about summary refs log tree commit diff
path: root/sysdeps/x86_64/fpu
diff options
context:
space:
mode:
Diffstat (limited to 'sysdeps/x86_64/fpu')
-rw-r--r--sysdeps/x86_64/fpu/Makeconfig1
-rw-r--r--sysdeps/x86_64/fpu/Versions2
-rw-r--r--sysdeps/x86_64/fpu/libm-test-ulps20
-rw-r--r--sysdeps/x86_64/fpu/multiarch/svml_d_cbrt2_core-sse2.S20
-rw-r--r--sysdeps/x86_64/fpu/multiarch/svml_d_cbrt2_core.c27
-rw-r--r--sysdeps/x86_64/fpu/multiarch/svml_d_cbrt2_core_sse4.S467
-rw-r--r--sysdeps/x86_64/fpu/multiarch/svml_d_cbrt4_core-sse.S20
-rw-r--r--sysdeps/x86_64/fpu/multiarch/svml_d_cbrt4_core.c27
-rw-r--r--sysdeps/x86_64/fpu/multiarch/svml_d_cbrt4_core_avx2.S505
-rw-r--r--sysdeps/x86_64/fpu/multiarch/svml_d_cbrt8_core-avx2.S20
-rw-r--r--sysdeps/x86_64/fpu/multiarch/svml_d_cbrt8_core.c27
-rw-r--r--sysdeps/x86_64/fpu/multiarch/svml_d_cbrt8_core_avx512.S253
-rw-r--r--sysdeps/x86_64/fpu/multiarch/svml_s_cbrtf16_core-avx2.S20
-rw-r--r--sysdeps/x86_64/fpu/multiarch/svml_s_cbrtf16_core.c28
-rw-r--r--sysdeps/x86_64/fpu/multiarch/svml_s_cbrtf16_core_avx512.S235
-rw-r--r--sysdeps/x86_64/fpu/multiarch/svml_s_cbrtf4_core-sse2.S20
-rw-r--r--sysdeps/x86_64/fpu/multiarch/svml_s_cbrtf4_core.c28
-rw-r--r--sysdeps/x86_64/fpu/multiarch/svml_s_cbrtf4_core_sse4.S490
-rw-r--r--sysdeps/x86_64/fpu/multiarch/svml_s_cbrtf8_core-sse.S20
-rw-r--r--sysdeps/x86_64/fpu/multiarch/svml_s_cbrtf8_core.c28
-rw-r--r--sysdeps/x86_64/fpu/multiarch/svml_s_cbrtf8_core_avx2.S509
-rw-r--r--sysdeps/x86_64/fpu/svml_d_cbrt2_core.S29
-rw-r--r--sysdeps/x86_64/fpu/svml_d_cbrt4_core.S29
-rw-r--r--sysdeps/x86_64/fpu/svml_d_cbrt4_core_avx.S25
-rw-r--r--sysdeps/x86_64/fpu/svml_d_cbrt8_core.S25
-rw-r--r--sysdeps/x86_64/fpu/svml_s_cbrtf16_core.S25
-rw-r--r--sysdeps/x86_64/fpu/svml_s_cbrtf4_core.S29
-rw-r--r--sysdeps/x86_64/fpu/svml_s_cbrtf8_core.S29
-rw-r--r--sysdeps/x86_64/fpu/svml_s_cbrtf8_core_avx.S25
-rw-r--r--sysdeps/x86_64/fpu/test-double-libmvec-cbrt-avx.c1
-rw-r--r--sysdeps/x86_64/fpu/test-double-libmvec-cbrt-avx2.c1
-rw-r--r--sysdeps/x86_64/fpu/test-double-libmvec-cbrt-avx512f.c1
-rw-r--r--sysdeps/x86_64/fpu/test-double-libmvec-cbrt.c3
-rw-r--r--sysdeps/x86_64/fpu/test-double-vlen2-wrappers.c1
-rw-r--r--sysdeps/x86_64/fpu/test-double-vlen4-avx2-wrappers.c1
-rw-r--r--sysdeps/x86_64/fpu/test-double-vlen4-wrappers.c1
-rw-r--r--sysdeps/x86_64/fpu/test-double-vlen8-wrappers.c1
-rw-r--r--sysdeps/x86_64/fpu/test-float-libmvec-cbrtf-avx.c1
-rw-r--r--sysdeps/x86_64/fpu/test-float-libmvec-cbrtf-avx2.c1
-rw-r--r--sysdeps/x86_64/fpu/test-float-libmvec-cbrtf-avx512f.c1
-rw-r--r--sysdeps/x86_64/fpu/test-float-libmvec-cbrtf.c3
-rw-r--r--sysdeps/x86_64/fpu/test-float-vlen16-wrappers.c1
-rw-r--r--sysdeps/x86_64/fpu/test-float-vlen4-wrappers.c1
-rw-r--r--sysdeps/x86_64/fpu/test-float-vlen8-avx2-wrappers.c1
-rw-r--r--sysdeps/x86_64/fpu/test-float-vlen8-wrappers.c1
45 files changed, 3003 insertions, 0 deletions
diff --git a/sysdeps/x86_64/fpu/Makeconfig b/sysdeps/x86_64/fpu/Makeconfig
index 81e9fc95b2..dde737c0d6 100644
--- a/sysdeps/x86_64/fpu/Makeconfig
+++ b/sysdeps/x86_64/fpu/Makeconfig
@@ -25,6 +25,7 @@ libmvec-funcs = \
   acos \
   asin \
   atan \
+  cbrt \
   cos \
   cosh \
   exp \
diff --git a/sysdeps/x86_64/fpu/Versions b/sysdeps/x86_64/fpu/Versions
index 2710446d12..b70aeb3e2f 100644
--- a/sysdeps/x86_64/fpu/Versions
+++ b/sysdeps/x86_64/fpu/Versions
@@ -17,6 +17,7 @@ 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_cbrt; _ZGVcN4v_cbrt; _ZGVdN4v_cbrt; _ZGVeN8v_cbrt;
     _ZGVbN2v_cosh; _ZGVcN4v_cosh; _ZGVdN4v_cosh; _ZGVeN8v_cosh;
     _ZGVbN2v_exp10; _ZGVcN4v_exp10; _ZGVdN4v_exp10; _ZGVeN8v_exp10;
     _ZGVbN2v_exp2; _ZGVcN4v_exp2; _ZGVdN4v_exp2; _ZGVeN8v_exp2;
@@ -26,6 +27,7 @@ libmvec {
     _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_cbrtf; _ZGVcN8v_cbrtf; _ZGVdN8v_cbrtf; _ZGVeN16v_cbrtf;
     _ZGVbN4v_coshf; _ZGVcN8v_coshf; _ZGVdN8v_coshf; _ZGVeN16v_coshf;
     _ZGVbN4v_exp10f; _ZGVcN8v_exp10f; _ZGVdN8v_exp10f; _ZGVeN16v_exp10f;
     _ZGVbN4v_exp2f; _ZGVcN8v_exp2f; _ZGVdN8v_exp2f; _ZGVeN16v_exp2f;
diff --git a/sysdeps/x86_64/fpu/libm-test-ulps b/sysdeps/x86_64/fpu/libm-test-ulps
index f4b313119d..e039a993df 100644
--- a/sysdeps/x86_64/fpu/libm-test-ulps
+++ b/sysdeps/x86_64/fpu/libm-test-ulps
@@ -583,6 +583,26 @@ float: 1
 float128: 1
 ldouble: 1
 
+Function: "cbrt_vlen16":
+float: 1
+
+Function: "cbrt_vlen2":
+double: 1
+
+Function: "cbrt_vlen4":
+double: 1
+float: 2
+
+Function: "cbrt_vlen4_avx2":
+double: 1
+
+Function: "cbrt_vlen8":
+double: 1
+float: 2
+
+Function: "cbrt_vlen8_avx2":
+float: 2
+
 Function: Real part of "ccos":
 double: 1
 float: 1
diff --git a/sysdeps/x86_64/fpu/multiarch/svml_d_cbrt2_core-sse2.S b/sysdeps/x86_64/fpu/multiarch/svml_d_cbrt2_core-sse2.S
new file mode 100644
index 0000000000..60f4c46a11
--- /dev/null
+++ b/sysdeps/x86_64/fpu/multiarch/svml_d_cbrt2_core-sse2.S
@@ -0,0 +1,20 @@
+/* SSE2 version of vectorized cbrt, 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
+   <https://www.gnu.org/licenses/>.  */
+
+#define _ZGVbN2v_cbrt _ZGVbN2v_cbrt_sse2
+#include "../svml_d_cbrt2_core.S"
diff --git a/sysdeps/x86_64/fpu/multiarch/svml_d_cbrt2_core.c b/sysdeps/x86_64/fpu/multiarch/svml_d_cbrt2_core.c
new file mode 100644
index 0000000000..07390b7150
--- /dev/null
+++ b/sysdeps/x86_64/fpu/multiarch/svml_d_cbrt2_core.c
@@ -0,0 +1,27 @@
+/* Multiple versions of vectorized cbrt, 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
+   <https://www.gnu.org/licenses/>.  */
+
+#define SYMBOL_NAME _ZGVbN2v_cbrt
+#include "ifunc-mathvec-sse4_1.h"
+
+libc_ifunc_redirected (REDIRECT_NAME, SYMBOL_NAME, IFUNC_SELECTOR ());
+
+#ifdef SHARED
+__hidden_ver1 (_ZGVbN2v_cbrt, __GI__ZGVbN2v_cbrt, __redirect__ZGVbN2v_cbrt)
+  __attribute__ ((visibility ("hidden")));
+#endif
diff --git a/sysdeps/x86_64/fpu/multiarch/svml_d_cbrt2_core_sse4.S b/sysdeps/x86_64/fpu/multiarch/svml_d_cbrt2_core_sse4.S
new file mode 100644
index 0000000000..72ecb25e05
--- /dev/null
+++ b/sysdeps/x86_64/fpu/multiarch/svml_d_cbrt2_core_sse4.S
@@ -0,0 +1,467 @@
+/* Function cbrt 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:
+ *
+ *   x=2^{3*k+j} * 1.b1 b2 ... b5 b6 ... b52
+ *   Let r=(x*2^{-3k-j} - 1.b1 b2 ... b5 1)* rcp[b1 b2 ..b5],
+ *   where rcp[b1 b2 .. b5]=1/(1.b1 b2 b3 b4 b5 1) in double precision
+ *   cbrt(2^j * 1. b1 b2 .. b5 1) is approximated as T[j][b1..b5]+D[j][b1..b5]
+ *   (T stores the high 53 bits, D stores the low order bits)
+ *   Result=2^k*T+(2^k*T*r)*P+2^k*D
+ *   where P=p1+p2*r+..+p8*r^7
+ *
+ */
+
+/* Offsets for data table __svml_dcbrt_data_internal
+ */
+#define _dRcp                         	0
+#define _dCbrtHiLo                    	256
+#define _dA7                          	1024
+#define _dA6                          	1040
+#define _dA5                          	1056
+#define _dA4                          	1072
+#define _dA3                          	1088
+#define _dA2                          	1104
+#define _dA1                          	1120
+#define _dNeg65Div64                  	1136
+#define _dSgnf6Mask                   	1152
+#define _dNegOne                      	1168
+#define _dMantissaMask                	1184
+#define _lExpHiMask                   	1200
+#define _lExpLoMask                   	1216
+#define _l1556                        	1232
+#define _iRcpIndexMask                	1248
+#define _iAbsMask                     	1264
+#define _iSignMask                    	1280
+#define _iBias                        	1296
+#define _iSub                         	1312
+#define _iCmp                         	1328
+
+#include <sysdep.h>
+
+        .text
+	.section .text.sse4,"ax",@progbits
+ENTRY(_ZGVbN2v_cbrt_sse4)
+        subq      $72, %rsp
+        cfi_def_cfa_offset(80)
+
+/* Calculate CbrtIndex */
+        movaps    %xmm0, %xmm10
+        psrlq     $52, %xmm10
+
+/* Load 1/(1+iRcpIndex/32+1/64) reciprocal table value */
+        lea       __svml_dcbrt_data_internal(%rip), %r8
+        pand      _lExpLoMask+__svml_dcbrt_data_internal(%rip), %xmm10
+        movdqu    _l1556+__svml_dcbrt_data_internal(%rip), %xmm9
+        pmuludq   %xmm10, %xmm9
+
+/* If the exponent field is zero - go to callout to process denormals */
+        movq      _iAbsMask+__svml_dcbrt_data_internal(%rip), %xmm7
+
+/* Calculate Rcp table index */
+        movq      _iRcpIndexMask+__svml_dcbrt_data_internal(%rip), %xmm13
+
+/* Get iX - high part of argument */
+        pshufd    $221, %xmm0, %xmm4
+
+/*
+ * Declarations
+ * Load constants
+ */
+        movq      _iSignMask+__svml_dcbrt_data_internal(%rip), %xmm1
+        pand      %xmm4, %xmm7
+        pand      %xmm4, %xmm13
+
+/* Compute 2^k */
+        psrld     $20, %xmm4
+        movq      _iBias+__svml_dcbrt_data_internal(%rip), %xmm2
+        pand      %xmm1, %xmm4
+        pshufd    $136, %xmm9, %xmm15
+        por       %xmm2, %xmm4
+        psrld     $14, %xmm15
+        psrld     $12, %xmm13
+        paddd     %xmm15, %xmm4
+        pxor      %xmm2, %xmm2
+        pslld     $20, %xmm4
+        movdqa    %xmm15, %xmm11
+        movd      %xmm13, %edx
+        paddd     %xmm15, %xmm11
+        pshufd    $1, %xmm13, %xmm8
+        punpckldq %xmm4, %xmm2
+
+/*
+ * VAND( L, l2k, = l2k, lExpHiMask );
+ * Argument reduction Z
+ */
+        movups    _dMantissaMask+__svml_dcbrt_data_internal(%rip), %xmm1
+        movups    _dSgnf6Mask+__svml_dcbrt_data_internal(%rip), %xmm4
+        andps     %xmm0, %xmm1
+        movd      %xmm8, %ecx
+        andps     %xmm0, %xmm4
+        orps      _dNegOne+__svml_dcbrt_data_internal(%rip), %xmm1
+        orps      _dNeg65Div64+__svml_dcbrt_data_internal(%rip), %xmm4
+        movslq    %edx, %rdx
+        subpd     %xmm4, %xmm1
+        movslq    %ecx, %rcx
+        movsd     (%r8,%rdx), %xmm3
+        movq      _iSub+__svml_dcbrt_data_internal(%rip), %xmm5
+        psubd     %xmm5, %xmm7
+        movhpd    (%r8,%rcx), %xmm3
+        mulpd     %xmm1, %xmm3
+
+/* Polynomial */
+        movups    _dA7+__svml_dcbrt_data_internal(%rip), %xmm5
+        mulpd     %xmm3, %xmm5
+        addpd     _dA6+__svml_dcbrt_data_internal(%rip), %xmm5
+        mulpd     %xmm3, %xmm5
+        addpd     _dA5+__svml_dcbrt_data_internal(%rip), %xmm5
+        mulpd     %xmm3, %xmm5
+        addpd     _dA4+__svml_dcbrt_data_internal(%rip), %xmm5
+        mulpd     %xmm3, %xmm5
+        addpd     _dA3+__svml_dcbrt_data_internal(%rip), %xmm5
+        pshufd    $136, %xmm10, %xmm12
+        psubd     %xmm15, %xmm12
+        psubd     %xmm11, %xmm12
+        mulpd     %xmm3, %xmm5
+        pslld     $8, %xmm12
+        paddd     %xmm12, %xmm13
+
+/* Load cbrt(2^j*(1+iRcpIndex/32+1/64)) Hi & Lo values */
+        movd      %xmm13, %esi
+        pshufd    $1, %xmm13, %xmm14
+        movq      _iCmp+__svml_dcbrt_data_internal(%rip), %xmm6
+        movd      %xmm14, %edi
+        pcmpgtd   %xmm6, %xmm7
+        movmskps  %xmm7, %eax
+        addpd     _dA2+__svml_dcbrt_data_internal(%rip), %xmm5
+        movslq    %esi, %rsi
+        movslq    %edi, %rdi
+        mulpd     %xmm3, %xmm5
+        movsd     256(%r8,%rsi), %xmm6
+        movhpd    256(%r8,%rdi), %xmm6
+
+/* THi*2^k, TLo*2^k */
+        mulpd     %xmm2, %xmm6
+        addpd     _dA1+__svml_dcbrt_data_internal(%rip), %xmm5
+
+/* THi*2^k*Z */
+        mulpd     %xmm6, %xmm3
+
+/* Final reconstruction */
+        mulpd     %xmm3, %xmm5
+        addpd     %xmm5, %xmm6
+        andl      $3, %eax
+
+/* Go to special inputs processing branch */
+        jne       L(SPECIAL_VALUES_BRANCH)
+                                # LOE rbx rbp r12 r13 r14 r15 eax xmm0 xmm6
+
+/* Restore registers
+ * and exit the function
+ */
+
+L(EXIT):
+        movaps    %xmm6, %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    %xmm6, 48(%rsp)
+                                # LOE rbx rbp r12 r13 r14 r15 eax xmm6
+
+        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), %xmm6
+
+/* 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 xmm6
+
+/* Scalar math fucntion call
+ * to process special input
+ */
+
+L(SCALAR_MATH_CALL):
+        movl      %r12d, %r14d
+        movsd     32(%rsp,%r14,8), %xmm0
+        call      cbrt@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_cbrt_sse4)
+
+        .section .rodata, "a"
+        .align 16
+
+#ifdef __svml_dcbrt_data_internal_typedef
+typedef unsigned int VUINT32;
+typedef struct
+{
+        __declspec(align(16)) VUINT32 _dRcp[32][2];
+        __declspec(align(16)) VUINT32 _dCbrtHiLo[96][2];
+        __declspec(align(16)) VUINT32 _dA7[2][2];
+        __declspec(align(16)) VUINT32 _dA6[2][2];
+        __declspec(align(16)) VUINT32 _dA5[2][2];
+        __declspec(align(16)) VUINT32 _dA4[2][2];
+        __declspec(align(16)) VUINT32 _dA3[2][2];
+        __declspec(align(16)) VUINT32 _dA2[2][2];
+        __declspec(align(16)) VUINT32 _dA1[2][2];
+        __declspec(align(16)) VUINT32 _dNeg65Div64[2][2];
+        __declspec(align(16)) VUINT32 _dSgnf6Mask[2][2];
+        __declspec(align(16)) VUINT32 _dNegOne[2][2];
+        __declspec(align(16)) VUINT32 _dMantissaMask[2][2];
+        __declspec(align(16)) VUINT32 _lExpHiMask[2][2];
+        __declspec(align(16)) VUINT32 _lExpLoMask[2][2];
+        __declspec(align(16)) VUINT32 _l1556[2][2];
+        __declspec(align(16)) VUINT32 _iRcpIndexMask[4][1];
+        __declspec(align(16)) VUINT32 _iAbsMask[4][1];
+        __declspec(align(16)) VUINT32 _iSignMask[4][1];
+        __declspec(align(16)) VUINT32 _iBias[4][1];
+        __declspec(align(16)) VUINT32 _iSub[4][1];
+        __declspec(align(16)) VUINT32 _iCmp[4][1];
+} __svml_dcbrt_data_internal;
+#endif
+__svml_dcbrt_data_internal:
+        /*== _dRcp ==*/
+        .quad 0xBFEF81F81F81F820  /* (1/(1+0/32+1/64)) = -.984615 */
+        .quad 0xBFEE9131ABF0B767  /* (1/(1+1/32+1/64)) = -.955224 */
+        .quad 0xBFEDAE6076B981DB  /* (1/(1+2/32+1/64)) = -.927536 */
+        .quad 0xBFECD85689039B0B  /* (1/(1+3/32+1/64)) = -.901408 */
+        .quad 0xBFEC0E070381C0E0  /* (1/(1+4/32+1/64)) = -.876712 */
+        .quad 0xBFEB4E81B4E81B4F  /* (1/(1+5/32+1/64)) = -.853333 */
+        .quad 0xBFEA98EF606A63BE  /* (1/(1+6/32+1/64)) = -.831169 */
+        .quad 0xBFE9EC8E951033D9  /* (1/(1+7/32+1/64)) = -.810127 */
+        .quad 0xBFE948B0FCD6E9E0  /* (1/(1+8/32+1/64)) = -.790123 */
+        .quad 0xBFE8ACB90F6BF3AA  /* (1/(1+9/32+1/64)) = -.771084 */
+        .quad 0xBFE8181818181818  /* (1/(1+10/32+1/64)) = -.752941 */
+        .quad 0xBFE78A4C8178A4C8  /* (1/(1+11/32+1/64)) = -.735632 */
+        .quad 0xBFE702E05C0B8170  /* (1/(1+12/32+1/64)) = -.719101 */
+        .quad 0xBFE6816816816817  /* (1/(1+13/32+1/64)) = -.703297 */
+        .quad 0xBFE6058160581606  /* (1/(1+14/32+1/64)) = -.688172 */
