Vector cosf for x86_64.

Here is implementation of vectorized cosf containing SSE, AVX,
AVX2 and AVX512 versions according to Vector ABI
<https://groups.google.com/forum/#!topic/x86-64-abi/LmppCfN1rZ4>.

    * 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.

20 files changed, 2458 insertions, 3 deletions

diff --git a/ChangeLog b/ChangeLog
index 47318a709b..c3e52b25a9 100644
--- a/ChangeLog
+++ b/ChangeLog
@@ -47,6 +47,27 @@
 	* sysdeps/x86_64/fpu/test-double-vlen8-wrappers.c: New file.
 	* sysdeps/x86_64/fpu/test-double-vlen8.c: New file.
 
+	* 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.
+
 2015-06-09 Marko Myllynen  <myllynen@redhat.com>
 
 	* locale/C-ctype.c (PREDEFINED_CLASSES): Remove.
diff --git a/NEWS b/NEWS
index 5e223a1e39..53f244d80f 100644
--- a/NEWS
+++ b/NEWS
@@ -52,7 +52,7 @@ Version 2.22
   condition in some applications.
 
 * Added vector math library named libmvec with the following vectorized x86_64
-  implementations: cos.
+  implementations: cos, cosf.
   The library can be disabled with --disable-mathvec. Use of the functions is
   enabled with -fopenmp -ffast-math starting from -O1 for GCC version >= 4.9.0.
   The library is linked in as needed when using -lm (no need to specify -lmvec
diff --git a/sysdeps/unix/sysv/linux/x86_64/libmvec.abilist b/sysdeps/unix/sysv/linux/x86_64/libmvec.abilist
index be6eaedafd..acabb8ae3a 100644
--- a/sysdeps/unix/sysv/linux/x86_64/libmvec.abilist
+++ b/sysdeps/unix/sysv/linux/x86_64/libmvec.abilist
@@ -1,6 +1,10 @@
 GLIBC_2.22
  GLIBC_2.22 A
  _ZGVbN2v_cos F
+ _ZGVbN4v_cosf F
  _ZGVcN4v_cos F
+ _ZGVcN8v_cosf F
  _ZGVdN4v_cos F
+ _ZGVdN8v_cosf F
+ _ZGVeN16v_cosf F
  _ZGVeN8v_cos F
diff --git a/sysdeps/x86/fpu/bits/math-vector.h b/sysdeps/x86/fpu/bits/math-vector.h
index 27294ce9fa..b3ef8330e8 100644
--- a/sysdeps/x86/fpu/bits/math-vector.h
+++ b/sysdeps/x86/fpu/bits/math-vector.h
@@ -30,5 +30,7 @@
 #  define __DECL_SIMD_x86_64 _Pragma ("omp declare simd notinbranch")
 #  undef __DECL_SIMD_cos
 #  define __DECL_SIMD_cos __DECL_SIMD_x86_64
+#  undef __DECL_SIMD_cosf
+#  define __DECL_SIMD_cosf __DECL_SIMD_x86_64
 # endif
 #endif
diff --git a/sysdeps/x86_64/fpu/Makefile b/sysdeps/x86_64/fpu/Makefile
index 2d5fcf894c..af0d2af261 100644
--- a/sysdeps/x86_64/fpu/Makefile
+++ b/sysdeps/x86_64/fpu/Makefile
@@ -1,7 +1,9 @@
 ifeq ($(subdir),mathvec)
 libmvec-support += svml_d_cos2_core svml_d_cos4_core_avx \
 		   svml_d_cos4_core svml_d_cos8_core \
-		   svml_d_cos_data init-arch
+		   svml_d_cos_data svml_s_cosf4_core svml_s_cosf8_core_avx \
+		   svml_s_cosf8_core svml_s_cosf16_core svml_s_cosf_data \
+		   init-arch
 endif
 
 # Variables for libmvec tests.
diff --git a/sysdeps/x86_64/fpu/Versions b/sysdeps/x86_64/fpu/Versions
index 4290e73589..f85c28b924 100644
--- a/sysdeps/x86_64/fpu/Versions
+++ b/sysdeps/x86_64/fpu/Versions
@@ -1,5 +1,6 @@
 libmvec {
   GLIBC_2.22 {
     _ZGVbN2v_cos; _ZGVcN4v_cos; _ZGVdN4v_cos; _ZGVeN8v_cos;
+    _ZGVbN4v_cosf; _ZGVcN8v_cosf; _ZGVdN8v_cosf; _ZGVeN16v_cosf;
   }
 }
diff --git a/sysdeps/x86_64/fpu/multiarch/Makefile b/sysdeps/x86_64/fpu/multiarch/Makefile
index b2f3266490..6b50475a2a 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 0000000000..91564de22a
--- /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
+   <http://www.gnu.org/licenses/>.  */
+
+#include <sysdep.h>
+#include <init-arch.h>
+
+	.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 0000000000..a78ae2ee7e
--- /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
+   <http://www.gnu.org/licenses/>.  */
+
+#include <sysdep.h>
+#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