ + pushq %rbp + cfi_adjust_cfa_offset (8) + cfi_rel_offset (%rbp, 0) + movq %rsp, %rbp + cfi_def_cfa_register (%rbp) + andq $-64, %rsp + subq $448, %rsp + movq __svml_scos_data@GOTPCREL(%rip), %rax + vmovaps %ymm0, %ymm2 + vmovups __sRShifter(%rax), %ymm5 + vmovups __sPI1_FMA(%rax), %ymm7 + +/* b) Add Pi/2 value to argument X for Cos to Sin transformation */ + vaddps __sHalfPI(%rax), %ymm2, %ymm4 + +/* + 1) Range reduction to [-Pi/2; +Pi/2] interval + c) Getting octant Y by 1/Pi multiplication + d) Add "Right Shifter" (0x4B000000) value + */ + vfmadd132ps __sInvPI(%rax), %ymm5, %ymm4 + +/* f) Subtract "Right Shifter" (0x4B000000) value */ + vsubps %ymm5, %ymm4, %ymm6 + +/* + e) Treat obtained value as integer for destination sign setting. + Shift first bit of this value to the last (sign) position (S << 31) + */ + vpslld $31, %ymm4, %ymm0 + +/* g) Subtract 0.5 from result for octant correction */ + vsubps __sOneHalf(%rax), %ymm6, %ymm4 + +/* Check for large and special arguments */ + vandps __sAbsMask(%rax), %ymm2, %ymm3 + vcmpnle_uqps __sRangeReductionVal(%rax), %ymm3, %ymm1 + +/* + h) Subtract Y*PI from X argument, where PI divided to 4 parts: + X = X - Y*PI1 - Y*PI2 - Y*PI3 + */ + vmovaps %ymm2, %ymm3 + vfnmadd231ps %ymm4, %ymm7, %ymm3 + vfnmadd231ps __sPI2_FMA(%rax), %ymm4, %ymm3 + vfnmadd132ps __sPI3_FMA(%rax), %ymm3, %ymm4 + +/* a) Calculate X^2 = X * X */ + vmulps %ymm4, %ymm4, %ymm5 + +/* + 3) Destination sign setting + a) Set shifted destination sign using XOR operation: + R = XOR( R, S ); + */ + vxorps %ymm0, %ymm4, %ymm6 + vmovups __sA9_FMA(%rax), %ymm0 + +/* + b) Calculate polynomial: + R = X + X * X^2 * (A3 + x^2 * (A5 + x^2 * (A7 + x^2 * (A9)))) + */ + vfmadd213ps __sA7_FMA(%rax), %ymm5, %ymm0 + vfmadd213ps __sA5_FMA(%rax), %ymm5, %ymm0 + vfmadd213ps __sA3(%rax), %ymm5, %ymm0 + vmulps %ymm5, %ymm0, %ymm0 + vmovmskps %ymm1, %ecx + vfmadd213ps %ymm6, %ymm6, %ymm0 + testl %ecx, %ecx + jne .LBL_1_3 + +.LBL_1_2: + cfi_remember_state + movq %rbp, %rsp + cfi_def_cfa_register (%rsp) + popq %rbp + cfi_adjust_cfa_offset (-8) + cfi_restore (%rbp) + ret + +.LBL_1_3: + cfi_restore_state + vmovups %ymm2, 320(%rsp) + vmovups %ymm0, 384(%rsp) + je .LBL_1_2 + + xorb %dl, %dl + xorl %eax, %eax + vmovups %ymm8, 224(%rsp) + vmovups %ymm9, 192(%rsp) + vmovups %ymm10, 160(%rsp) + vmovups %ymm11, 128(%rsp) + vmovups %ymm12, 96(%rsp) + vmovups %ymm13, 64(%rsp) + vmovups %ymm14, 32(%rsp) + vmovups %ymm15, (%rsp) + movq %rsi, 264(%rsp) + movq %rdi, 256(%rsp) + movq %r12, 296(%rsp) + cfi_offset_rel_rsp (12, 296) + movb %dl, %r12b + movq %r13, 288(%rsp) + cfi_offset_rel_rsp (13, 288) + movl %ecx, %r13d + movq %r14, 280(%rsp) + cfi_offset_rel_rsp (14, 280) + movl %eax, %r14d + movq %r15, 272(%rsp) + cfi_offset_rel_rsp (15, 272) + cfi_remember_state + +.LBL_1_6: + btl %r14d, %r13d + jc .LBL_1_12 + +.LBL_1_7: + lea 1(%r14), %esi + btl %esi, %r13d + jc .LBL_1_10 + +.LBL_1_8: + incb %r12b + addl $2, %r14d + cmpb $16, %r12b + jb .LBL_1_6 + + vmovups 224(%rsp), %ymm8 + vmovups 192(%rsp), %ymm9 + vmovups 160(%rsp), %ymm10 + vmovups 128(%rsp), %ymm11 + vmovups 96(%rsp), %ymm12 + vmovups 64(%rsp), %ymm13 + vmovups 32(%rsp), %ymm14 + vmovups (%rsp), %ymm15 + vmovups 384(%rsp), %ymm0 + movq 264(%rsp), %rsi + movq 256(%rsp), %rdi + movq 296(%rsp), %r12 + cfi_restore (%r12) + movq 288(%rsp), %r13 + cfi_restore (%r13) + movq 280(%rsp), %r14 + cfi_restore (%r14) + movq 272(%rsp), %r15 + cfi_restore (%r15) + jmp .LBL_1_2 + +.LBL_1_10: + cfi_restore_state + movzbl %r12b, %r15d + vmovss 324(%rsp,%r15,8), %xmm0 + vzeroupper + + call cosf@PLT + + vmovss %xmm0, 388(%rsp,%r15,8) + jmp .LBL_1_8 + +.LBL_1_12: + movzbl %r12b, %r15d + vmovss 320(%rsp,%r15,8), %xmm0 + vzeroupper + + call cosf@PLT + + vmovss %xmm0, 384(%rsp,%r15,8) + jmp .LBL_1_7 + +END (_ZGVdN8v_cosf_avx2) -- cgit v1.1