+        .quad 0xBFE58ED2308158ED  /* (1/(1+15/32+1/64)) = -.673684 */
+        .quad 0xBFE51D07EAE2F815  /* (1/(1+16/32+1/64)) = -.659794 */
+        .quad 0xBFE4AFD6A052BF5B  /* (1/(1+17/32+1/64)) = -.646465 */
+        .quad 0xBFE446F86562D9FB  /* (1/(1+18/32+1/64)) = -.633663 */
+        .quad 0xBFE3E22CBCE4A902  /* (1/(1+19/32+1/64)) = -.621359 */
+        .quad 0xBFE3813813813814  /* (1/(1+20/32+1/64)) = -.609524 */
+        .quad 0xBFE323E34A2B10BF  /* (1/(1+21/32+1/64)) = -.598131 */
+        .quad 0xBFE2C9FB4D812CA0  /* (1/(1+22/32+1/64)) = -.587156 */
+        .quad 0xBFE27350B8812735  /* (1/(1+23/32+1/64)) = -.576577 */
+        .quad 0xBFE21FB78121FB78  /* (1/(1+24/32+1/64)) = -.566372 */
+        .quad 0xBFE1CF06ADA2811D  /* (1/(1+25/32+1/64)) = -.556522 */
+        .quad 0xBFE1811811811812  /* (1/(1+26/32+1/64)) = -.547009 */
+        .quad 0xBFE135C81135C811  /* (1/(1+27/32+1/64)) = -.537815 */
+        .quad 0xBFE0ECF56BE69C90  /* (1/(1+28/32+1/64)) = -.528926 */
+        .quad 0xBFE0A6810A6810A7  /* (1/(1+29/32+1/64)) = -.520325 */
+        .quad 0xBFE0624DD2F1A9FC  /* (1/(1+30/32+1/64)) = -.512    */
+        .quad 0xBFE0204081020408  /* (1/(1+31/32+1/64)) = -.503937 */
+        /*== _dCbrtHiLo ==*/
+        .align 16
+        .quad 0x3FF01539221D4C97    /* HI((2^0*(1+0/32+1/64))^(1/3)) = 1.005181 */
+        .quad 0x3FF03F06771A2E33    /* HI((2^0*(1+1/32+1/64))^(1/3)) = 1.015387 */
+        .quad 0x3FF06800E629D671    /* HI((2^0*(1+2/32+1/64))^(1/3)) = 1.025391 */
+        .quad 0x3FF090328731DEB2    /* HI((2^0*(1+3/32+1/64))^(1/3)) = 1.035204 */
+        .quad 0x3FF0B7A4B1BD64AC    /* HI((2^0*(1+4/32+1/64))^(1/3)) = 1.044835 */
+        .quad 0x3FF0DE601024FB87    /* HI((2^0*(1+5/32+1/64))^(1/3)) = 1.054291 */
+        .quad 0x3FF1046CB0597000    /* HI((2^0*(1+6/32+1/64))^(1/3)) = 1.06358  */
+        .quad 0x3FF129D212A9BA9B    /* HI((2^0*(1+7/32+1/64))^(1/3)) = 1.07271  */
+        .quad 0x3FF14E9736CDAF38    /* HI((2^0*(1+8/32+1/64))^(1/3)) = 1.081687 */
+        .quad 0x3FF172C2A772F507    /* HI((2^0*(1+9/32+1/64))^(1/3)) = 1.090518 */
+        .quad 0x3FF1965A848001D3    /* HI((2^0*(1+10/32+1/64))^(1/3)) = 1.099207 */
+        .quad 0x3FF1B9648C38C55D    /* HI((2^0*(1+11/32+1/64))^(1/3)) = 1.107762 */
+        .quad 0x3FF1DBE6236A0C45    /* HI((2^0*(1+12/32+1/64))^(1/3)) = 1.116186 */
+        .quad 0x3FF1FDE45CBB1F9F    /* HI((2^0*(1+13/32+1/64))^(1/3)) = 1.124485 */
+        .quad 0x3FF21F63FF409042    /* HI((2^0*(1+14/32+1/64))^(1/3)) = 1.132664 */
+        .quad 0x3FF240698C6746E5    /* HI((2^0*(1+15/32+1/64))^(1/3)) = 1.140726 */
+        .quad 0x3FF260F9454BB99B    /* HI((2^0*(1+16/32+1/64))^(1/3)) = 1.148675 */
+        .quad 0x3FF281172F8E7073    /* HI((2^0*(1+17/32+1/64))^(1/3)) = 1.156516 */
+        .quad 0x3FF2A0C719B4B6D0    /* HI((2^0*(1+18/32+1/64))^(1/3)) = 1.164252 */
+        .quad 0x3FF2C00C9F2263EC    /* HI((2^0*(1+19/32+1/64))^(1/3)) = 1.171887 */
+        .quad 0x3FF2DEEB2BB7FB78    /* HI((2^0*(1+20/32+1/64))^(1/3)) = 1.179423 */
+        .quad 0x3FF2FD65FF1EFBBC    /* HI((2^0*(1+21/32+1/64))^(1/3)) = 1.186865 */
+        .quad 0x3FF31B802FCCF6A2    /* HI((2^0*(1+22/32+1/64))^(1/3)) = 1.194214 */
+        .quad 0x3FF3393CADC50708    /* HI((2^0*(1+23/32+1/64))^(1/3)) = 1.201474 */
+        .quad 0x3FF3569E451E4C2A    /* HI((2^0*(1+24/32+1/64))^(1/3)) = 1.208647 */
+        .quad 0x3FF373A7A0554CDE    /* HI((2^0*(1+25/32+1/64))^(1/3)) = 1.215736 */
+        .quad 0x3FF3905B4A6D76CE    /* HI((2^0*(1+26/32+1/64))^(1/3)) = 1.222743 */
+        .quad 0x3FF3ACBBB0E756B6    /* HI((2^0*(1+27/32+1/64))^(1/3)) = 1.229671 */
+        .quad 0x3FF3C8CB258FA340    /* HI((2^0*(1+28/32+1/64))^(1/3)) = 1.236522 */
+        .quad 0x3FF3E48BE02AC0CE    /* HI((2^0*(1+29/32+1/64))^(1/3)) = 1.243297 */
+        .quad 0x3FF4000000000000    /* HI((2^0*(1+30/32+1/64))^(1/3)) = 1.25     */
+        .quad 0x3FF41B298D47800E    /* HI((2^0*(1+31/32+1/64))^(1/3)) = 1.256631 */
+        .quad 0x3FF443604B34D9B2    /* HI((2^1*(1+0/32+1/64))^(1/3)) = 1.266449 */
+        .quad 0x3FF4780B20906571    /* HI((2^1*(1+1/32+1/64))^(1/3)) = 1.279307 */
+        .quad 0x3FF4ABAC3EE06706    /* HI((2^1*(1+2/32+1/64))^(1/3)) = 1.291912 */
+        .quad 0x3FF4DE505DA66B8D    /* HI((2^1*(1+3/32+1/64))^(1/3)) = 1.304276 */
+        .quad 0x3FF51003420A5C07    /* HI((2^1*(1+4/32+1/64))^(1/3)) = 1.316409 */
+        .quad 0x3FF540CFD6FD11C1    /* HI((2^1*(1+5/32+1/64))^(1/3)) = 1.328323 */
+        .quad 0x3FF570C04260716B    /* HI((2^1*(1+6/32+1/64))^(1/3)) = 1.340027 */
+        .quad 0x3FF59FDDF7A45F38    /* HI((2^1*(1+7/32+1/64))^(1/3)) = 1.35153  */
+        .quad 0x3FF5CE31C83539DF    /* HI((2^1*(1+8/32+1/64))^(1/3)) = 1.36284  */
+        .quad 0x3FF5FBC3F20966A4    /* HI((2^1*(1+9/32+1/64))^(1/3)) = 1.373966 */
+        .quad 0x3FF6289C2C8F1B70    /* HI((2^1*(1+10/32+1/64))^(1/3)) = 1.384915 */
+        .quad 0x3FF654C1B4316DCF    /* HI((2^1*(1+11/32+1/64))^(1/3)) = 1.395693 */
+        .quad 0x3FF6803B54A34E44    /* HI((2^1*(1+12/32+1/64))^(1/3)) = 1.406307 */
+        .quad 0x3FF6AB0F72182659    /* HI((2^1*(1+13/32+1/64))^(1/3)) = 1.416763 */
+        .quad 0x3FF6D544118C08BC    /* HI((2^1*(1+14/32+1/64))^(1/3)) = 1.427067 */
+        .quad 0x3FF6FEDEE0388D4A    /* HI((2^1*(1+15/32+1/64))^(1/3)) = 1.437224 */
+        .quad 0x3FF727E53A4F645E    /* HI((2^1*(1+16/32+1/64))^(1/3)) = 1.44724  */
+        .quad 0x3FF7505C31104114    /* HI((2^1*(1+17/32+1/64))^(1/3)) = 1.457119 */
+        .quad 0x3FF77848904CD549    /* HI((2^1*(1+18/32+1/64))^(1/3)) = 1.466866 */
+        .quad 0x3FF79FAEE36B2534    /* HI((2^1*(1+19/32+1/64))^(1/3)) = 1.476485 */
+        .quad 0x3FF7C69379F4605B    /* HI((2^1*(1+20/32+1/64))^(1/3)) = 1.48598  */
+        .quad 0x3FF7ECFA6BBCA391    /* HI((2^1*(1+21/32+1/64))^(1/3)) = 1.495356 */
+        .quad 0x3FF812E79CAE7EB9    /* HI((2^1*(1+22/32+1/64))^(1/3)) = 1.504615 */
+        .quad 0x3FF8385EC043C71D    /* HI((2^1*(1+23/32+1/64))^(1/3)) = 1.513762 */
+        .quad 0x3FF85D635CB41B9D    /* HI((2^1*(1+24/32+1/64))^(1/3)) = 1.5228   */
+        .quad 0x3FF881F8CDE083DB    /* HI((2^1*(1+25/32+1/64))^(1/3)) = 1.531731 */
+        .quad 0x3FF8A6224802B8A8    /* HI((2^1*(1+26/32+1/64))^(1/3)) = 1.54056  */
+        .quad 0x3FF8C9E2DA25E5E4    /* HI((2^1*(1+27/32+1/64))^(1/3)) = 1.549289 */
+        .quad 0x3FF8ED3D706E1010    /* HI((2^1*(1+28/32+1/64))^(1/3)) = 1.55792  */
+        .quad 0x3FF91034D632B6DF    /* HI((2^1*(1+29/32+1/64))^(1/3)) = 1.566457 */
+        .quad 0x3FF932CBB7F0CF2D    /* HI((2^1*(1+30/32+1/64))^(1/3)) = 1.574901 */
+        .quad 0x3FF95504A517BF3A    /* HI((2^1*(1+31/32+1/64))^(1/3)) = 1.583256 */
+        .quad 0x3FF987AF34F8BB19    /* HI((2^2*(1+0/32+1/64))^(1/3)) = 1.595626 */
+        .quad 0x3FF9CA0A8337B317    /* HI((2^2*(1+1/32+1/64))^(1/3)) = 1.611826 */
+        .quad 0x3FFA0B1709CC13D5    /* HI((2^2*(1+2/32+1/64))^(1/3)) = 1.627708 */
+        .quad 0x3FFA4AE4CE6419ED    /* HI((2^2*(1+3/32+1/64))^(1/3)) = 1.643285 */
+        .quad 0x3FFA8982A5567031    /* HI((2^2*(1+4/32+1/64))^(1/3)) = 1.658572 */
+        .quad 0x3FFAC6FE500AB570    /* HI((2^2*(1+5/32+1/64))^(1/3)) = 1.673582 */
+        .quad 0x3FFB036497A15A17    /* HI((2^2*(1+6/32+1/64))^(1/3)) = 1.688328 */
+        .quad 0x3FFB3EC164671755    /* HI((2^2*(1+7/32+1/64))^(1/3)) = 1.702821 */
+        .quad 0x3FFB791FD288C46F    /* HI((2^2*(1+8/32+1/64))^(1/3)) = 1.717071 */
+        .quad 0x3FFBB28A44693BE4    /* HI((2^2*(1+9/32+1/64))^(1/3)) = 1.731089 */
+        .quad 0x3FFBEB0A72EB6E31    /* HI((2^2*(1+10/32+1/64))^(1/3)) = 1.744883 */
+        .quad 0x3FFC22A97BF5F697    /* HI((2^2*(1+11/32+1/64))^(1/3)) = 1.758462 */
+        .quad 0x3FFC596FEF6AF983    /* HI((2^2*(1+12/32+1/64))^(1/3)) = 1.771835 */
+        .quad 0x3FFC8F65DAC655A3    /* HI((2^2*(1+13/32+1/64))^(1/3)) = 1.785009 */
+        .quad 0x3FFCC492D38CE8D9    /* HI((2^2*(1+14/32+1/64))^(1/3)) = 1.797992 */
+        .quad 0x3FFCF8FE00B19367    /* HI((2^2*(1+15/32+1/64))^(1/3)) = 1.810789 */
+        .quad 0x3FFD2CAE230F8709    /* HI((2^2*(1+16/32+1/64))^(1/3)) = 1.823408 */
+        .quad 0x3FFD5FA99D15208F    /* HI((2^2*(1+17/32+1/64))^(1/3)) = 1.835855 */
+        .quad 0x3FFD91F679B6E505    /* HI((2^2*(1+18/32+1/64))^(1/3)) = 1.848135 */
+        .quad 0x3FFDC39A72BF2302    /* HI((2^2*(1+19/32+1/64))^(1/3)) = 1.860255 */
+        .quad 0x3FFDF49AF68C1570    /* HI((2^2*(1+20/32+1/64))^(1/3)) = 1.872218 */
+        .quad 0x3FFE24FD2D4C23B8    /* HI((2^2*(1+21/32+1/64))^(1/3)) = 1.884031 */
+        .quad 0x3FFE54C5FDC5EC73    /* HI((2^2*(1+22/32+1/64))^(1/3)) = 1.895697 */
+        .quad 0x3FFE83FA11B81DBB    /* HI((2^2*(1+23/32+1/64))^(1/3)) = 1.907221 */
+        .quad 0x3FFEB29DD9DBAF25    /* HI((2^2*(1+24/32+1/64))^(1/3)) = 1.918608 */
+        .quad 0x3FFEE0B59191D374    /* HI((2^2*(1+25/32+1/64))^(1/3)) = 1.929861 */
+        .quad 0x3FFF0E454245E4BF    /* HI((2^2*(1+26/32+1/64))^(1/3)) = 1.940984 */
+        .quad 0x3FFF3B50C68A9DD3    /* HI((2^2*(1+27/32+1/64))^(1/3)) = 1.951981 */
+        .quad 0x3FFF67DBCCF922DC    /* HI((2^2*(1+28/32+1/64))^(1/3)) = 1.962856 */
+        .quad 0x3FFF93E9DAD7A4A6    /* HI((2^2*(1+29/32+1/64))^(1/3)) = 1.973612 */
+        .quad 0x3FFFBF7E4E8CC9CB    /* HI((2^2*(1+30/32+1/64))^(1/3)) = 1.984251 */
+        .quad 0x3FFFEA9C61E47CD3    /* HI((2^2*(1+31/32+1/64))^(1/3)) = 1.994778 */
+        .align 16
+        .quad 0x3F93750AD588F115, 0x3F93750AD588F115      /* _dA7 */
+        .align 16
+        .quad 0xBF98090D6221A247, 0xBF98090D6221A247      /* _dA6 */
+        .align 16
+        .quad 0x3F9EE7113506AC12, 0x3F9EE7113506AC12      /* _dA5 */
+        .align 16
+        .quad 0xBFA511E8D2B3183B, 0xBFA511E8D2B3183B      /* _dA4 */
+        .align 16
+        .quad 0x3FAF9ADD3C0CA458, 0x3FAF9ADD3C0CA458      /* _dA3 */
+        .align 16
+        .quad 0xBFBC71C71C71C71C, 0xBFBC71C71C71C71C      /* _dA2 */
+        .align 16
+        .quad 0x3FD5555555555555, 0x3FD5555555555555      /* _dA1 */
+        .align 16
+        .quad 0xBFF0400000000000, 0xBFF0400000000000        /* _dNeg65Div64 */
+        .align 16
+        .quad 0x000FC00000000000, 0x000FC00000000000        /* _dSgnf6Mask */
+        .align 16
+        .quad 0xBFF0000000000000, 0xBFF0000000000000        /* _dNegOne */
+        .align 16
+        .quad 0x000FFFFFFFFFFFFF, 0x000FFFFFFFFFFFFF        /* _dMantissaMask */
+        .align 16
+        .quad 0xFFF0000000000000, 0xFFF0000000000000        /* _lExpHiMask */
+        .align 16
+        .quad 0x00000000000007FF, 0x00000000000007FF        /* _lExpLoMask */
+        .align 16
+        .quad 0x0000000000001556, 0x0000000000001556        /* _l1556 */
+        .align 16
+        .long 0x000F8000, 0x000F8000, 0x000F8000, 0x000F8000    /* _iRcpIndexMask */
+        .align 16
+        .long 0x7FFFFFFF, 0x7FFFFFFF, 0x7FFFFFFF, 0x7FFFFFFF    /* _iAbsMask */
+        .align 16
+        .long 0x00000800, 0x00000800, 0x00000800, 0x00000800    /* _iSignMask */
+        .align 16
+        .long 0x000002AA, 0x000002AA, 0x000002AA, 0x000002AA    /* _iBias */
+        .align 16
+        .long 0x80100000, 0x80100000, 0x80100000, 0x80100000    /* _iSub */
+        .align 16
+        .long 0xffdfffff, 0xffdfffff, 0xffdfffff, 0xffdfffff    /* _iCmp */
+        .align 16
+        .type	__svml_dcbrt_data_internal,@object
+        .size	__svml_dcbrt_data_internal,.-__svml_dcbrt_data_internal
diff --git a/sysdeps/x86_64/fpu/multiarch/svml_d_cbrt4_core-sse.S b/sysdeps/x86_64/fpu/multiarch/svml_d_cbrt4_core-sse.S
new file mode 100644
index 0000000000..3b54f31fbc
--- /dev/null
+++ b/sysdeps/x86_64/fpu/multiarch/svml_d_cbrt4_core-sse.S
@@ -0,0 +1,20 @@
+/* SSE version of vectorized cbrt, 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
+   <https://www.gnu.org/licenses/>.  */
+
+#define _ZGVdN4v_cbrt _ZGVdN4v_cbrt_sse_wrapper
+#include "../svml_d_cbrt4_core.S"
diff --git a/sysdeps/x86_64/fpu/multiarch/svml_d_cbrt4_core.c b/sysdeps/x86_64/fpu/multiarch/svml_d_cbrt4_core.c
new file mode 100644
index 0000000000..0b135877aa
--- /dev/null
+++ b/sysdeps/x86_64/fpu/multiarch/svml_d_cbrt4_core.c
@@ -0,0 +1,27 @@
+/* Multiple versions of vectorized cbrt, 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
+   <https://www.gnu.org/licenses/>.  */
+
+#define SYMBOL_NAME _ZGVdN4v_cbrt
+#include "ifunc-mathvec-avx2.h"
+
+libc_ifunc_redirected (REDIRECT_NAME, SYMBOL_NAME, IFUNC_SELECTOR ());
+
+#ifdef SHARED
+__hidden_ver1 (_ZGVdN4v_cbrt, __GI__ZGVdN4v_cbrt, __redirect__ZGVdN4v_cbrt)
+  __attribute__ ((visibility ("hidden")));
+#endif
diff --git a/sysdeps/x86_64/fpu/multiarch/svml_d_cbrt4_core_avx2.S b/sysdeps/x86_64/fpu/multiarch/svml_d_cbrt4_core_avx2.S
new file mode 100644
index 0000000000..2223c5309f
--- /dev/null
+++ b/sysdeps/x86_64/fpu/multiarch/svml_d_cbrt4_core_avx2.S
@@ -0,0 +1,505 @@
+/* Function cbrt 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:
+ *
+ *   x=2^{3*k+j} * 1.b1 b2 ... b5 b6 ... b52
+ *   Let r=(x*2^{-3k-j} - 1.b1 b2 ... b5 1)* rcp[b1 b2 ..b5],
+ *   where rcp[b1 b2 .. b5]=1/(1.b1 b2 b3 b4 b5 1) in double precision
+ *   cbrt(2^j * 1. b1 b2 .. b5 1) is approximated as T[j][b1..b5]+D[j][b1..b5]
+ *   (T stores the high 53 bits, D stores the low order bits)
+ *   Result=2^k*T+(2^k*T*r)*P+2^k*D
+ *   where P=p1+p2*r+..+p8*r^7
+ *
+ */
+
+/* Offsets for data table __svml_dcbrt_data_internal
+ */
+#define _dRcp                         	0
+#define _dCbrtHiLo                    	256
+#define _dA7                          	1024
+#define _dA6                          	1056
+#define _dA5                          	1088
+#define _dA4                          	1120
+#define _dA3                          	1152
+#define _dA2                          	1184
+#define _dA1                          	1216
+#define _dNeg65Div64                  	1248
+#define _dSgnf6Mask                   	1280
+#define _dNegOne                      	1312
+#define _dMantissaMask                	1344
+#define _lExpHiMask                   	1376
+#define _lExpLoMask                   	1408
+#define _l1556                        	1440
+#define _iRcpIndexMask                	1472
+#define _iAbsMask                     	1504
+#define _iSignMask                    	1536
+#define _iBias                        	1568
+#define _iSub                         	1600
+#define _iCmp                         	1632
+
+#include <sysdep.h>
+
+        .text
+	.section .text.avx2,"ax",@progbits
+ENTRY(_ZGVdN4v_cbrt_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
+
+/* Load 1/(1+iRcpIndex/32+1/64) reciprocal table value */
+        lea       __svml_dcbrt_data_internal(%rip), %rax
+        vmovapd   %ymm0, %ymm5
+
+/*
+ * Declarations
+ * Load constants
+ * Get iX - high part of argument
+ */
+        vextractf128 $1, %ymm5, %xmm6
+
+/* Calculate CbrtIndex */
+        vpsrlq    $52, %ymm5, %ymm15
+        vshufps   $221, %xmm6, %xmm5, %xmm4
+
+/* Calculate Rcp table index */
+        vandps    _iRcpIndexMask+__svml_dcbrt_data_internal(%rip), %xmm4, %xmm10
+        vpsrld    $12, %xmm10, %xmm3
+        vmovd     %xmm3, %ecx
+
+/* If the exponent field is zero - go to callout to process denormals */
+        vandps    _iAbsMask+__svml_dcbrt_data_internal(%rip), %xmm4, %xmm7
+
+/* Compute 2^k */
+        vpsrld    $20, %xmm4, %xmm4
+        vpsubd    _iSub+__svml_dcbrt_data_internal(%rip), %xmm7, %xmm8
+        vandps    _lExpLoMask+__svml_dcbrt_data_internal(%rip), %ymm15, %ymm0
+        vpmuludq  _l1556+__svml_dcbrt_data_internal(%rip), %ymm0, %ymm6
+        vpextrd   $2, %xmm3, %edi
+        movslq    %ecx, %rcx
+        vpextrd   $1, %xmm3, %esi
+        movslq    %edi, %rdi
+        vpextrd   $3, %xmm3, %r8d
+        movslq    %esi, %rsi
+        movslq    %r8d, %r8
+        vpcmpgtd  _iCmp+__svml_dcbrt_data_internal(%rip), %xmm8, %xmm9
+        vmovsd    (%rax,%rcx), %xmm11
+        vmovmskps %xmm9, %edx
+        vmovsd    (%rax,%rdi), %xmm13
+        vmovhpd   (%rax,%rsi), %xmm11, %xmm12
+        vmovhpd   (%rax,%r8), %xmm13, %xmm14
+        vextractf128 $1, %ymm6, %xmm7
+        vshufps   $136, %xmm7, %xmm6, %xmm8
+        vmovups   __VUNPACK_ODD_ind1.613.0.1(%rip), %ymm7
+        vextractf128 $1, %ymm0, %xmm1
+        vshufps   $136, %xmm1, %xmm0, %xmm9
+        vpsrld    $14, %xmm8, %xmm1
+        vpsubd    %xmm1, %xmm9, %xmm10
+        vpaddd    %xmm1, %xmm1, %xmm11
+
+/*
+ * VAND( L, l2k, = l2k, lExpHiMask );
+ * Argument reduction Z
+ */
+        vandpd    _dMantissaMask+__svml_dcbrt_data_internal(%rip), %ymm5, %ymm9
+        vinsertf128 $1, %xmm14, %ymm12, %ymm2
+        vpsubd    %xmm11, %xmm10, %xmm12
+        vpslld    $8, %xmm12, %xmm13
+        vpaddd    %xmm13, %xmm3, %xmm15
+
+/* Load cbrt(2^j*(1+iRcpIndex/32+1/64)) Hi & Lo values */
+        vmovd     %xmm15, %r9d
+        vpextrd   $2, %xmm15, %r11d
+        movslq    %r9d, %r9
+        vpextrd   $1, %xmm15, %r10d
+        movslq    %r11d, %r11
+        vpextrd   $3, %xmm15, %ecx
+        movslq    %r10d, %r10
+        movslq    %ecx, %rcx
+        vmovsd    256(%rax,%r9), %xmm3
+        vmovsd    256(%rax,%r11), %xmm0
+        vandpd    _dSgnf6Mask+__svml_dcbrt_data_internal(%rip), %ymm5, %ymm10
+        vmovhpd   256(%rax,%r10), %xmm3, %xmm14
+        vmovhpd   256(%rax,%rcx), %xmm0, %xmm3
+        vorpd     _dNegOne+__svml_dcbrt_data_internal(%rip), %ymm9, %ymm11
+        vorpd     _dNeg65Div64+__svml_dcbrt_data_internal(%rip), %ymm10, %ymm12
+        vsubpd    %ymm12, %ymm11, %ymm13
+        vmulpd    %ymm13, %ymm2, %ymm2
+        vinsertf128 $1, %xmm3, %ymm14, %ymm0
+        vpand     _iSignMask+__svml_dcbrt_data_internal(%rip), %xmm4, %xmm3
+        vpor      _iBias+__svml_dcbrt_data_internal(%rip), %xmm3, %xmm4
+        vpaddd    %xmm1, %xmm4, %xmm1
+        vpslld    $20, %xmm1, %xmm6
+
+/* Polynomial */
+        vmovupd   _dA7+__svml_dcbrt_data_internal(%rip), %ymm1
+        vfmadd213pd _dA6+__svml_dcbrt_data_internal(%rip), %ymm2, %ymm1
+        vfmadd213pd _dA5+__svml_dcbrt_data_internal(%rip), %ymm2, %ymm1
+        vfmadd213pd _dA4+__svml_dcbrt_data_internal(%rip), %ymm2, %ymm1
+        vfmadd213pd _dA3+__svml_dcbrt_data_internal(%rip), %ymm2, %ymm1
+        vfmadd213pd _dA2+__svml_dcbrt_data_internal(%rip), %ymm2, %ymm1
+        vfmadd213pd _dA1+__svml_dcbrt_data_internal(%rip), %ymm2, %ymm1
+        vpermps   %ymm6, %ymm7, %ymm8
+        vandps    __VUNPACK_ODD_mask.613.0.1(%rip), %ymm8, %ymm14
+
+/* THi*2^k, TLo*2^k */
+        vmulpd    %ymm14, %ymm0, %ymm0
+
+/* THi*2^k*Z */
+        vmulpd    %ymm0, %ymm2, %ymm2
+
+/* Final reconstruction */
+        vmulpd    %ymm2, %ymm1, %ymm3
+        vaddpd    %ymm3, %ymm0, %ymm0
+        testl     %edx, %edx
+
+/* Go to special inputs processing branch */
+        jne       L(SPECIAL_VALUES_BRANCH)
+                                # LOE rbx r12 r13 r14 r15 edx ymm0 ymm5
+
+/* 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   %ymm5, 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      cbrt@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_cbrt_avx2)
+        .section .rodata, "a"
+        .align 32
+
+__VUNPACK_ODD_ind1.613.0.1:
+	.rept	3
+        .long	0
+	.endr
+        .long	1
+        .long	0
+        .long	2
+        .long	0
+        .long	3
+        .align 32
+
+__VUNPACK_ODD_mask.613.0.1:
+        .long	0
+        .long	-1
+        .long	0
+        .long	-1
+        .long	0
+        .long	-1
+        .long	0
+        .long	-1
+
+        .section .rodata, "a"
+        .align 32
+
+#ifdef __svml_dcbrt_data_internal_typedef
+typedef unsigned int VUINT32;
+typedef struct
+{
+        __declspec(align(32)) VUINT32 _dRcp[32][2];
+        __declspec(align(32)) VUINT32 _dCbrtHiLo[96][2];
+        __declspec(align(32)) VUINT32 _dA7[4][2];
+        __declspec(align(32)) VUINT32 _dA6[4][2];
+        __declspec(align(32)) VUINT32 _dA5[4][2];
+        __declspec(align(32)) VUINT32 _dA4[4][2];
+        __declspec(align(32)) VUINT32 _dA3[4][2];
+        __declspec(align(32)) VUINT32 _dA2[4][2];
+        __declspec(align(32)) VUINT32 _dA1[4][2];
+        __declspec(align(32)) VUINT32 _dNeg65Div64[4][2];
+        __declspec(align(32)) VUINT32 _dSgnf6Mask[4][2];
+        __declspec(align(32)) VUINT32 _dNegOne[4][2];
+        __declspec(align(32)) VUINT32 _dMantissaMask[4][2];
+        __declspec(align(32)) VUINT32 _lExpHiMask[4][2];
+        __declspec(align(32)) VUINT32 _lExpLoMask[4][2];
+        __declspec(align(32)) VUINT32 _l1556[4][2];
+        __declspec(align(32)) VUINT32 _iRcpIndexMask[8][1];
+        __declspec(align(32)) VUINT32 _iAbsMask[8][1];
+        __declspec(align(32)) VUINT32 _iSignMask[8][1];
+        __declspec(align(32)) VUINT32 _iBias[8][1];
+        __declspec(align(32)) VUINT32 _iSub[8][1];
+        __declspec(align(32)) VUINT32 _iCmp[8][1];
+} __svml_dcbrt_data_internal;
+#endif
+__svml_dcbrt_data_internal:
+        /*== _dRcp ==*/
+        .quad 0xBFEF81F81F81F820  /* (1/(1+0/32+1/64)) = -.984615 */
+        .quad 0xBFEE9131ABF0B767  /* (1/(1+1/32+1/64)) = -.955224 */
+        .quad 0xBFEDAE6076B981DB  /* (1/(1+2/32+1/64)) = -.927536 */
+        .quad 0xBFECD85689039B0B  /* (1/(1+3/32+1/64)) = -.901408 */
+        .quad 0xBFEC0E070381C0E0  /* (1/(1+4/32+1/64)) = -.876712 */
+        .quad 0xBFEB4E81B4E81B4F  /* (1/(1+5/32+1/64)) = -.853333 */
+        .quad 0xBFEA98EF606A63BE  /* (1/(1+6/32+1/64)) = -.831169 */
+        .quad 0xBFE9EC8E951033D9  /* (1/(1+7/32+1/64)) = -.810127 */
+        .quad 0xBFE948B0FCD6E9E0  /* (1/(1+8/32+1/64)) = -.790123 */
+        .quad 0xBFE8ACB90F6BF3AA  /* (1/(1+9/32+1/64)) = -.771084 */
+        .quad 0xBFE8181818181818  /* (1/(1+10/32+1/64)) = -.752941 */
+        .quad 0xBFE78A4C8178A4C8  /* (1/(1+11/32+1/64)) = -.735632 */
+        .quad 0xBFE702E05C0B8170  /* (1/(1+12/32+1/64)) = -.719101 */
+        .quad 0xBFE6816816816817  /* (1/(1+13/32+1/64)) = -.703297 */
+        .quad 0xBFE6058160581606  /* (1/(1+14/32+1/64)) = -.688172 */
+        .quad 0xBFE58ED2308158ED  /* (1/(1+15/32+1/64)) = -.673684 */
+        .quad 0xBFE51D07EAE2F815  /* (1/(1+16/32+1/64)) = -.659794 */
+        .quad 0xBFE4AFD6A052BF5B  /* (1/(1+17/32+1/64)) = -.646465 */
+        .quad 0xBFE446F86562D9FB  /* (1/(1+18/32+1/64)) = -.633663 */
+        .quad 0xBFE3E22CBCE4A902  /* (1/(1+19/32+1/64)) = -.621359 */
+        .quad 0xBFE3813813813814  /* (1/(1+20/32+1/64)) = -.609524 */
+        .quad 0xBFE323E34A2B10BF  /* (1/(1+21/32+1/64)) = -.598131 */
+        .quad 0xBFE2C9FB4D812CA0  /* (1/(1+22/32+1/64)) = -.587156 */
+        .quad 0xBFE27350B8812735  /* (1/(1+23/32+1/64)) = -.576577 */
+        .quad 0xBFE21FB78121FB78  /* (1/(1+24/32+1/64)) = -.566372 */
+        .quad 0xBFE1CF06ADA2811D  /* (1/(1+25/32+1/64)) = -.556522 */
+        .quad 0xBFE1811811811812  /* (1/(1+26/32+1/64)) = -.547009 */
+        .quad 0xBFE135C81135C811  /* (1/(1+27/32+1/64)) = -.537815 */
+        .quad 0xBFE0ECF56BE69C90  /* (1/(1+28/32+1/64)) = -.528926 */
+        .quad 0xBFE0A6810A6810A7  /* (1/(1+29/32+1/64)) = -.520325 */
+        .quad 0xBFE0624DD2F1A9FC  /* (1/(1+30/32+1/64)) = -.512    */
+        .quad 0xBFE0204081020408  /* (1/(1+31/32+1/64)) = -.503937 */
+        /*== _dCbrtHiLo ==*/
+        .align 32
+        .quad 0x3FF01539221D4C97    /* HI((2^0*(1+0/32+1/64))^(1/3)) = 1.005181 */
+        .quad 0x3FF03F06771A2E33    /* HI((2^0*(1+1/32+1/64))^(1/3)) = 1.015387 */
+        .quad 0x3FF06800E629D671    /* HI((2^0*(1+2/32+1/64))^(1/3)) = 1.025391 */
+        .quad 0x3FF090328731DEB2    /* HI((2^0*(1+3/32+1/64))^(1/3)) = 1.035204 */
+        .quad 0x3FF0B7A4B1BD64AC    /* HI((2^0*(1+4/32+1/64))^(1/3)) = 1.044835 */
+        .quad 0x3FF0DE601024FB87    /* HI((2^0*(1+5/32+1/64))^(1/3)) = 1.054291 */
+        .quad 0x3FF1046CB0597000    /* HI((2^0*(1+6/32+1/64))^(1/3)) = 1.06358  */
+        .quad 0x3FF129D212A9BA9B    /* HI((2^0*(1+7/32+1/64))^(1/3)) = 1.07271  */
+        .quad 0x3FF14E9736CDAF38    /* HI((2^0*(1+8/32+1/64))^(1/3)) = 1.081687 */
+        .quad 0x3FF172C2A772F507    /* HI((2^0*(1+9/32+1/64))^(1/3)) = 1.090518 */
+        .quad 0x3FF1965A848001D3    /* HI((2^0*(1+10/32+1/64))^(1/3)) = 1.099207 */
+        .quad 0x3FF1B9648C38C55D    /* HI((2^0*(1+11/32+1/64))^(1/3)) = 1.107762 */
+        .quad 0x3FF1DBE6236A0C45    /* HI((2^0*(1+12/32+1/64))^(1/3)) = 1.116186 */
+        .quad 0x3FF1FDE45CBB1F9F    /* HI((2^0*(1+13/32+1/64))^(1/3)) = 1.124485 */
+        .quad 0x3FF21F63FF409042    /* HI((2^0*(1+14/32+1/64))^(1/3)) = 1.132664 */
+        .quad 0x3FF240698C6746E5    /* HI((2^0*(1+15/32+1/64))^(1/3)) = 1.140726 */
+        .quad 0x3FF260F9454BB99B    /* HI((2^0*(1+16/32+1/64))^(1/3)) = 1.148675 */
+        .quad 0x3FF281172F8E7073    /* HI((2^0*(1+17/32+1/64))^(1/3)) = 1.156516 */
+        .quad 0x3FF2A0C719B4B6D0    /* HI((2^0*(1+18/32+1/64))^(1/3)) = 1.164252 */
+        .quad 0x3FF2C00C9F2263EC    /* HI((2^0*(1+19/32+1/64))^(1/3)) = 1.171887 */
+        .quad 0x3FF2DEEB2BB7FB78    /* HI((2^0*(1+20/32+1/64))^(1/3)) = 1.179423 */
+        .quad 0x3FF2FD65FF1EFBBC    /* HI((2^0*(1+21/32+1/64))^(1/3)) = 1.186865 */
+        .quad 0x3FF31B802FCCF6A2    /* HI((2^0*(1+22/32+1/64))^(1/3)) = 1.194214 */
+        .quad 0x3FF3393CADC50708    /* HI((2^0*(1+23/32+1/64))^(1/3)) = 1.201474 */
+        .quad 0x3FF3569E451E4C2A    /* HI((2^0*(1+24/32+1/64))^(1/3)) = 1.208647 */
+        .quad 0x3FF373A7A0554CDE    /* HI((2^0*(1+25/32+1/64))^(1/3)) = 1.215736 */
+        .quad 0x3FF3905B4A6D76CE    /* HI((2^0*(1+26/32+1/64))^(1/3)) = 1.222743 */
+        .quad 0x3FF3ACBBB0E756B6    /* HI((2^0*(1+27/32+1/64))^(1/3)) = 1.229671 */
+        .quad 0x3FF3C8CB258FA340    /* HI((2^0*(1+28/32+1/64))^(1/3)) = 1.236522 */
+        .quad 0x3FF3E48BE02AC0CE    /* HI((2^0*(1+29/32+1/64))^(1/3)) = 1.243297 */
+        .quad 0x3FF4000000000000    /* HI((2^0*(1+30/32+1/64))^(1/3)) = 1.25     */
+        .quad 0x3FF41B298D47800E    /* HI((2^0*(1+31/32+1/64))^(1/3)) = 1.256631 */
+        .quad 0x3FF443604B34D9B2    /* HI((2^1*(1+0/32+1/64))^(1/3)) = 1.266449 */
+        .quad 0x3FF4780B20906571    /* HI((2^1*(1+1/32+1/64))^(1/3)) = 1.279307 */
+        .quad 0x3FF4ABAC3EE06706    /* HI((2^1*(1+2/32+1/64))^(1/3)) = 1.291912 */
+        .quad 0x3FF4DE505DA66B8D    /* HI((2^1*(1+3/32+1/64))^(1/3)) = 1.304276 */
+        .quad 0x3FF51003420A5C07    /* HI((2^1*(1+4/32+1/64))^(1/3)) = 1.316409 */
+        .quad 0x3FF540CFD6FD11C1    /* HI((2^1*(1+5/32+1/64))^(1/3)) = 1.328323 */
+        .quad 0x3FF570C04260716B    /* HI((2^1*(1+6/32+1/64))^(1/3)) = 1.340027 */
+        .quad 0x3FF59FDDF7A45F38    /* HI((2^1*(1+7/32+1/64))^(1/3)) = 1.35153  */
+        .quad 0x3FF5CE31C83539DF    /* HI((2^1*(1+8/32+1/64))^(1/3)) = 1.36284  */
+        .quad 0x3FF5FBC3F20966A4    /* HI((2^1*(1+9/32+1/64))^(1/3)) = 1.373966 */
+        .quad 0x3FF6289C2C8F1B70    /* HI((2^1*(1+10/32+1/64))^(1/3)) = 1.384915 */
+        .quad 0x3FF654C1B4316DCF    /* HI((2^1*(1+11/32+1/64))^(1/3)) = 1.395693 */
+        .quad 0x3FF6803B54A34E44    /* HI((2^1*(1+12/32+1/64))^(1/3)) = 1.406307 */
+        .quad 0x3FF6AB0F72182659    /* HI((2^1*(1+13/32+1/64))^(1/3)) = 1.416763 */
+        .quad 0x3FF6D544118C08BC    /* HI((2^1*(1+14/32+1/64))^(1/3)) = 1.427067 */
+        .quad 0x3FF6FEDEE0388D4A    /* HI((2^1*(1+15/32+1/64))^(1/3)) = 1.437224 */
+        .quad 0x3FF727E53A4F645E    /* HI((2^1*(1+16/32+1/64))^(1/3)) = 1.44724  */
+        .quad 0x3FF7505C31104114    /* HI((2^1*(1+17/32+1/64))^(1/3)) = 1.457119 */
+        .quad 0x3FF77848904CD549    /* HI((2^1*(1+18/32+1/64))^(1/3)) = 1.466866 */
+        .quad 0x3FF79FAEE36B2534    /* HI((2^1*(1+19/32+1/64))^(1/3)) = 1.476485 */
+        .quad 0x3FF7C69379F4605B    /* HI((2^1*(1+20/32+1/64))^(1/3)) = 1.48598  */
+        .quad 0x3FF7ECFA6BBCA391    /* HI((2^1*(1+21/32+1/64))^(1/3)) = 1.495356 */
+        .quad 0x3FF812E79CAE7EB9    /* HI((2^1*(1+22/32+1/64))^(1/3)) = 1.504615 */
+        .quad 0x3FF8385EC043C71D    /* HI((2^1*(1+23/32+1/64))^(1/3)) = 1.513762 */
+        .quad 0x3FF85D635CB41B9D    /* HI((2^1*(1+24/32+1/64))^(1/3)) = 1.5228   */
+        .quad 0x3FF881F8CDE083DB    /* HI((2^1*(1+25/32+1/64))^(1/3)) = 1.531731 */
+        .quad 0x3FF8A6224802B8A8    /* HI((2^1*(1+26/32+1/64))^(1/3)) = 1.54056  */
+        .quad 0x3FF8C9E2DA25E5E4    /* HI((2^1*(1+27/32+1/64))^(1/3)) = 1.549289 */
+        .quad 0x3FF8ED3D706E1010    /* HI((2^1*(1+28/32+1/64))^(1/3)) = 1.55792  */
+        .quad 0x3FF91034D632B6DF    /* HI((2^1*(1+29/32+1/64))^(1/3)) = 1.566457 */
+        .quad 0x3FF932CBB7F0CF2D    /* HI((2^1*(1+30/32+1/64))^(1/3)) = 1.574901 */
+        .quad 0x3FF95504A517BF3A    /* HI((2^1*(1+31/32+1/64))^(1/3)) = 1.583256 */
+        .quad 0x3FF987AF34F8BB19    /* HI((2^2*(1+0/32+1/64))^(1/3)) = 1.595626 */
+        .quad 0x3FF9CA0A8337B317    /* HI((2^2*(1+1/32+1/64))^(1/3)) = 1.611826 */
+        .quad 0x3FFA0B1709CC13D5    /* HI((2^2*(1+2/32+1/64))^(1/3)) = 1.627708 */
+        .quad 0x3FFA4AE4CE6419ED    /* HI((2^2*(1+3/32+1/64))^(1/3)) = 1.643285 */
+        .quad 0x3FFA8982A5567031    /* HI((2^2*(1+4/32+1/64))^(1/3)) = 1.658572 */
+        .quad 0x3FFAC6FE500AB570    /* HI((2^2*(1+5/32+1/64))^(1/3)) = 1.673582 */
+        .quad 0x3FFB036497A15A17    /* HI((2^2*(1+6/32+1/64))^(1/3)) = 1.688328 */
+        .quad 0x3FFB3EC164671755    /* HI((2^2*(1+7/32+1/64))^(1/3)) = 1.702821 */
+        .quad 0x3FFB791FD288C46F    /* HI((2^2*(1+8/32+1/64))^(1/3)) = 1.717071 */
+        .quad 0x3FFBB28A44693BE4    /* HI((2^2*(1+9/32+1/64))^(1/3)) = 1.731089 */
+        .quad 0x3FFBEB0A72EB6E31    /* HI((2^2*(1+10/32+1/64))^(1/3)) = 1.744883 */
+        .quad 0x3FFC22A97BF5F697    /* HI((2^2*(1+11/32+1/64))^(1/3)) = 1.758462 */
+        .quad 0x3FFC596FEF6AF983    /* HI((2^2*(1+12/32+1/64))^(1/3)) = 1.771835 */
+        .quad 0x3FFC8F65DAC655A3    /* HI((2^2*(1+13/32+1/64))^(1/3)) = 1.785009 */
+        .quad 0x3FFCC492D38CE8D9    /* HI((2^2*(1+14/32+1/64))^(1/3)) = 1.797992 */
+        .quad 0x3FFCF8FE00B19367    /* HI((2^2*(1+15/32+1/64))^(1/3)) = 1.810789 */
+        .quad 0x3FFD2CAE230F8709    /* HI((2^2*(1+16/32+1/64))^(1/3)) = 1.823408 */
+        .quad 0x3FFD5FA99D15208F    /* HI((2^2*(1+17/32+1/64))^(1/3)) = 1.835855 */
+        .quad 0x3FFD91F679B6E505    /* HI((2^2*(1+18/32+1/64))^(1/3)) = 1.848135 */
+        .quad 0x3FFDC39A72BF2302    /* HI((2^2*(1+19/32+1/64))^(1/3)) = 1.860255 */
+        .quad 0x3FFDF49AF68C1570    /* HI((2^2*(1+20/32+1/64))^(1/3)) = 1.872218 */
+        .quad 0x3FFE24FD2D4C23B8    /* HI((2^2*(1+21/32+1/64))^(1/3)) = 1.884031 */
+        .quad 0x3FFE54C5FDC5EC73    /* HI((2^2*(1+22/32+1/64))^(1/3)) = 1.895697 */
+        .quad 0x3FFE83FA11B81DBB    /* HI((2^2*(1+23/32+1/64))^(1/3)) = 1.907221 */
+        .quad 0x3FFEB29DD9DBAF25    /* HI((2^2*(1+24/32+1/64))^(1/3)) = 1.918608 */
+        .quad 0x3FFEE0B59191D374    /* HI((2^2*(1+25/32+1/64))^(1/3)) = 1.929861 */
+        .quad 0x3FFF0E454245E4BF    /* HI((2^2*(1+26/32+1/64))^(1/3)) = 1.940984 */
+        .quad 0x3FFF3B50C68A9DD3    /* HI((2^2*(1+27/32+1/64))^(1/3)) = 1.951981 */
+        .quad 0x3FFF67DBCCF922DC    /* HI((2^2*(1+28/32+1/64))^(1/3)) = 1.962856 */
+        .quad 0x3FFF93E9DAD7A4A6    /* HI((2^2*(1+29/32+1/64))^(1/3)) = 1.973612 */
+        .quad 0x3FFFBF7E4E8CC9CB    /* HI((2^2*(1+30/32+1/64))^(1/3)) = 1.984251 */
+        .quad 0x3FFFEA9C61E47CD3    /* HI((2^2*(1+31/32+1/64))^(1/3)) = 1.994778 */
+        .align 32
+        .quad 0x3F93750AD588F115, 0x3F93750AD588F115, 0x3F93750AD588F115, 0x3F93750AD588F115      /* _dA7 */
+        .align 32
+        .quad 0xBF98090D6221A247, 0xBF98090D6221A247, 0xBF98090D6221A247, 0xBF98090D6221A247      /* _dA6 */
+        .align 32
+        .quad 0x3F9EE7113506AC12, 0x3F9EE7113506AC12, 0x3F9EE7113506AC12, 0x3F9EE7113506AC12      /* _dA5 */
+        .align 32
+        .quad 0xBFA511E8D2B3183B, 0xBFA511E8D2B3183B, 0xBFA511E8D2B3183B, 0xBFA511E8D2B3183B      /* _dA4 */
+        .align 32
+        .quad 0x3FAF9ADD3C0CA458, 0x3FAF9ADD3C0CA458, 0x3FAF9ADD3C0CA458, 0x3FAF9ADD3C0CA458      /* _dA3 */
+        .align 32
+        .quad 0xBFBC71C71C71C71C, 0xBFBC71C71C71C71C, 0xBFBC71C71C71C71C, 0xBFBC71C71C71C71C      /* _dA2 */
+        .align 32
+        .quad 0x3FD5555555555555, 0x3FD5555555555555, 0x3FD5555555555555, 0x3FD5555555555555      /* _dA1 */
+        .align 32
+        .quad 0xBFF0400000000000, 0xBFF0400000000000, 0xBFF0400000000000, 0xBFF0400000000000        /* _dNeg65Div64 */
+        .align 32
+        .quad 0x000FC00000000000, 0x000FC00000000000, 0x000FC00000000000, 0x000FC00000000000        /* _dSgnf6Mask */
+        .align 32
+        .quad 0xBFF0000000000000, 0xBFF0000000000000, 0xBFF0000000000000, 0xBFF0000000000000        /* _dNegOne */
+        .align 32
+        .quad 0x000FFFFFFFFFFFFF, 0x000FFFFFFFFFFFFF, 0x000FFFFFFFFFFFFF, 0x000FFFFFFFFFFFFF        /* _dMantissaMask */
+        .align 32
+        .quad 0xFFF0000000000000, 0xFFF0000000000000, 0xFFF0000000000000, 0xFFF0000000000000        /* _lExpHiMask */
+        .align 32
+        .quad 0x00000000000007FF, 0x00000000000007FF, 0x00000000000007FF, 0x00000000000007FF        /* _lExpLoMask */
+        .align 32
+        .quad 0x0000000000001556, 0x0000000000001556, 0x0000000000001556, 0x0000000000001556        /* _l1556 */
+        .align 32
+        .long 0x000F8000, 0x000F8000, 0x000F8000, 0x000F8000, 0x000F8000, 0x000F8000, 0x000F8000, 0x000F8000    /* _iRcpIndexMask */
+        .align 32
+        .long 0x7FFFFFFF, 0x7FFFFFFF, 0x7FFFFFFF, 0x7FFFFFFF, 0x7FFFFFFF, 0x7FFFFFFF, 0x7FFFFFFF, 0x7FFFFFFF    /* _iAbsMask */
+        .align 32
+        .long 0x00000800, 0x00000800, 0x00000800, 0x00000800, 0x00000800, 0x00000800, 0x00000800, 0x00000800    /* _iSignMask */
+        .align 32
+        .long 0x000002AA, 0x000002AA, 0x000002AA, 0x000002AA, 0x000002AA, 0x000002AA, 0x000002AA, 0x000002AA    /* _iBias */
+        .align 32
+        .long 0x80100000, 0x80100000, 0x80100000, 0x80100000, 0x80100000, 0x80100000, 0x80100000, 0x80100000    /* _iSub */
+        .align 32
+        .long 0xffdfffff, 0xffdfffff, 0xffdfffff, 0xffdfffff, 0xffdfffff, 0xffdfffff, 0xffdfffff, 0xffdfffff    /* _iCmp */
+        .align 32
+        .type	__svml_dcbrt_data_internal,@object
+        .size	__svml_dcbrt_data_internal,.-__svml_dcbrt_data_internal
diff --git a/sysdeps/x86_64/fpu/multiarch/svml_d_cbrt8_core-avx2.S b/sysdeps/x86_64/fpu/multiarch/svml_d_cbrt8_core-avx2.S
new file mode 100644
index 0000000000..3831e582ce
--- /dev/null
+++ b/sysdeps/x86_64/fpu/multiarch/svml_d_cbrt8_core-avx2.S
@@ -0,0 +1,20 @@
+/* AVX2 version of vectorized cbrt, 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
+   <https://www.gnu.org/licenses/>.  */
+
+#define _ZGVeN8v_cbrt _ZGVeN8v_cbrt_avx2_wrapper
+#include "../svml_d_cbrt8_core.S"
diff --git a/sysdeps/x86_64/fpu/multiarch/svml_d_cbrt8_core.c b/sysdeps/x86_64/fpu/multiarch/svml_d_cbrt8_core.c
new file mode 100644
index 0000000000..28c147216f
--- /dev/null
+++ b/sysdeps/x86_64/fpu/multiarch/svml_d_cbrt8_core.c
@@ -0,0 +1,27 @@
+/* Multiple versions of vectorized cbrt, 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
+   <https://www.gnu.org/licenses/>.  */
+
+#define SYMBOL_NAME _ZGVeN8v_cbrt
+#include "ifunc-mathvec-avx512-skx.h"
+
+libc_ifunc_redirected (REDIRECT_NAME, SYMBOL_NAME, IFUNC_SELECTOR ());
+
+#ifdef SHARED
+__hidden_ver1 (_ZGVeN8v_cbrt, __GI__ZGVeN8v_cbrt, __redirect__ZGVeN8v_cbrt)
+  __attribute__ ((visibility ("hidden")));
+#endif
diff --git a/sysdeps/x86_64/fpu/multiarch/svml_d_cbrt8_core_avx512.S b/sysdeps/x86_64/fpu/multiarch/svml_d_cbrt8_core_avx512.S
new file mode 100644
index 0000000000..b9c071b54c
--- /dev/null
+++ b/sysdeps/x86_64/fpu/multiarch/svml_d_cbrt8_core_avx512.S
@@ -0,0 +1,253 @@
+/* Function cbrt 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:
+ *
+ *   x=2^{3*k+j} * 1.b1 b2 ... b5 b6 ... b52
+ *   Let r=(x*2^{-3k-j} - 1.b1 b2 ... b5 1)* rcp[b1 b2 ..b5],
+ *   where rcp[b1 b2 .. b5]=1/(1.b1 b2 b3 b4 b5 1) in double precision
+ *   cbrt(2^j * 1. b1 b2 .. b5 1) is approximated as T[j][b1..b5]+D[j][b1..b5]
+ *   (T stores the high 53 bits, D stores the low order bits)
+ *   Result=2^k*T+(2^k*T*r)*P+2^k*D
+ *   where P=p1+p2*r+..+p8*r^7
+ *
+ */
+
+/* Offsets for data table __svml_dcbrt_data_internal_avx512
+ */
+#define etbl_H                        	0
+#define etbl_L                        	64
+#define cbrt_tbl_H                    	128
+#define BiasL                         	256
+#define SZero                         	320
+#define OneThird                      	384
+#define Bias3                         	448
+#define Three                         	512
+#define One                           	576
+#define poly_coeff10                  	640
+#define poly_coeff9                   	704
+#define poly_coeff8                   	768
+#define poly_coeff7                   	832
+#define poly_coeff6                   	896
+#define poly_coeff5                   	960
+#define poly_coeff4                   	1024
+#define poly_coeff3                   	1088
+#define poly_coeff2                   	1152
+#define poly_coeff1                   	1216
+
+#include <sysdep.h>
+
+        .text
+	.section .text.evex512,"ax",@progbits
+ENTRY(_ZGVeN8v_cbrt_skx)
+        vgetmantpd $0, {sae}, %zmm0, %zmm14
+
+/* GetExp(x) */
+        vgetexppd {sae}, %zmm0, %zmm7
+        vmovups   BiasL+__svml_dcbrt_data_internal_avx512(%rip), %zmm8
+
+/* exponent/3 */
+        vmovups   OneThird+__svml_dcbrt_data_internal_avx512(%rip), %zmm9
+        vmovups   Bias3+__svml_dcbrt_data_internal_avx512(%rip), %zmm10
+
+/* Reduced argument: R = DblRcp*Mantissa - 1 */
+        vmovups   One+__svml_dcbrt_data_internal_avx512(%rip), %zmm2
+
+/* exponent%3 (to be used as index) */
+        vmovups   Three+__svml_dcbrt_data_internal_avx512(%rip), %zmm11
+
+/* DblRcp ~ 1/Mantissa */
+        vrcp14pd  %zmm14, %zmm13
+        vaddpd    {rn-sae}, %zmm8, %zmm7, %zmm12
+        vandpd    SZero+__svml_dcbrt_data_internal_avx512(%rip), %zmm0, %zmm6
+
+/* round DblRcp to 3 fractional bits (RN mode, no Precision exception) */
+        vrndscalepd $72, {sae}, %zmm13, %zmm15
+        vfmsub231pd {rn-sae}, %zmm12, %zmm9, %zmm10
+
+/* polynomial */
+        vmovups   poly_coeff10+__svml_dcbrt_data_internal_avx512(%rip), %zmm0
+        vmovups   poly_coeff8+__svml_dcbrt_data_internal_avx512(%rip), %zmm7
+        vmovups   poly_coeff7+__svml_dcbrt_data_internal_avx512(%rip), %zmm9
+        vfmsub231pd {rn-sae}, %zmm15, %zmm14, %zmm2
+        vrndscalepd $9, {sae}, %zmm10, %zmm5
+
+/* Table lookup */
+        vmovups   cbrt_tbl_H+__svml_dcbrt_data_internal_avx512(%rip), %zmm10
+        vmovups   poly_coeff6+__svml_dcbrt_data_internal_avx512(%rip), %zmm8
+        vmovups   poly_coeff3+__svml_dcbrt_data_internal_avx512(%rip), %zmm13
+        vfmadd231pd {rn-sae}, %zmm2, %zmm7, %zmm9
+        vfnmadd231pd {rn-sae}, %zmm5, %zmm11, %zmm12
+        vmovups   poly_coeff5+__svml_dcbrt_data_internal_avx512(%rip), %zmm11
+        vmovups   poly_coeff1+__svml_dcbrt_data_internal_avx512(%rip), %zmm14
+
+/* Prepare table index */
+        vpsrlq    $49, %zmm15, %zmm1
+
+/* Table lookup: 2^(exponent%3) */
+        vpermpd   __svml_dcbrt_data_internal_avx512(%rip), %zmm12, %zmm4
+        vpermpd   etbl_L+__svml_dcbrt_data_internal_avx512(%rip), %zmm12, %zmm3
+        vpermt2pd cbrt_tbl_H+64+__svml_dcbrt_data_internal_avx512(%rip), %zmm1, %zmm10
+        vmovups   poly_coeff9+__svml_dcbrt_data_internal_avx512(%rip), %zmm1
+        vfmadd231pd {rn-sae}, %zmm2, %zmm8, %zmm11
+        vmovups   poly_coeff2+__svml_dcbrt_data_internal_avx512(%rip), %zmm12
+        vscalefpd {rn-sae}, %zmm5, %zmm10, %zmm15
+        vfmadd231pd {rn-sae}, %zmm2, %zmm0, %zmm1
+        vmovups   poly_coeff4+__svml_dcbrt_data_internal_avx512(%rip), %zmm5
+        vfmadd231pd {rn-sae}, %zmm2, %zmm12, %zmm14
+        vmulpd    {rn-sae}, %zmm2, %zmm2, %zmm0
+        vfmadd231pd {rn-sae}, %zmm2, %zmm5, %zmm13
+
+/* Sh*R */
+        vmulpd    {rn-sae}, %zmm2, %zmm4, %zmm2
+        vfmadd213pd {rn-sae}, %zmm9, %zmm0, %zmm1
+        vfmadd213pd {rn-sae}, %zmm11, %zmm0, %zmm1
+        vfmadd213pd {rn-sae}, %zmm13, %zmm0, %zmm1
+        vfmadd213pd {rn-sae}, %zmm14, %zmm0, %zmm1
+
+/* Sl + (Sh*R)*Poly */
+        vfmadd213pd {rn-sae}, %zmm3, %zmm1, %zmm2
+
+/*
+ * branch-free
+ * scaled_Th*(Sh+Sl+Sh*R*Poly)
+ */
+        vaddpd    {rn-sae}, %zmm4, %zmm2, %zmm3
+        vmulpd    {rn-sae}, %zmm15, %zmm3, %zmm4
+        vorpd     %zmm6, %zmm4, %zmm0
+        ret
+
+END(_ZGVeN8v_cbrt_skx)
+
+        .section .rodata, "a"
+        .align 64
+
+#ifdef __svml_dcbrt_data_internal_avx512_typedef
+typedef unsigned int VUINT32;
+typedef struct {
+        __declspec(align(64)) VUINT32 etbl_H[8][2];
+        __declspec(align(64)) VUINT32 etbl_L[8][2];
+        __declspec(align(64)) VUINT32 cbrt_tbl_H[16][2];
+        __declspec(align(64)) VUINT32 BiasL[8][2];
+        __declspec(align(64)) VUINT32 SZero[8][2];
+        __declspec(align(64)) VUINT32 OneThird[8][2];
+        __declspec(align(64)) VUINT32 Bias3[8][2];
+        __declspec(align(64)) VUINT32 Three[8][2];
+        __declspec(align(64)) VUINT32 One[8][2];
+        __declspec(align(64)) VUINT32 poly_coeff10[8][2];
+        __declspec(align(64)) VUINT32 poly_coeff9[8][2];
+        __declspec(align(64)) VUINT32 poly_coeff8[8][2];
+        __declspec(align(64)) VUINT32 poly_coeff7[8][2];
+        __declspec(align(64)) VUINT32 poly_coeff6[8][2];
+        __declspec(align(64)) VUINT32 poly_coeff5[8][2];
+        __declspec(align(64)) VUINT32 poly_coeff4[8][2];
+        __declspec(align(64)) VUINT32 poly_coeff3[8][2];
+        __declspec(align(64)) VUINT32 poly_coeff2[8][2];
+        __declspec(align(64)) VUINT32 poly_coeff1[8][2];
+    } __svml_dcbrt_data_internal_avx512;
+#endif
+__svml_dcbrt_data_internal_avx512:
+        /*== etbl_H ==*/
+        .quad 0x3ff0000000000000
+        .quad 0x3ff428a2f98d728b
+        .quad 0x3ff965fea53d6e3d
+        .quad 0x0000000000000000
+        .quad 0xbff0000000000000
+        .quad 0xbff428a2f98d728b
+        .quad 0xbff965fea53d6e3d
+        .quad 0x0000000000000000
+        /*== etbl_L ==*/
+        .align 64
+        .quad 0x0000000000000000
+        .quad 0xbc7ddc22548ea41e
+        .quad 0xbc9f53e999952f09
+        .quad 0x0000000000000000
+        .quad 0x0000000000000000
+        .quad 0x3c7ddc22548ea41e
+        .quad 0x3c9f53e999952f09
+        .quad 0x0000000000000000
+        /*== cbrt_tbl_H ==*/
+        .align 64
+        .quad 0x3ff428a2f98d728b
+        .quad 0x3ff361f35ca116ff
+        .quad 0x3ff2b6b5edf6b54a
+        .quad 0x3ff220e6dd675180
+        .quad 0x3ff19c3b38e975a8
+        .quad 0x3ff12589c21fb842
+        .quad 0x3ff0ba6ee5f9aad4
+        .quad 0x3ff059123d3a9848
+        .quad 0x3ff0000000000000
+        .quad 0x0000000000000000
+        .quad 0x0000000000000000
+        .quad 0x0000000000000000
+        .quad 0x0000000000000000
+        .quad 0x0000000000000000
+        .quad 0x0000000000000000
+        .quad 0x0000000000000000
+        /*== BiasL ==*/
+        .align 64
+        .quad 0x4338000000000000, 0x4338000000000000, 0x4338000000000000, 0x4338000000000000, 0x4338000000000000, 0x4338000000000000, 0x4338000000000000, 0x4338000000000000
+        /*== Zero ==*/
+        .align 64
+        .quad 0x8000000000000000, 0x8000000000000000, 0x8000000000000000, 0x8000000000000000, 0x8000000000000000, 0x8000000000000000, 0x8000000000000000, 0x8000000000000000
+        /*== OneThird ==*/
+        .align 64
+        .quad 0x3fd5555555555556, 0x3fd5555555555556, 0x3fd5555555555556, 0x3fd5555555555556, 0x3fd5555555555556, 0x3fd5555555555556, 0x3fd5555555555556, 0x3fd5555555555556
+        /*== Bias3 ==*/
+        .align 64
+        .quad 0x4320000000000000, 0x4320000000000000, 0x4320000000000000, 0x4320000000000000, 0x4320000000000000, 0x4320000000000000, 0x4320000000000000, 0x4320000000000000
+        /*== Three ==*/
+        .align 64
+        .quad 0x4008000000000000, 0x4008000000000000, 0x4008000000000000, 0x4008000000000000, 0x4008000000000000, 0x4008000000000000, 0x4008000000000000, 0x4008000000000000
+        /*==One ==*/
+        .align 64
+        .quad 0x3ff0000000000000, 0x3ff0000000000000, 0x3ff0000000000000, 0x3ff0000000000000, 0x3ff0000000000000, 0x3ff0000000000000, 0x3ff0000000000000, 0x3ff0000000000000
+        /*== poly_coeff10 ==*/
+        .align 64
+        .quad 0xbf882e3b6adeca62, 0xbf882e3b6adeca62, 0xbf882e3b6adeca62, 0xbf882e3b6adeca62, 0xbf882e3b6adeca62, 0xbf882e3b6adeca62, 0xbf882e3b6adeca62, 0xbf882e3b6adeca62
+        /*== poly_coeff9 ==*/
+        .align 64
+        .quad 0x3f8bda24bae48875, 0x3f8bda24bae48875, 0x3f8bda24bae48875, 0x3f8bda24bae48875, 0x3f8bda24bae48875, 0x3f8bda24bae48875, 0x3f8bda24bae48875, 0x3f8bda24bae48875
+        /*== poly_coeff8 ==*/
+        .align 64
+        .quad 0xbf9036b87c71d55f, 0xbf9036b87c71d55f, 0xbf9036b87c71d55f, 0xbf9036b87c71d55f, 0xbf9036b87c71d55f, 0xbf9036b87c71d55f, 0xbf9036b87c71d55f, 0xbf9036b87c71d55f
+        /*== poly_coeff7 ==*/
+        .align 64
+        .quad 0x3f9374ed9398b914, 0x3f9374ed9398b914, 0x3f9374ed9398b914, 0x3f9374ed9398b914, 0x3f9374ed9398b914, 0x3f9374ed9398b914, 0x3f9374ed9398b914, 0x3f9374ed9398b914
+        /*== poly_coeff6 ==*/
+        .align 64
+        .quad 0xbf98090d77f2468e, 0xbf98090d77f2468e, 0xbf98090d77f2468e, 0xbf98090d77f2468e, 0xbf98090d77f2468e, 0xbf98090d77f2468e, 0xbf98090d77f2468e, 0xbf98090d77f2468e
+        /*== poly_coeff5 ==*/
+        .align 64
+        .quad 0x3f9ee71141dcf569, 0x3f9ee71141dcf569, 0x3f9ee71141dcf569, 0x3f9ee71141dcf569, 0x3f9ee71141dcf569, 0x3f9ee71141dcf569, 0x3f9ee71141dcf569, 0x3f9ee71141dcf569
+        /*== poly_coeff4 ==*/
+        .align 64
+        .quad 0xbfa511e8d2b0363e, 0xbfa511e8d2b0363e, 0xbfa511e8d2b0363e, 0xbfa511e8d2b0363e, 0xbfa511e8d2b0363e, 0xbfa511e8d2b0363e, 0xbfa511e8d2b0363e, 0xbfa511e8d2b0363e
+        /*== poly_coeff3 ==*/
+        .align 64
+        .quad 0x3faf9add3c0b7e31, 0x3faf9add3c0b7e31, 0x3faf9add3c0b7e31, 0x3faf9add3c0b7e31, 0x3faf9add3c0b7e31, 0x3faf9add3c0b7e31, 0x3faf9add3c0b7e31, 0x3faf9add3c0b7e31
+        /*== poly_coeff2 ==*/
+        .align 64
+        .quad 0xbfbc71c71c71c741, 0xbfbc71c71c71c741, 0xbfbc71c71c71c741, 0xbfbc71c71c71c741, 0xbfbc71c71c71c741, 0xbfbc71c71c71c741, 0xbfbc71c71c71c741, 0xbfbc71c71c71c741
+        /*== poly_coeff1 ==*/
+        .align 64
+        .quad 0x3fd5555555555557, 0x3fd5555555555557, 0x3fd5555555555557, 0x3fd5555555555557, 0x3fd5555555555557, 0x3fd5555555555557, 0x3fd5555555555557, 0x3fd5555555555557
+        .align 64
+        .type	__svml_dcbrt_data_internal_avx512,@object
+        .size	__svml_dcbrt_data_internal_avx512,.-__svml_dcbrt_data_internal_avx512
diff --git a/sysdeps/x86_64/fpu/multiarch/svml_s_cbrtf16_core-avx2.S b/sysdeps/x86_64/fpu/multiarch/svml_s_cbrtf16_core-avx2.S
new file mode 100644
index 0000000000..faa847fba6
--- /dev/null
+++ b/sysdeps/x86_64/fpu/multiarch/svml_s_cbrtf16_core-avx2.S
@@ -0,0 +1,20 @@
+/* AVX2 version of vectorized cbrtf.
+   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/>.  */
+
+#define _ZGVeN16v_cbrtf _ZGVeN16v_cbrtf_avx2_wrapper
+#include "../svml_s_cbrtf16_core.S"
diff --git a/sysdeps/x86_64/fpu/multiarch/svml_s_cbrtf16_core.c b/sysdeps/x86_64/fpu/multiarch/svml_s_cbrtf16_core.c
new file mode 100644
index 0000000000..785a68cc0d
--- /dev/null
+++ b/sysdeps/x86_64/fpu/multiarch/svml_s_cbrtf16_core.c
@@ -0,0 +1,28 @@
+/* Multiple versions of vectorized cbrtf, 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
+   <https://www.gnu.org/licenses/>.  */
+
+#define SYMBOL_NAME _ZGVeN16v_cbrtf
+#include "ifunc-mathvec-avx512-skx.h"
+
+libc_ifunc_redirected (REDIRECT_NAME, SYMBOL_NAME, IFUNC_SELECTOR ());
+
+#ifdef SHARED
+__hidden_ver1 (_ZGVeN16v_cbrtf, __GI__ZGVeN16v_cbrtf,
+	       __redirect__ZGVeN16v_cbrtf)
+  __attribute__ ((visibility ("hidden")));
+#endif
diff --git a/sysdeps/x86_64/fpu/multiarch/svml_s_cbrtf16_core_avx512.S b/sysdeps/x86_64/fpu/multiarch/svml_s_cbrtf16_core_avx512.S
new file mode 100644
index 0000000000..55b017682b
--- /dev/null
+++ b/sysdeps/x86_64/fpu/multiarch/svml_s_cbrtf16_core_avx512.S
@@ -0,0 +1,235 @@
+/* Function cbrtf 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:
+ *
+ *     x=2^{3*k+j} * 1.b1 b2 ... b5 b6 ... b52
+ *     Let r=(x*2^{-3k-j} - 1.b1 b2 ... b5 1)* rcp[b1 b2 ..b5],
+ *     where rcp[b1 b2 .. b5]=1/(1.b1 b2 b3 b4 b5 1) in single precision
+ *     cbrtf(2^j * 1. b1 b2 .. b5 1) is approximated as T[j][b1..b5]+D[j][b1..b5]
+ *     (T stores the high 24 bits, D stores the low order bits)
+ *     Result=2^k*T+(2^k*T*r)*P+2^k*D
+ *      where P=p1+p2*r+..
+ *
+ */
+
+/* Offsets for data table __svml_scbrt_data_internal_avx512
+ */
+#define etbl_H                        	0
+#define etbl_L                        	64
+#define cbrt_tbl_H                    	128
+#define BiasL                         	256
+#define SZero                         	320
+#define OneThird                      	384
+#define Bias3                         	448
+#define Three                         	512
+#define One                           	576
+#define poly_coeff3                   	640
+#define poly_coeff2                   	704
+#define poly_coeff1                   	768
+
+#include <sysdep.h>
+
+        .text
+	.section .text.exex512,"ax",@progbits
+ENTRY(_ZGVeN16v_cbrtf_skx)
+        vgetmantps $0, {sae}, %zmm0, %zmm8
+
+/* GetExp(x) */
+        vgetexpps {sae}, %zmm0, %zmm1
+        vmovups   BiasL+__svml_scbrt_data_internal_avx512(%rip), %zmm2
+
+/* exponent/3 */
+        vmovups   OneThird+__svml_scbrt_data_internal_avx512(%rip), %zmm3
+        vmovups   Bias3+__svml_scbrt_data_internal_avx512(%rip), %zmm4
+        vmovups   One+__svml_scbrt_data_internal_avx512(%rip), %zmm15
+
+/* exponent%3 (to be used as index) */
+        vmovups   Three+__svml_scbrt_data_internal_avx512(%rip), %zmm5
+
+/* polynomial */
+        vmovups   poly_coeff3+__svml_scbrt_data_internal_avx512(%rip), %zmm11
+        vmovups   poly_coeff1+__svml_scbrt_data_internal_avx512(%rip), %zmm14
+
+/* Table lookup */
+        vmovups   cbrt_tbl_H+__svml_scbrt_data_internal_avx512(%rip), %zmm12
+
+/* DblRcp ~ 1/Mantissa */
+        vrcp14ps  %zmm8, %zmm7
+        vaddps    {rn-sae}, %zmm2, %zmm1, %zmm6
+        vandps    SZero+__svml_scbrt_data_internal_avx512(%rip), %zmm0, %zmm0
+
+/* round DblRcp to 3 fractional bits (RN mode, no Precision exception) */
+        vrndscaleps $88, {sae}, %zmm7, %zmm9
+        vfmsub231ps {rn-sae}, %zmm6, %zmm3, %zmm4
+        vmovups   poly_coeff2+__svml_scbrt_data_internal_avx512(%rip), %zmm7
+
+/* Reduced argument: R = DblRcp*Mantissa - 1 */
+        vfmsub231ps {rn-sae}, %zmm9, %zmm8, %zmm15
+        vrndscaleps $9, {sae}, %zmm4, %zmm13
+
+/* Prepare table index */
+        vpsrld    $19, %zmm9, %zmm10
+        vfmadd231ps {rn-sae}, %zmm15, %zmm11, %zmm7
+        vfnmadd231ps {rn-sae}, %zmm13, %zmm5, %zmm6
+        vpermt2ps cbrt_tbl_H+64+__svml_scbrt_data_internal_avx512(%rip), %zmm10, %zmm12
+        vfmadd213ps {rn-sae}, %zmm14, %zmm15, %zmm7
+        vscalefps {rn-sae}, %zmm13, %zmm12, %zmm2
+
+/* Table lookup: 2^(exponent%3) */
+        vpermps   __svml_scbrt_data_internal_avx512(%rip), %zmm6, %zmm1
+        vpermps   etbl_L+__svml_scbrt_data_internal_avx512(%rip), %zmm6, %zmm6
+
+/* Sh*R */
+        vmulps    {rn-sae}, %zmm15, %zmm1, %zmm14
+
+/* Sl + (Sh*R)*Poly */
+        vfmadd213ps {rn-sae}, %zmm6, %zmm7, %zmm14
+
+/*
+ * branch-free
+ * scaled_Th*(Sh+Sl+Sh*R*Poly)
+ */
+        vaddps    {rn-sae}, %zmm1, %zmm14, %zmm15
+        vmulps    {rn-sae}, %zmm2, %zmm15, %zmm3
+        vorps     %zmm0, %zmm3, %zmm0
+        ret
+
+END(_ZGVeN16v_cbrtf_skx)
+
+        .section .rodata, "a"
+        .align 64
+
+#ifdef __svml_scbrt_data_internal_avx512_typedef
+typedef unsigned int VUINT32;
+typedef struct {
+        __declspec(align(64)) VUINT32 etbl_H[16][1];
+        __declspec(align(64)) VUINT32 etbl_L[16][1];
+        __declspec(align(64)) VUINT32 cbrt_tbl_H[32][1];
+        __declspec(align(64)) VUINT32 BiasL[16][1];
+        __declspec(align(64)) VUINT32 SZero[16][1];
+        __declspec(align(64)) VUINT32 OneThird[16][1];
+        __declspec(align(64)) VUINT32 Bias3[16][1];
+        __declspec(align(64)) VUINT32 Three[16][1];
+        __declspec(align(64)) VUINT32 One[16][1];
+        __declspec(align(64)) VUINT32 poly_coeff3[16][1];
+        __declspec(align(64)) VUINT32 poly_coeff2[16][1];
+        __declspec(align(64)) VUINT32 poly_coeff1[16][1];
+    } __svml_scbrt_data_internal_avx512;
+#endif
+__svml_scbrt_data_internal_avx512:
+        /*== etbl_H ==*/
+        .long 0x3f800000
+        .long 0x3fa14518
+        .long 0x3fcb2ff5
+        .long 0x00000000
+        .long 0x00000000
+        .long 0x00000000
+        .long 0x00000000
+        .long 0x00000000
+        .long 0x00000000
+        .long 0x00000000
+        .long 0x00000000
+        .long 0x00000000
+        .long 0x00000000
+        .long 0x00000000
+        .long 0x00000000
+        .long 0x00000000
+        /*== etbl_L ==*/
+        .align 64
+        .long 0x00000000
+        .long 0xb2ce51af
+        .long 0x32a7adc8
+        .long 0x00000000
+        .long 0x00000000
+        .long 0x00000000
+        .long 0x00000000
+        .long 0x00000000
+        .long 0x00000000
+        .long 0x00000000
+        .long 0x00000000
+        .long 0x00000000
+        .long 0x00000000
+        .long 0x00000000
+        .long 0x00000000
+        .long 0x00000000
+        /*== cbrt_tbl_H ==*/
+        .align 64
+        .long 0x3fa14518
+        .long 0x3f9e0b2b
+        .long 0x3f9b0f9b
+        .long 0x3f984a9a
+        .long 0x3f95b5af
+        .long 0x3f934b6c
+        .long 0x3f910737
+        .long 0x3f8ee526
+        .long 0x3f8ce1da
+        .long 0x3f8afa6a
+        .long 0x3f892c4e
+        .long 0x3f87754e
+        .long 0x3f85d377
+        .long 0x3f844510
+        .long 0x3f82c892
+        .long 0x3f815c9f
+        .long 0x3f800000
+        .long 0x00000000
+        .long 0x00000000
+        .long 0x00000000
+        .long 0x00000000
+        .long 0x00000000
+        .long 0x00000000
+        .long 0x00000000
+        .long 0x00000000
+        .long 0x00000000
+        .long 0x00000000
+        .long 0x00000000
+        .long 0x00000000
+        .long 0x00000000
+        .long 0x00000000
+        .long 0x00000000
+        /*== BiasL ==*/
+        .align 64
+        .long 0x4b400000, 0x4b400000, 0x4b400000, 0x4b400000, 0x4b400000, 0x4b400000, 0x4b400000, 0x4b400000, 0x4b400000, 0x4b400000, 0x4b400000, 0x4b400000, 0x4b400000, 0x4b400000, 0x4b400000, 0x4b400000
+        /*== Zero ==*/
+        .align 64
+        .long 0x80000000, 0x80000000, 0x80000000, 0x80000000, 0x80000000, 0x80000000, 0x80000000, 0x80000000, 0x80000000, 0x80000000, 0x80000000, 0x80000000, 0x80000000, 0x80000000, 0x80000000, 0x80000000
+        /*== OneThird ==*/
+        .align 64
+        .long 0x3eaaaaab, 0x3eaaaaab, 0x3eaaaaab, 0x3eaaaaab, 0x3eaaaaab, 0x3eaaaaab, 0x3eaaaaab, 0x3eaaaaab, 0x3eaaaaab, 0x3eaaaaab, 0x3eaaaaab, 0x3eaaaaab, 0x3eaaaaab, 0x3eaaaaab, 0x3eaaaaab, 0x3eaaaaab
+        /*== Bias3 ==*/
+        .align 64
+        .long 0x4a800000, 0x4a800000, 0x4a800000, 0x4a800000, 0x4a800000, 0x4a800000, 0x4a800000, 0x4a800000, 0x4a800000, 0x4a800000, 0x4a800000, 0x4a800000, 0x4a800000, 0x4a800000, 0x4a800000, 0x4a800000
+        /*== Three ==*/
+        .align 64
+        .long 0x40400000, 0x40400000, 0x40400000, 0x40400000, 0x40400000, 0x40400000, 0x40400000, 0x40400000, 0x40400000, 0x40400000, 0x40400000, 0x40400000, 0x40400000, 0x40400000, 0x40400000, 0x40400000
+        /*==One ==*/
+        .align 64
+        .long 0x3f800000, 0x3f800000, 0x3f800000, 0x3f800000, 0x3f800000, 0x3f800000, 0x3f800000, 0x3f800000, 0x3f800000, 0x3f800000, 0x3f800000, 0x3f800000, 0x3f800000, 0x3f800000, 0x3f800000, 0x3f800000
+        /*== poly_coeff3 ==*/
+        .align 64
+        .long 0x3d7d057c, 0x3d7d057c, 0x3d7d057c, 0x3d7d057c, 0x3d7d057c, 0x3d7d057c, 0x3d7d057c, 0x3d7d057c, 0x3d7d057c, 0x3d7d057c, 0x3d7d057c, 0x3d7d057c, 0x3d7d057c, 0x3d7d057c, 0x3d7d057c, 0x3d7d057c
+        /*== poly_coeff2 ==*/
+        .align 64
+        .long 0xbde3a363, 0xbde3a363, 0xbde3a363, 0xbde3a363, 0xbde3a363, 0xbde3a363, 0xbde3a363, 0xbde3a363, 0xbde3a363, 0xbde3a363, 0xbde3a363, 0xbde3a363, 0xbde3a363, 0xbde3a363, 0xbde3a363, 0xbde3a363
+        /*== poly_coeff1 ==*/
+        .align 64
+        .long 0x3eaaaaaa, 0x3eaaaaaa, 0x3eaaaaaa, 0x3eaaaaaa, 0x3eaaaaaa, 0x3eaaaaaa, 0x3eaaaaaa, 0x3eaaaaaa, 0x3eaaaaaa, 0x3eaaaaaa, 0x3eaaaaaa, 0x3eaaaaaa, 0x3eaaaaaa, 0x3eaaaaaa, 0x3eaaaaaa, 0x3eaaaaaa
+        .align 64
+        .type	__svml_scbrt_data_internal_avx512,@object
+        .size	__svml_scbrt_data_internal_avx512,.-__svml_scbrt_data_internal_avx512
diff --git a/sysdeps/x86_64/fpu/multiarch/svml_s_cbrtf4_core-sse2.S b/sysdeps/x86_64/fpu/multiarch/svml_s_cbrtf4_core-sse2.S
new file mode 100644
index 0000000000..76fc254e7a
--- /dev/null
+++ b/sysdeps/x86_64/fpu/multiarch/svml_s_cbrtf4_core-sse2.S
@@ -0,0 +1,20 @@
+/* SSE2 version of vectorized cbrtf, 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
+   <https://www.gnu.org/licenses/>.  */
+
+#define _ZGVbN4v_cbrtf _ZGVbN4v_cbrtf_sse2
+#include "../svml_s_cbrtf4_core.S"
diff --git a/sysdeps/x86_64/fpu/multiarch/svml_s_cbrtf4_core.c b/sysdeps/x86_64/fpu/multiarch/svml_s_cbrtf4_core.c
new file mode 100644
index 0000000000..564a549b39
--- /dev/null
+++ b/sysdeps/x86_64/fpu/multiarch/svml_s_cbrtf4_core.c
@@ -0,0 +1,28 @@
+/* Multiple versions of vectorized cbrtf, 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
+   <https://www.gnu.org/licenses/>.  */
+
+#define SYMBOL_NAME _ZGVbN4v_cbrtf
+#include "ifunc-mathvec-sse4_1.h"
+
+libc_ifunc_redirected (REDIRECT_NAME, SYMBOL_NAME, IFUNC_SELECTOR ());
+
+#ifdef SHARED
+__hidden_ver1 (_ZGVbN4v_cbrtf, __GI__ZGVbN4v_cbrtf,
+	       __redirect__ZGVbN4v_cbrtf)
+  __attribute__ ((visibility ("hidden")));
+#endif
diff --git a/sysdeps/x86_64/fpu/multiarch/svml_s_cbrtf4_core_sse4.S b/sysdeps/x86_64/fpu/multiarch/svml_s_cbrtf4_core_sse4.S
new file mode 100644
index 0000000000..af42dd5164
--- /dev/null
+++ b/sysdeps/x86_64/fpu/multiarch/svml_s_cbrtf4_core_sse4.S
@@ -0,0 +1,490 @@
+/* Function cbrtf 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:
+ *
+ *     x=2^{3*k+j} * 1.b1 b2 ... b5 b6 ... b52
+ *     Let r=(x*2^{-3k-j} - 1.b1 b2 ... b5 1)* rcp[b1 b2 ..b5],
+ *     where rcp[b1 b2 .. b5]=1/(1.b1 b2 b3 b4 b5 1) in single precision
+ *     cbrtf(2^j * 1. b1 b2 .. b5 1) is approximated as T[j][b1..b5]+D[j][b1..b5]
+ *     (T stores the high 24 bits, D stores the low order bits)
+ *     Result=2^k*T+(2^k*T*r)*P+2^k*D
+ *      where P=p1+p2*r+..
+ *
+ */
+
+/* Offsets for data table __svml_scbrt_data_internal
+ */
+#define _sRcp                         	0
+#define _sCbrtHL                      	128
+#define _sP2                          	512
+#define _sP1                          	528
+#define _sMantissaMask                	544
+#define _sMantissaMask1               	560
+#define _sExpMask                     	576
+#define _sExpMask1                    	592
+#define _iRcpIndexMask                	608
+#define _iBExpMask                    	624
+#define _iSignMask                    	640
+#define _iBias                        	656
+#define _iOne                         	672
+#define _i555                         	688
+#define _iAbsMask                     	704
+#define _iSubConst                    	720
+#define _iCmpConst                    	736
+
+#include <sysdep.h>
+
+        .text
+	.section .text.sse4,"ax",@progbits
+ENTRY(_ZGVbN4v_cbrtf_sse4)
+        subq      $72, %rsp
+        cfi_def_cfa_offset(80)
+
+/*
+ * Load constants
+ * Reciprocal index calculation
+ */
+        movaps    %xmm0, %xmm2
+        movdqu    _iRcpIndexMask+__svml_scbrt_data_internal(%rip), %xmm3
+        psrld     $16, %xmm2
+        pand      %xmm2, %xmm3
+
+/* Load reciprocal value */
+        lea       __svml_scbrt_data_internal(%rip), %rdx
+        pshufd    $1, %xmm3, %xmm5
+
+/* Get signed biased exponent */
+        psrld     $7, %xmm2
+        movd      %xmm3, %eax
+        movd      %xmm5, %ecx
+
+/* Get absolute biased exponent */
+        movdqu    _iBExpMask+__svml_scbrt_data_internal(%rip), %xmm15
+
+/*
+ * Calculate exponent/3
+ * i555Exp=(2^{12}-1)/3*exponent
+ */
+        movdqu    _i555+__svml_scbrt_data_internal(%rip), %xmm14
+        pand      %xmm2, %xmm15
+        movslq    %eax, %rax
+        movdqa    %xmm14, %xmm5
+        movslq    %ecx, %rcx
+        psrlq     $32, %xmm14
+        pmuludq   %xmm15, %xmm5
+        movd      (%rdx,%rax), %xmm4
+        movd      (%rdx,%rcx), %xmm6
+        punpckldq %xmm6, %xmm4
+        movdqa    %xmm15, %xmm6
+        psrlq     $32, %xmm15
+        pmuludq   %xmm14, %xmm15
+        pshufd    $2, %xmm3, %xmm7
+        psllq     $32, %xmm15
+        pshufd    $3, %xmm3, %xmm8
+        movd      %xmm7, %esi
+        movd      %xmm8, %edi
+
+/* Argument reduction */
+        movups    _sMantissaMask+__svml_scbrt_data_internal(%rip), %xmm12
+        movups    _sMantissaMask1+__svml_scbrt_data_internal(%rip), %xmm11
+        andps     %xmm0, %xmm12
+        pand      .FLT_17(%rip), %xmm5
+        andps     %xmm0, %xmm11
+        movslq    %esi, %rsi
+        por       %xmm15, %xmm5
+        movslq    %edi, %rdi
+
+/* Get K (exponent=3*k+j) */
+        psrld     $12, %xmm5
+        orps      _sExpMask+__svml_scbrt_data_internal(%rip), %xmm12
+        orps      _sExpMask1+__svml_scbrt_data_internal(%rip), %xmm11
+        psubd     _iOne+__svml_scbrt_data_internal(%rip), %xmm6
+
+/* r=y-y` */
+        subps     %xmm11, %xmm12
+
+/* Get J */
+        psubd     %xmm5, %xmm6
+        movdqu    _iAbsMask+__svml_scbrt_data_internal(%rip), %xmm1
+        psubd     %xmm5, %xmm6
+        movd      (%rdx,%rsi), %xmm10
+        pand      %xmm0, %xmm1
+        movd      (%rdx,%rdi), %xmm9
+        psubd     %xmm5, %xmm6
+        punpckldq %xmm9, %xmm10
+
+/* Get 128*J */
+        pslld     $7, %xmm6
+        punpcklqdq %xmm10, %xmm4
+
+/*
+ * iCbrtIndex=4*l+128*j
+ * Zero index if callout expected
+ */
+        paddd     %xmm6, %xmm3
+        psubd     _iSubConst+__svml_scbrt_data_internal(%rip), %xmm1
+        pcmpgtd   _iCmpConst+__svml_scbrt_data_internal(%rip), %xmm1
+
+/* r=(y-y`)*rcp_table(y`) */
+        mulps     %xmm12, %xmm4
+        movmskps  %xmm1, %eax
+
+/* Biased exponent-1 */
+        movdqu    _iSignMask+__svml_scbrt_data_internal(%rip), %xmm13
+        pandn     %xmm3, %xmm1
+
+/*
+ * Add 2/3*(bias-1)+1 to (k+1/3*(bias-1))
+ * Attach sign to exponent
+ */
+        movdqu    _iBias+__svml_scbrt_data_internal(%rip), %xmm12
+        pand      %xmm13, %xmm2
+        paddd     %xmm5, %xmm12
+
+/* Load Cbrt table Hi & Lo values */
+        movd      %xmm1, %r8d
+        por       %xmm2, %xmm12
+        pshufd    $1, %xmm1, %xmm2
+        pslld     $23, %xmm12
+        pshufd    $2, %xmm1, %xmm7
+        pshufd    $3, %xmm1, %xmm1
+        movd      %xmm2, %r9d
+        movd      %xmm7, %r10d
+        movd      %xmm1, %r11d
+
+/* Polynomial:    p1+r*(p2*r+r*(p3+r*p4)) */
+        movups    _sP2+__svml_scbrt_data_internal(%rip), %xmm11
+        mulps     %xmm4, %xmm11
+        movslq    %r8d, %r8
+        addps     _sP1+__svml_scbrt_data_internal(%rip), %xmm11
+        movslq    %r9d, %r9
+        movslq    %r10d, %r10
+        movslq    %r11d, %r11
+        movd      128(%rdx,%r8), %xmm10
+        movd      128(%rdx,%r9), %xmm3
+        movd      128(%rdx,%r10), %xmm9
+        movd      128(%rdx,%r11), %xmm8
+        punpckldq %xmm3, %xmm10
+        punpckldq %xmm8, %xmm9
+        punpcklqdq %xmm9, %xmm10
+
+/* sCbrtHi *= 2^k */
+        mulps     %xmm10, %xmm12
+
+/* T`*r */
+        mulps     %xmm12, %xmm4
+
+/* (T`*r)*P */
+        mulps     %xmm4, %xmm11
+
+/*
+ * T`*r*P+D`
+ * result = T`+(T`*r*P+D`)
+ */
+        addps     %xmm11, %xmm12
+        testl     %eax, %eax
+
+/* Go to special inputs processing branch */
+        jne       L(SPECIAL_VALUES_BRANCH)
+                                # LOE rbx rbp r12 r13 r14 r15 eax xmm0 xmm12
+
+/* Restore registers
+ * and exit the function
+ */
+
+L(EXIT):
+        movaps    %xmm12, %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    %xmm12, 48(%rsp)
+                                # LOE rbx rbp r12 r13 r14 r15 eax
+
+        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      $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), %xmm12
+
+/* 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 xmm12
+
+/* Scalar math fucntion call
+ * to process special input
+ */
+
+L(SCALAR_MATH_CALL):
+        movl      %r12d, %r14d
+        movss     32(%rsp,%r14,4), %xmm0
+        call      cbrtf@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_cbrtf_sse4)
+
+        .section .rodata, "a"
+        .align 16
+
+#ifdef __svml_scbrt_data_internal_typedef
+typedef unsigned int VUINT32;
+typedef struct
+{
+        __declspec(align(16)) VUINT32 _sRcp[32][1];
+        __declspec(align(16)) VUINT32 _sCbrtHL[96][1];
+        __declspec(align(16)) VUINT32 _sP2[4][1];
+        __declspec(align(16)) VUINT32 _sP1[4][1];
+        __declspec(align(16)) VUINT32 _sMantissaMask[4][1];
+        __declspec(align(16)) VUINT32 _sMantissaMask1[4][1];
+        __declspec(align(16)) VUINT32 _sExpMask[4][1];
+        __declspec(align(16)) VUINT32 _sExpMask1[4][1];
+        __declspec(align(16)) VUINT32 _iRcpIndexMask[4][1];
+        __declspec(align(16)) VUINT32 _iBExpMask[4][1];
+        __declspec(align(16)) VUINT32 _iSignMask[4][1];
+        __declspec(align(16)) VUINT32 _iBias[4][1];
+        __declspec(align(16)) VUINT32 _iOne[4][1];
+        __declspec(align(16)) VUINT32 _i555[4][1];
+        __declspec(align(16)) VUINT32 _iAbsMask[4][1];
+        __declspec(align(16)) VUINT32 _iSubConst[4][1];
+        __declspec(align(16)) VUINT32 _iCmpConst[4][1];
+} __svml_scbrt_data_internal;
+#endif
+__svml_scbrt_data_internal:
+        /*== _sRcp ==*/
+        .long 0xBF7C0FC1  /* (1/(1+0/32+1/64)) = -.984615 */
+        .long 0xBF74898D  /* (1/(1+1/32+1/64)) = -.955224 */
+        .long 0xBF6D7304  /* (1/(1+2/32+1/64)) = -.927536 */
+        .long 0xBF66C2B4  /* (1/(1+3/32+1/64)) = -.901408 */
+        .long 0xBF607038  /* (1/(1+4/32+1/64)) = -.876712 */
+        .long 0xBF5A740E  /* (1/(1+5/32+1/64)) = -.853333 */
+        .long 0xBF54C77B  /* (1/(1+6/32+1/64)) = -.831169 */
+        .long 0xBF4F6475  /* (1/(1+7/32+1/64)) = -.810127 */
+        .long 0xBF4A4588  /* (1/(1+8/32+1/64)) = -.790123 */
+        .long 0xBF4565C8  /* (1/(1+9/32+1/64)) = -.771084 */
+        .long 0xBF40C0C1  /* (1/(1+10/32+1/64)) = -.752941 */
+        .long 0xBF3C5264  /* (1/(1+11/32+1/64)) = -.735632 */
+        .long 0xBF381703  /* (1/(1+12/32+1/64)) = -.719101 */
+        .long 0xBF340B41  /* (1/(1+13/32+1/64)) = -.703297 */
+        .long 0xBF302C0B  /* (1/(1+14/32+1/64)) = -.688172 */
+        .long 0xBF2C7692  /* (1/(1+15/32+1/64)) = -.673684 */
+        .long 0xBF28E83F  /* (1/(1+16/32+1/64)) = -.659794 */
+        .long 0xBF257EB5  /* (1/(1+17/32+1/64)) = -.646465 */
+        .long 0xBF2237C3  /* (1/(1+18/32+1/64)) = -.633663 */
+        .long 0xBF1F1166  /* (1/(1+19/32+1/64)) = -.621359 */
+        .long 0xBF1C09C1  /* (1/(1+20/32+1/64)) = -.609524 */
+        .long 0xBF191F1A  /* (1/(1+21/32+1/64)) = -.598131 */
+        .long 0xBF164FDA  /* (1/(1+22/32+1/64)) = -.587156 */
+        .long 0xBF139A86  /* (1/(1+23/32+1/64)) = -.576577 */
+        .long 0xBF10FDBC  /* (1/(1+24/32+1/64)) = -.566372 */
+        .long 0xBF0E7835  /* (1/(1+25/32+1/64)) = -.556522 */
+        .long 0xBF0C08C1  /* (1/(1+26/32+1/64)) = -.547009 */
+        .long 0xBF09AE41  /* (1/(1+27/32+1/64)) = -.537815 */
+        .long 0xBF0767AB  /* (1/(1+28/32+1/64)) = -.528926 */
+        .long 0xBF053408  /* (1/(1+29/32+1/64)) = -.520325 */
+        .long 0xBF03126F  /* (1/(1+30/32+1/64)) = -.512    */
+        .long 0xBF010204  /* (1/(1+31/32+1/64)) = -.503937 */
+        /*== _sCbrtHL ==*/
+        .align 16
+        .long 0x3F80A9C9    /* HI((2^0*(1+0/32+1/64))^(1/3)) = 1.005181 */
+        .long 0x3F81F833    /* HI((2^0*(1+1/32+1/64))^(1/3)) = 1.015387 */
+        .long 0x3F834007    /* HI((2^0*(1+2/32+1/64))^(1/3)) = 1.025391 */
+        .long 0x3F848194    /* HI((2^0*(1+3/32+1/64))^(1/3)) = 1.035204 */
+        .long 0x3F85BD25    /* HI((2^0*(1+4/32+1/64))^(1/3)) = 1.044835 */
+        .long 0x3F86F300    /* HI((2^0*(1+5/32+1/64))^(1/3)) = 1.054291 */
+        .long 0x3F882365    /* HI((2^0*(1+6/32+1/64))^(1/3)) = 1.06358  */
+        .long 0x3F894E90    /* HI((2^0*(1+7/32+1/64))^(1/3)) = 1.07271  */
+        .long 0x3F8A74B9    /* HI((2^0*(1+8/32+1/64))^(1/3)) = 1.081687 */
+        .long 0x3F8B9615    /* HI((2^0*(1+9/32+1/64))^(1/3)) = 1.090518 */
+        .long 0x3F8CB2D4    /* HI((2^0*(1+10/32+1/64))^(1/3)) = 1.099207 */
+        .long 0x3F8DCB24    /* HI((2^0*(1+11/32+1/64))^(1/3)) = 1.107762 */
+        .long 0x3F8EDF31    /* HI((2^0*(1+12/32+1/64))^(1/3)) = 1.116186 */
+        .long 0x3F8FEF22    /* HI((2^0*(1+13/32+1/64))^(1/3)) = 1.124485 */
+        .long 0x3F90FB1F    /* HI((2^0*(1+14/32+1/64))^(1/3)) = 1.132664 */
+        .long 0x3F92034C    /* HI((2^0*(1+15/32+1/64))^(1/3)) = 1.140726 */
+        .long 0x3F9307CA    /* HI((2^0*(1+16/32+1/64))^(1/3)) = 1.148675 */
+        .long 0x3F9408B9    /* HI((2^0*(1+17/32+1/64))^(1/3)) = 1.156516 */
+        .long 0x3F950638    /* HI((2^0*(1+18/32+1/64))^(1/3)) = 1.164252 */
+        .long 0x3F960064    /* HI((2^0*(1+19/32+1/64))^(1/3)) = 1.171887 */
+        .long 0x3F96F759    /* HI((2^0*(1+20/32+1/64))^(1/3)) = 1.179423 */
+        .long 0x3F97EB2F    /* HI((2^0*(1+21/32+1/64))^(1/3)) = 1.186865 */
+        .long 0x3F98DC01    /* HI((2^0*(1+22/32+1/64))^(1/3)) = 1.194214 */
+        .long 0x3F99C9E5    /* HI((2^0*(1+23/32+1/64))^(1/3)) = 1.201474 */
+        .long 0x3F9AB4F2    /* HI((2^0*(1+24/32+1/64))^(1/3)) = 1.208647 */
+        .long 0x3F9B9D3D    /* HI((2^0*(1+25/32+1/64))^(1/3)) = 1.215736 */
+        .long 0x3F9C82DA    /* HI((2^0*(1+26/32+1/64))^(1/3)) = 1.222743 */
+        .long 0x3F9D65DD    /* HI((2^0*(1+27/32+1/64))^(1/3)) = 1.229671 */
+        .long 0x3F9E4659    /* HI((2^0*(1+28/32+1/64))^(1/3)) = 1.236522 */
+        .long 0x3F9F245F    /* HI((2^0*(1+29/32+1/64))^(1/3)) = 1.243297 */
+        .long 0x3FA00000    /* HI((2^0*(1+30/32+1/64))^(1/3)) = 1.25     */
+        .long 0x3FA0D94C    /* HI((2^0*(1+31/32+1/64))^(1/3)) = 1.256631 */
+        .long 0x3FA21B02    /* HI((2^1*(1+0/32+1/64))^(1/3)) = 1.266449 */
+        .long 0x3FA3C059    /* HI((2^1*(1+1/32+1/64))^(1/3)) = 1.279307 */
+        .long 0x3FA55D61    /* HI((2^1*(1+2/32+1/64))^(1/3)) = 1.291912 */
+        .long 0x3FA6F282    /* HI((2^1*(1+3/32+1/64))^(1/3)) = 1.304276 */
+        .long 0x3FA8801A    /* HI((2^1*(1+4/32+1/64))^(1/3)) = 1.316409 */
+        .long 0x3FAA067E    /* HI((2^1*(1+5/32+1/64))^(1/3)) = 1.328323 */
+        .long 0x3FAB8602    /* HI((2^1*(1+6/32+1/64))^(1/3)) = 1.340027 */
+        .long 0x3FACFEEF    /* HI((2^1*(1+7/32+1/64))^(1/3)) = 1.35153  */
+        .long 0x3FAE718E    /* HI((2^1*(1+8/32+1/64))^(1/3)) = 1.36284  */
+        .long 0x3FAFDE1F    /* HI((2^1*(1+9/32+1/64))^(1/3)) = 1.373966 */
+        .long 0x3FB144E1    /* HI((2^1*(1+10/32+1/64))^(1/3)) = 1.384915 */
+        .long 0x3FB2A60D    /* HI((2^1*(1+11/32+1/64))^(1/3)) = 1.395692 */
+        .long 0x3FB401DA    /* HI((2^1*(1+12/32+1/64))^(1/3)) = 1.406307 */
+        .long 0x3FB5587B    /* HI((2^1*(1+13/32+1/64))^(1/3)) = 1.416763 */
+        .long 0x3FB6AA20    /* HI((2^1*(1+14/32+1/64))^(1/3)) = 1.427067 */
+        .long 0x3FB7F6F7    /* HI((2^1*(1+15/32+1/64))^(1/3)) = 1.437224 */
+        .long 0x3FB93F29    /* HI((2^1*(1+16/32+1/64))^(1/3)) = 1.44724  */
+        .long 0x3FBA82E1    /* HI((2^1*(1+17/32+1/64))^(1/3)) = 1.457119 */
+        .long 0x3FBBC244    /* HI((2^1*(1+18/32+1/64))^(1/3)) = 1.466866 */
+        .long 0x3FBCFD77    /* HI((2^1*(1+19/32+1/64))^(1/3)) = 1.476485 */
+        .long 0x3FBE349B    /* HI((2^1*(1+20/32+1/64))^(1/3)) = 1.48598  */
+        .long 0x3FBF67D3    /* HI((2^1*(1+21/32+1/64))^(1/3)) = 1.495356 */
+        .long 0x3FC0973C    /* HI((2^1*(1+22/32+1/64))^(1/3)) = 1.504615 */
+        .long 0x3FC1C2F6    /* HI((2^1*(1+23/32+1/64))^(1/3)) = 1.513762 */
+        .long 0x3FC2EB1A    /* HI((2^1*(1+24/32+1/64))^(1/3)) = 1.5228   */
+        .long 0x3FC40FC6    /* HI((2^1*(1+25/32+1/64))^(1/3)) = 1.531731 */
+        .long 0x3FC53112    /* HI((2^1*(1+26/32+1/64))^(1/3)) = 1.54056  */
+        .long 0x3FC64F16    /* HI((2^1*(1+27/32+1/64))^(1/3)) = 1.549289 */
+        .long 0x3FC769EB    /* HI((2^1*(1+28/32+1/64))^(1/3)) = 1.55792  */
+        .long 0x3FC881A6    /* HI((2^1*(1+29/32+1/64))^(1/3)) = 1.566457 */
+        .long 0x3FC9965D    /* HI((2^1*(1+30/32+1/64))^(1/3)) = 1.574901 */
+        .long 0x3FCAA825    /* HI((2^1*(1+31/32+1/64))^(1/3)) = 1.583256 */
+        .long 0x3FCC3D79    /* HI((2^2*(1+0/32+1/64))^(1/3)) = 1.595626 */
+        .long 0x3FCE5054    /* HI((2^2*(1+1/32+1/64))^(1/3)) = 1.611826 */
+        .long 0x3FD058B8    /* HI((2^2*(1+2/32+1/64))^(1/3)) = 1.627707 */
+        .long 0x3FD25726    /* HI((2^2*(1+3/32+1/64))^(1/3)) = 1.643285 */
+        .long 0x3FD44C15    /* HI((2^2*(1+4/32+1/64))^(1/3)) = 1.658572 */
+        .long 0x3FD637F2    /* HI((2^2*(1+5/32+1/64))^(1/3)) = 1.673582 */
+        .long 0x3FD81B24    /* HI((2^2*(1+6/32+1/64))^(1/3)) = 1.688328 */
+        .long 0x3FD9F60B    /* HI((2^2*(1+7/32+1/64))^(1/3)) = 1.702821 */
+        .long 0x3FDBC8FE    /* HI((2^2*(1+8/32+1/64))^(1/3)) = 1.717071 */
+        .long 0x3FDD9452    /* HI((2^2*(1+9/32+1/64))^(1/3)) = 1.731089 */
+        .long 0x3FDF5853    /* HI((2^2*(1+10/32+1/64))^(1/3)) = 1.744883 */
+        .long 0x3FE1154B    /* HI((2^2*(1+11/32+1/64))^(1/3)) = 1.758462 */
+        .long 0x3FE2CB7F    /* HI((2^2*(1+12/32+1/64))^(1/3)) = 1.771835 */
+        .long 0x3FE47B2E    /* HI((2^2*(1+13/32+1/64))^(1/3)) = 1.785009 */
+        .long 0x3FE62496    /* HI((2^2*(1+14/32+1/64))^(1/3)) = 1.797992 */
+        .long 0x3FE7C7F0    /* HI((2^2*(1+15/32+1/64))^(1/3)) = 1.810789 */
+        .long 0x3FE96571    /* HI((2^2*(1+16/32+1/64))^(1/3)) = 1.823408 */
+        .long 0x3FEAFD4C    /* HI((2^2*(1+17/32+1/64))^(1/3)) = 1.835855 */
+        .long 0x3FEC8FB3    /* HI((2^2*(1+18/32+1/64))^(1/3)) = 1.848135 */
+        .long 0x3FEE1CD3    /* HI((2^2*(1+19/32+1/64))^(1/3)) = 1.860255 */
+        .long 0x3FEFA4D7    /* HI((2^2*(1+20/32+1/64))^(1/3)) = 1.872218 */
+        .long 0x3FF127E9    /* HI((2^2*(1+21/32+1/64))^(1/3)) = 1.88403  */
+        .long 0x3FF2A62F    /* HI((2^2*(1+22/32+1/64))^(1/3)) = 1.895697 */
+        .long 0x3FF41FD0    /* HI((2^2*(1+23/32+1/64))^(1/3)) = 1.907221 */
+        .long 0x3FF594EE    /* HI((2^2*(1+24/32+1/64))^(1/3)) = 1.918607 */
+        .long 0x3FF705AC    /* HI((2^2*(1+25/32+1/64))^(1/3)) = 1.929861 */
+        .long 0x3FF8722A    /* HI((2^2*(1+26/32+1/64))^(1/3)) = 1.940984 */
+        .long 0x3FF9DA86    /* HI((2^2*(1+27/32+1/64))^(1/3)) = 1.951981 */
+        .long 0x3FFB3EDE    /* HI((2^2*(1+28/32+1/64))^(1/3)) = 1.962856 */
+        .long 0x3FFC9F4E    /* HI((2^2*(1+29/32+1/64))^(1/3)) = 1.973612 */
+        .long 0x3FFDFBF2    /* HI((2^2*(1+30/32+1/64))^(1/3)) = 1.984251 */
+        .long 0x3FFF54E3    /* HI((2^2*(1+31/32+1/64))^(1/3)) = 1.994778 */
+        .align 16
+        .long 0xBDE3A962, 0xBDE3A962, 0xBDE3A962, 0xBDE3A962  /* _sP2 */
+        .align 16
+        .long 0x3EAAAC91, 0x3EAAAC91, 0x3EAAAC91, 0x3EAAAC91  /* _sP1 */
+        .align 16
+        .long 0x007fffff, 0x007fffff, 0x007fffff, 0x007fffff  /* _sMantissaMask (EXP_MSK3) */
+        .align 16
+        .long 0x007e0000, 0x007e0000, 0x007e0000, 0x007e0000  /* _sMantissaMask1 (SIG_MASK) */
+        .align 16
+        .long 0xBF800000, 0xBF800000, 0xBF800000, 0xBF800000  /* _sExpMask  (EXP_MASK) */
+        .align 16
+        .long 0xBF820000, 0xBF820000, 0xBF820000, 0xBF820000  /* _sExpMask1 (EXP_MASK2) */
+        .align 16
+        .long 0x0000007c, 0x0000007c, 0x0000007c, 0x0000007c  /* _iRcpIndexMask */
+        .align 16
+        .long 0x000000ff, 0x000000ff, 0x000000ff, 0x000000ff  /* _iBExpMask */
+        .align 16
+        .long 0x00000100, 0x00000100, 0x00000100, 0x00000100  /* _iSignMask */
+        .align 16
+        .long 0x00000055, 0x00000055, 0x00000055, 0x00000055  /* _iBias */
+        .align 16
+        .long 0x00000001, 0x00000001, 0x00000001, 0x00000001  /* _iOne */
+        .align 16
+        .long 0x00000555, 0x00000555, 0x00000555, 0x00000555  /* _i555 */
+        .align 16
+        .long 0x7fffffff, 0x7fffffff, 0x7fffffff, 0x7fffffff  /* _iAbsMask */
+        .align 16
+        .long 0x80800000, 0x80800000, 0x80800000, 0x80800000  /* _iSubConst */
+        .align 16
+        .long 0xFEFFFFFF, 0xFEFFFFFF, 0xFEFFFFFF, 0xFEFFFFFF  /* _iCmpConst */
+        .align 16
+        .type	__svml_scbrt_data_internal,@object
+        .size	__svml_scbrt_data_internal,.-__svml_scbrt_data_internal
+        .align 16
+
+.FLT_17:
+        .long	0xffffffff,0x00000000,0xffffffff,0x00000000
+        .type	.FLT_17,@object
+        .size	.FLT_17,16
diff --git a/sysdeps/x86_64/fpu/multiarch/svml_s_cbrtf8_core-sse.S b/sysdeps/x86_64/fpu/multiarch/svml_s_cbrtf8_core-sse.S
new file mode 100644
index 0000000000..8eaa457fa6
--- /dev/null
+++ b/sysdeps/x86_64/fpu/multiarch/svml_s_cbrtf8_core-sse.S
@@ -0,0 +1,20 @@
+/* SSE version of vectorized cbrtf, 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
+   <https://www.gnu.org/licenses/>.  */
+
+#define _ZGVdN8v_cbrtf _ZGVdN8v_cbrtf_sse_wrapper
+#include "../svml_s_cbrtf8_core.S"
diff --git a/sysdeps/x86_64/fpu/multiarch/svml_s_cbrtf8_core.c b/sysdeps/x86_64/fpu/multiarch/svml_s_cbrtf8_core.c
new file mode 100644
index 0000000000..089d28461f
--- /dev/null
+++ b/sysdeps/x86_64/fpu/multiarch/svml_s_cbrtf8_core.c
@@ -0,0 +1,28 @@
+/* Multiple versions of vectorized cbrtf, 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
+   <https://www.gnu.org/licenses/>.  */
+
+#define SYMBOL_NAME _ZGVdN8v_cbrtf
+#include "ifunc-mathvec-avx2.h"
+
+libc_ifunc_redirected (REDIRECT_NAME, SYMBOL_NAME, IFUNC_SELECTOR ());
+
+#ifdef SHARED
+__hidden_ver1 (_ZGVdN8v_cbrtf, __GI__ZGVdN8v_cbrtf,
+	       __redirect__ZGVdN8v_cbrtf)
+  __attribute__ ((visibility ("hidden")));
+#endif
diff --git a/sysdeps/x86_64/fpu/multiarch/svml_s_cbrtf8_core_avx2.S b/sysdeps/x86_64/fpu/multiarch/svml_s_cbrtf8_core_avx2.S
new file mode 100644
index 0000000000..acd20d9db8
--- /dev/null
+++ b/sysdeps/x86_64/fpu/multiarch/svml_s_cbrtf8_core_avx2.S
@@ -0,0 +1,509 @@
+/* Function cbrtf 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:
+ *
+ *     x=2^{3*k+j} * 1.b1 b2 ... b5 b6 ... b52
+ *     Let r=(x*2^{-3k-j} - 1.b1 b2 ... b5 1)* rcp[b1 b2 ..b5],
+ *     where rcp[b1 b2 .. b5]=1/(1.b1 b2 b3 b4 b5 1) in single precision
+ *     cbrtf(2^j * 1. b1 b2 .. b5 1) is approximated as T[j][b1..b5]+D[j][b1..b5]
+ *     (T stores the high 24 bits, D stores the low order bits)
+ *     Result=2^k*T+(2^k*T*r)*P+2^k*D
+ *      where P=p1+p2*r+..
+ *
+ */
+
+/* Offsets for data table __svml_scbrt_data_internal
+ */
+#define _sRcp                         	0
+#define _sCbrtHL                      	128
+#define _sP2                          	512
+#define _sP1                          	544
+#define _sMantissaMask                	576
+#define _sMantissaMask1               	608
+#define _sExpMask                     	640
+#define _sExpMask1                    	672
+#define _iRcpIndexMask                	704
+#define _iBExpMask                    	736
+#define _iSignMask                    	768
+#define _iBias                        	800
+#define _iOne                         	832
+#define _i555                         	864
+#define _iAbsMask                     	896
+#define _iSubConst                    	928
+#define _iCmpConst                    	960
+
+#include <sysdep.h>
+
+        .text
+	.section .text.avx2,"ax",@progbits
+ENTRY(_ZGVdN8v_cbrtf_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
+
+/* Load reciprocal value */
+        lea       __svml_scbrt_data_internal(%rip), %rdx
+        vmovaps   %ymm0, %ymm5
+
+/*
+ * Load constants
+ * Reciprocal index calculation
+ */
+        vpsrld    $16, %ymm5, %ymm3
+        vpand     _iRcpIndexMask+__svml_scbrt_data_internal(%rip), %ymm3, %ymm4
+        vextractf128 $1, %ymm4, %xmm15
+        vmovd     %xmm4, %eax
+        vmovd     %xmm15, %r8d
+        vpextrd   $1, %xmm15, %r9d
+        vpextrd   $2, %xmm15, %r10d
+        vpextrd   $3, %xmm15, %r11d
+        movslq    %r8d, %r8
+        movslq    %r9d, %r9
+        movslq    %r10d, %r10
+        movslq    %r11d, %r11
+        vpextrd   $1, %xmm4, %ecx
+        vpextrd   $2, %xmm4, %esi
+        vpextrd   $3, %xmm4, %edi
+        movslq    %eax, %rax
+        movslq    %ecx, %rcx
+        movslq    %esi, %rsi
+        movslq    %edi, %rdi
+        vmovd     (%rdx,%r8), %xmm13
+        vmovd     (%rdx,%r9), %xmm14
+        vmovd     (%rdx,%r10), %xmm1
+        vmovd     (%rdx,%r11), %xmm0
+        vpunpckldq %xmm14, %xmm13, %xmm2
+        vpunpckldq %xmm0, %xmm1, %xmm13
+
+/* Get signed biased exponent */
+        vpsrld    $7, %ymm3, %ymm0
+        vmovd     (%rdx,%rax), %xmm6
+        vmovd     (%rdx,%rcx), %xmm7
+        vmovd     (%rdx,%rsi), %xmm8
+        vmovd     (%rdx,%rdi), %xmm9
+        vpunpckldq %xmm7, %xmm6, %xmm10
+        vpunpckldq %xmm9, %xmm8, %xmm11
+        vpunpcklqdq %xmm11, %xmm10, %xmm12
+        vpunpcklqdq %xmm13, %xmm2, %xmm6
+        vandps    _iAbsMask+__svml_scbrt_data_internal(%rip), %ymm5, %ymm3
+
+/* Argument reduction */
+        vandps    _sMantissaMask+__svml_scbrt_data_internal(%rip), %ymm5, %ymm8
+        vandps    _sMantissaMask1+__svml_scbrt_data_internal(%rip), %ymm5, %ymm9
+        vpsubd    _iSubConst+__svml_scbrt_data_internal(%rip), %ymm3, %ymm7
+        vorps     _sExpMask+__svml_scbrt_data_internal(%rip), %ymm8, %ymm10
+        vorps     _sExpMask1+__svml_scbrt_data_internal(%rip), %ymm9, %ymm11
+
+/* r=y-y` */
+        vsubps    %ymm11, %ymm10, %ymm15
+
+/* Biased exponent-1 */
+        vpand     _iSignMask+__svml_scbrt_data_internal(%rip), %ymm0, %ymm8
+        vpcmpgtd  _iCmpConst+__svml_scbrt_data_internal(%rip), %ymm7, %ymm2
+        vmovmskps %ymm2, %eax
+        vinsertf128 $1, %xmm6, %ymm12, %ymm14
+
+/* Get absolute biased exponent */
+        vpand     _iBExpMask+__svml_scbrt_data_internal(%rip), %ymm0, %ymm6
+
+/* r=(y-y`)*rcp_table(y`) */
+        vmulps    %ymm15, %ymm14, %ymm1
+        vpsubd    _iOne+__svml_scbrt_data_internal(%rip), %ymm6, %ymm10
+
+/*
+ * Calculate exponent/3
+ * i555Exp=(2^{12}-1)/3*exponent
+ */
+        vpmulld   _i555+__svml_scbrt_data_internal(%rip), %ymm6, %ymm3
+
+/* Get K (exponent=3*k+j) */
+        vpsrld    $12, %ymm3, %ymm13
+
+/* Get J */
+        vpsubd    %ymm13, %ymm10, %ymm11
+
+/* Add 2/3*(bias-1)+1 to (k+1/3*(bias-1)) */
+        vpaddd    _iBias+__svml_scbrt_data_internal(%rip), %ymm13, %ymm7
+        vpsubd    %ymm13, %ymm11, %ymm12
+
+/* Attach sign to exponent */
+        vpor      %ymm8, %ymm7, %ymm9
+        vpsubd    %ymm13, %ymm12, %ymm14
+        vpslld    $23, %ymm9, %ymm0
+
+/* Get 128*J */
+        vpslld    $7, %ymm14, %ymm15
+
+/* iCbrtIndex=4*l+128*j */
+        vpaddd    %ymm15, %ymm4, %ymm4
+
+/* Zero index if callout expected */
+        vpandn    %ymm4, %ymm2, %ymm4
+
+/* Load Cbrt table Hi & Lo values */
+        vmovd     %xmm4, %ecx
+        vextractf128 $1, %ymm4, %xmm13
+        vpextrd   $1, %xmm4, %esi
+        movslq    %ecx, %rcx
+        movslq    %esi, %rsi
+        vmovd     %xmm13, %r9d
+        vmovd     128(%rdx,%rcx), %xmm2
+        vpextrd   $2, %xmm4, %edi
+        vpextrd   $3, %xmm4, %r8d
+        vmovd     128(%rdx,%rsi), %xmm3
+        vpextrd   $1, %xmm13, %r10d
+        vpextrd   $2, %xmm13, %ecx
+        vpextrd   $3, %xmm13, %esi
+        movslq    %edi, %rdi
+        movslq    %r8d, %r8
+        movslq    %r9d, %r9
+        movslq    %r10d, %r10
+        movslq    %ecx, %rcx
+        movslq    %esi, %rsi
+        vmovd     128(%rdx,%rdi), %xmm6
+        vmovd     128(%rdx,%r8), %xmm7
+        vmovd     128(%rdx,%r9), %xmm11
+        vmovd     128(%rdx,%r10), %xmm12
+        vmovd     128(%rdx,%rcx), %xmm14
+        vmovd     128(%rdx,%rsi), %xmm15
+        vpunpckldq %xmm3, %xmm2, %xmm8
+        vpunpckldq %xmm7, %xmm6, %xmm9
+        vpunpckldq %xmm12, %xmm11, %xmm4
+        vpunpckldq %xmm15, %xmm14, %xmm11
+        vpunpcklqdq %xmm9, %xmm8, %xmm10
+        vpunpcklqdq %xmm11, %xmm4, %xmm2
+        vinsertf128 $1, %xmm2, %ymm10, %ymm3
+
+/* sCbrtHi *= 2^k */
+        vmulps    %ymm3, %ymm0, %ymm2
+
+/* Polynomial:    p1+r*(p2*r+r*(p3+r*p4)) */
+        vmovups   _sP2+__svml_scbrt_data_internal(%rip), %ymm0
+        vfmadd213ps _sP1+__svml_scbrt_data_internal(%rip), %ymm1, %ymm0
+
+/* T`*r */
+        vmulps    %ymm2, %ymm1, %ymm1
+
+/* (T`*r)*P */
+        vmulps    %ymm1, %ymm0, %ymm0
+
+/*
+ * T`*r*P+D`
+ * result = T`+(T`*r*P+D`)
+ */
+        vaddps    %ymm0, %ymm2, %ymm0
+        testl     %eax, %eax
+
+/* Go to special inputs processing branch */
+        jne       L(SPECIAL_VALUES_BRANCH)
+                                # LOE rbx r12 r13 r14 r15 eax ymm0 ymm5
+
+/* 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   %ymm5, 32(%rsp)
+        vmovups   %ymm0, 64(%rsp)
+                                # LOE rbx r12 r13 r14 r15 eax ymm0
+
+        xorl      %edx, %edx
+                                # 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      %edx, %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      %eax, %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      cbrtf@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_cbrtf_avx2)
+
+        .section .rodata, "a"
+        .align 32
+
+#ifdef __svml_scbrt_data_internal_typedef
+typedef unsigned int VUINT32;
+typedef struct
+{
+        __declspec(align(32)) VUINT32 _sRcp[32][1];
+        __declspec(align(32)) VUINT32 _sCbrtHL[96][1];
+        __declspec(align(32)) VUINT32 _sP2[8][1];
+        __declspec(align(32)) VUINT32 _sP1[8][1];
+        __declspec(align(32)) VUINT32 _sMantissaMask[8][1];
+        __declspec(align(32)) VUINT32 _sMantissaMask1[8][1];
+        __declspec(align(32)) VUINT32 _sExpMask[8][1];
+        __declspec(align(32)) VUINT32 _sExpMask1[8][1];
+        __declspec(align(32)) VUINT32 _iRcpIndexMask[8][1];
+        __declspec(align(32)) VUINT32 _iBExpMask[8][1];
+        __declspec(align(32)) VUINT32 _iSignMask[8][1];
+        __declspec(align(32)) VUINT32 _iBias[8][1];
+        __declspec(align(32)) VUINT32 _iOne[8][1];
+        __declspec(align(32)) VUINT32 _i555[8][1];
+        __declspec(align(32)) VUINT32 _iAbsMask[8][1];
+        __declspec(align(32)) VUINT32 _iSubConst[8][1];
+        __declspec(align(32)) VUINT32 _iCmpConst[8][1];
+} __svml_scbrt_data_internal;
+#endif
+__svml_scbrt_data_internal:
+        /*== _sRcp ==*/
+        .long 0xBF7C0FC1  /* (1/(1+0/32+1/64)) = -.984615 */
+        .long 0xBF74898D  /* (1/(1+1/32+1/64)) = -.955224 */
+        .long 0xBF6D7304  /* (1/(1+2/32+1/64)) = -.927536 */
+        .long 0xBF66C2B4  /* (1/(1+3/32+1/64)) = -.901408 */
+        .long 0xBF607038  /* (1/(1+4/32+1/64)) = -.876712 */
+        .long 0xBF5A740E  /* (1/(1+5/32+1/64)) = -.853333 */
+        .long 0xBF54C77B  /* (1/(1+6/32+1/64)) = -.831169 */
+        .long 0xBF4F6475  /* (1/(1+7/32+1/64)) = -.810127 */
+        .long 0xBF4A4588  /* (1/(1+8/32+1/64)) = -.790123 */
+        .long 0xBF4565C8  /* (1/(1+9/32+1/64)) = -.771084 */
+        .long 0xBF40C0C1  /* (1/(1+10/32+1/64)) = -.752941 */
+        .long 0xBF3C5264  /* (1/(1+11/32+1/64)) = -.735632 */
+        .long 0xBF381703  /* (1/(1+12/32+1/64)) = -.719101 */
+        .long 0xBF340B41  /* (1/(1+13/32+1/64)) = -.703297 */
+        .long 0xBF302C0B  /* (1/(1+14/32+1/64)) = -.688172 */
+        .long 0xBF2C7692  /* (1/(1+15/32+1/64)) = -.673684 */
+        .long 0xBF28E83F  /* (1/(1+16/32+1/64)) = -.659794 */
+        .long 0xBF257EB5  /* (1/(1+17/32+1/64)) = -.646465 */
+        .long 0xBF2237C3  /* (1/(1+18/32+1/64)) = -.633663 */
+        .long 0xBF1F1166  /* (1/(1+19/32+1/64)) = -.621359 */
+        .long 0xBF1C09C1  /* (1/(1+20/32+1/64)) = -.609524 */
+        .long 0xBF191F1A  /* (1/(1+21/32+1/64)) = -.598131 */
+        .long 0xBF164FDA  /* (1/(1+22/32+1/64)) = -.587156 */
+        .long 0xBF139A86  /* (1/(1+23/32+1/64)) = -.576577 */
+        .long 0xBF10FDBC  /* (1/(1+24/32+1/64)) = -.566372 */
+        .long 0xBF0E7835  /* (1/(1+25/32+1/64)) = -.556522 */
+        .long 0xBF0C08C1  /* (1/(1+26/32+1/64)) = -.547009 */
+        .long 0xBF09AE41  /* (1/(1+27/32+1/64)) = -.537815 */
+        .long 0xBF0767AB  /* (1/(1+28/32+1/64)) = -.528926 */
+        .long 0xBF053408  /* (1/(1+29/32+1/64)) = -.520325 */
+        .long 0xBF03126F  /* (1/(1+30/32+1/64)) = -.512    */
+        .long 0xBF010204  /* (1/(1+31/32+1/64)) = -.503937 */
+        /*== _sCbrtHL ==*/
+        .align 32
+        .long 0x3F80A9C9    /* HI((2^0*(1+0/32+1/64))^(1/3)) = 1.005181 */
+        .long 0x3F81F833    /* HI((2^0*(1+1/32+1/64))^(1/3)) = 1.015387 */
+        .long 0x3F834007    /* HI((2^0*(1+2/32+1/64))^(1/3)) = 1.025391 */
+        .long 0x3F848194    /* HI((2^0*(1+3/32+1/64))^(1/3)) = 1.035204 */
+        .long 0x3F85BD25    /* HI((2^0*(1+4/32+1/64))^(1/3)) = 1.044835 */
+        .long 0x3F86F300    /* HI((2^0*(1+5/32+1/64))^(1/3)) = 1.054291 */
+        .long 0x3F882365    /* HI((2^0*(1+6/32+1/64))^(1/3)) = 1.06358  */
+        .long 0x3F894E90    /* HI((2^0*(1+7/32+1/64))^(1/3)) = 1.07271  */
+        .long 0x3F8A74B9    /* HI((2^0*(1+8/32+1/64))^(1/3)) = 1.081687 */
+        .long 0x3F8B9615    /* HI((2^0*(1+9/32+1/64))^(1/3)) = 1.090518 */
+        .long 0x3F8CB2D4    /* HI((2^0*(1+10/32+1/64))^(1/3)) = 1.099207 */
+        .long 0x3F8DCB24    /* HI((2^0*(1+11/32+1/64))^(1/3)) = 1.107762 */
+        .long 0x3F8EDF31    /* HI((2^0*(1+12/32+1/64))^(1/3)) = 1.116186 */
+        .long 0x3F8FEF22    /* HI((2^0*(1+13/32+1/64))^(1/3)) = 1.124485 */
+        .long 0x3F90FB1F    /* HI((2^0*(1+14/32+1/64))^(1/3)) = 1.132664 */
+        .long 0x3F92034C    /* HI((2^0*(1+15/32+1/64))^(1/3)) = 1.140726 */
+        .long 0x3F9307CA    /* HI((2^0*(1+16/32+1/64))^(1/3)) = 1.148675 */
+        .long 0x3F9408B9    /* HI((2^0*(1+17/32+1/64))^(1/3)) = 1.156516 */
+        .long 0x3F950638    /* HI((2^0*(1+18/32+1/64))^(1/3)) = 1.164252 */
+        .long 0x3F960064    /* HI((2^0*(1+19/32+1/64))^(1/3)) = 1.171887 */
+        .long 0x3F96F759    /* HI((2^0*(1+20/32+1/64))^(1/3)) = 1.179423 */
+        .long 0x3F97EB2F    /* HI((2^0*(1+21/32+1/64))^(1/3)) = 1.186865 */
+        .long 0x3F98DC01    /* HI((2^0*(1+22/32+1/64))^(1/3)) = 1.194214 */
+        .long 0x3F99C9E5    /* HI((2^0*(1+23/32+1/64))^(1/3)) = 1.201474 */
+        .long 0x3F9AB4F2    /* HI((2^0*(1+24/32+1/64))^(1/3)) = 1.208647 */
+        .long 0x3F9B9D3D    /* HI((2^0*(1+25/32+1/64))^(1/3)) = 1.215736 */
+        .long 0x3F9C82DA    /* HI((2^0*(1+26/32+1/64))^(1/3)) = 1.222743 */
+        .long 0x3F9D65DD    /* HI((2^0*(1+27/32+1/64))^(1/3)) = 1.229671 */
+        .long 0x3F9E4659    /* HI((2^0*(1+28/32+1/64))^(1/3)) = 1.236522 */
+        .long 0x3F9F245F    /* HI((2^0*(1+29/32+1/64))^(1/3)) = 1.243297 */
+        .long 0x3FA00000    /* HI((2^0*(1+30/32+1/64))^(1/3)) = 1.25     */
+        .long 0x3FA0D94C    /* HI((2^0*(1+31/32+1/64))^(1/3)) = 1.256631 */
+        .long 0x3FA21B02    /* HI((2^1*(1+0/32+1/64))^(1/3)) = 1.266449 */
+        .long 0x3FA3C059    /* HI((2^1*(1+1/32+1/64))^(1/3)) = 1.279307 */
+        .long 0x3FA55D61    /* HI((2^1*(1+2/32+1/64))^(1/3)) = 1.291912 */
+        .long 0x3FA6F282    /* HI((2^1*(1+3/32+1/64))^(1/3)) = 1.304276 */
+        .long 0x3FA8801A    /* HI((2^1*(1+4/32+1/64))^(1/3)) = 1.316409 */
+        .long 0x3FAA067E    /* HI((2^1*(1+5/32+1/64))^(1/3)) = 1.328323 */
+        .long 0x3FAB8602    /* HI((2^1*(1+6/32+1/64))^(1/3)) = 1.340027 */
+        .long 0x3FACFEEF    /* HI((2^1*(1+7/32+1/64))^(1/3)) = 1.35153  */
+        .long 0x3FAE718E    /* HI((2^1*(1+8/32+1/64))^(1/3)) = 1.36284  */
+        .long 0x3FAFDE1F    /* HI((2^1*(1+9/32+1/64))^(1/3)) = 1.373966 */
+        .long 0x3FB144E1    /* HI((2^1*(1+10/32+1/64))^(1/3)) = 1.384915 */
+        .long 0x3FB2A60D    /* HI((2^1*(1+11/32+1/64))^(1/3)) = 1.395692 */
+        .long 0x3FB401DA    /* HI((2^1*(1+12/32+1/64))^(1/3)) = 1.406307 */
+        .long 0x3FB5587B    /* HI((2^1*(1+13/32+1/64))^(1/3)) = 1.416763 */
+        .long 0x3FB6AA20    /* HI((2^1*(1+14/32+1/64))^(1/3)) = 1.427067 */
+        .long 0x3FB7F6F7    /* HI((2^1*(1+15/32+1/64))^(1/3)) = 1.437224 */
+        .long 0x3FB93F29    /* HI((2^1*(1+16/32+1/64))^(1/3)) = 1.44724  */
+        .long 0x3FBA82E1    /* HI((2^1*(1+17/32+1/64))^(1/3)) = 1.457119 */
+        .long 0x3FBBC244    /* HI((2^1*(1+18/32+1/64))^(1/3)) = 1.466866 */
+        .long 0x3FBCFD77    /* HI((2^1*(1+19/32+1/64))^(1/3)) = 1.476485 */
+        .long 0x3FBE349B    /* HI((2^1*(1+20/32+1/64))^(1/3)) = 1.48598  */
+        .long 0x3FBF67D3    /* HI((2^1*(1+21/32+1/64))^(1/3)) = 1.495356 */
+        .long 0x3FC0973C    /* HI((2^1*(1+22/32+1/64))^(1/3)) = 1.504615 */
+        .long 0x3FC1C2F6    /* HI((2^1*(1+23/32+1/64))^(1/3)) = 1.513762 */
+        .long 0x3FC2EB1A    /* HI((2^1*(1+24/32+1/64))^(1/3)) = 1.5228   */
+        .long 0x3FC40FC6    /* HI((2^1*(1+25/32+1/64))^(1/3)) = 1.531731 */
+        .long 0x3FC53112    /* HI((2^1*(1+26/32+1/64))^(1/3)) = 1.54056  */
+        .long 0x3FC64F16    /* HI((2^1*(1+27/32+1/64))^(1/3)) = 1.549289 */
+        .long 0x3FC769EB    /* HI((2^1*(1+28/32+1/64))^(1/3)) = 1.55792  */
+        .long 0x3FC881A6    /* HI((2^1*(1+29/32+1/64))^(1/3)) = 1.566457 */
+        .long 0x3FC9965D    /* HI((2^1*(1+30/32+1/64))^(1/3)) = 1.574901 */
+        .long 0x3FCAA825    /* HI((2^1*(1+31/32+1/64))^(1/3)) = 1.583256 */
+        .long 0x3FCC3D79    /* HI((2^2*(1+0/32+1/64))^(1/3)) = 1.595626 */
+        .long 0x3FCE5054    /* HI((2^2*(1+1/32+1/64))^(1/3)) = 1.611826 */
+        .long 0x3FD058B8    /* HI((2^2*(1+2/32+1/64))^(1/3)) = 1.627707 */
+        .long 0x3FD25726    /* HI((2^2*(1+3/32+1/64))^(1/3)) = 1.643285 */
+        .long 0x3FD44C15    /* HI((2^2*(1+4/32+1/64))^(1/3)) = 1.658572 */
+        .long 0x3FD637F2    /* HI((2^2*(1+5/32+1/64))^(1/3)) = 1.673582 */
+        .long 0x3FD81B24    /* HI((2^2*(1+6/32+1/64))^(1/3)) = 1.688328 */
+        .long 0x3FD9F60B    /* HI((2^2*(1+7/32+1/64))^(1/3)) = 1.702821 */
+        .long 0x3FDBC8FE    /* HI((2^2*(1+8/32+1/64))^(1/3)) = 1.717071 */
+        .long 0x3FDD9452    /* HI((2^2*(1+9/32+1/64))^(1/3)) = 1.731089 */
+        .long 0x3FDF5853    /* HI((2^2*(1+10/32+1/64))^(1/3)) = 1.744883 */
+        .long 0x3FE1154B    /* HI((2^2*(1+11/32+1/64))^(1/3)) = 1.758462 */
+        .long 0x3FE2CB7F    /* HI((2^2*(1+12/32+1/64))^(1/3)) = 1.771835 */
+        .long 0x3FE47B2E    /* HI((2^2*(1+13/32+1/64))^(1/3)) = 1.785009 */
+        .long 0x3FE62496    /* HI((2^2*(1+14/32+1/64))^(1/3)) = 1.797992 */
+        .long 0x3FE7C7F0    /* HI((2^2*(1+15/32+1/64))^(1/3)) = 1.810789 */
+        .long 0x3FE96571    /* HI((2^2*(1+16/32+1/64))^(1/3)) = 1.823408 */
+        .long 0x3FEAFD4C    /* HI((2^2*(1+17/32+1/64))^(1/3)) = 1.835855 */
+        .long 0x3FEC8FB3    /* HI((2^2*(1+18/32+1/64))^(1/3)) = 1.848135 */
+        .long 0x3FEE1CD3    /* HI((2^2*(1+19/32+1/64))^(1/3)) = 1.860255 */
+        .long 0x3FEFA4D7    /* HI((2^2*(1+20/32+1/64))^(1/3)) = 1.872218 */
+        .long 0x3FF127E9    /* HI((2^2*(1+21/32+1/64))^(1/3)) = 1.88403  */
+        .long 0x3FF2A62F    /* HI((2^2*(1+22/32+1/64))^(1/3)) = 1.895697 */
+        .long 0x3FF41FD0    /* HI((2^2*(1+23/32+1/64))^(1/3)) = 1.907221 */
+        .long 0x3FF594EE    /* HI((2^2*(1+24/32+1/64))^(1/3)) = 1.918607 */
+        .long 0x3FF705AC    /* HI((2^2*(1+25/32+1/64))^(1/3)) = 1.929861 */
+        .long 0x3FF8722A    /* HI((2^2*(1+26/32+1/64))^(1/3)) = 1.940984 */
+        .long 0x3FF9DA86    /* HI((2^2*(1+27/32+1/64))^(1/3)) = 1.951981 */
+        .long 0x3FFB3EDE    /* HI((2^2*(1+28/32+1/64))^(1/3)) = 1.962856 */
+        .long 0x3FFC9F4E    /* HI((2^2*(1+29/32+1/64))^(1/3)) = 1.973612 */
+        .long 0x3FFDFBF2    /* HI((2^2*(1+30/32+1/64))^(1/3)) = 1.984251 */
+        .long 0x3FFF54E3    /* HI((2^2*(1+31/32+1/64))^(1/3)) = 1.994778 */
+        .align 32
+        .long 0xBDE3A962, 0xBDE3A962, 0xBDE3A962, 0xBDE3A962, 0xBDE3A962, 0xBDE3A962, 0xBDE3A962, 0xBDE3A962  /* _sP2 */
+        .align 32
+        .long 0x3EAAAC91, 0x3EAAAC91, 0x3EAAAC91, 0x3EAAAC91, 0x3EAAAC91, 0x3EAAAC91, 0x3EAAAC91, 0x3EAAAC91  /* _sP1 */
+        .align 32
+        .long 0x007fffff, 0x007fffff, 0x007fffff, 0x007fffff, 0x007fffff, 0x007fffff, 0x007fffff, 0x007fffff  /* _sMantissaMask (EXP_MSK3) */
+        .align 32
+        .long 0x007e0000, 0x007e0000, 0x007e0000, 0x007e0000, 0x007e0000, 0x007e0000, 0x007e0000, 0x007e0000  /* _sMantissaMask1 (SIG_MASK) */
+        .align 32
+        .long 0xBF800000, 0xBF800000, 0xBF800000, 0xBF800000, 0xBF800000, 0xBF800000, 0xBF800000, 0xBF800000  /* _sExpMask  (EXP_MASK) */
+        .align 32
+        .long 0xBF820000, 0xBF820000, 0xBF820000, 0xBF820000, 0xBF820000, 0xBF820000, 0xBF820000, 0xBF820000  /* _sExpMask1 (EXP_MASK2) */
+        .align 32
+        .long 0x0000007c, 0x0000007c, 0x0000007c, 0x0000007c, 0x0000007c, 0x0000007c, 0x0000007c, 0x0000007c  /* _iRcpIndexMask */
+        .align 32
+        .long 0x000000ff, 0x000000ff, 0x000000ff, 0x000000ff, 0x000000ff, 0x000000ff, 0x000000ff, 0x000000ff  /* _iBExpMask */
+        .align 32
+        .long 0x00000100, 0x00000100, 0x00000100, 0x00000100, 0x00000100, 0x00000100, 0x00000100, 0x00000100  /* _iSignMask */
+        .align 32
+        .long 0x00000055, 0x00000055, 0x00000055, 0x00000055, 0x00000055, 0x00000055, 0x00000055, 0x00000055  /* _iBias */
+        .align 32
+        .long 0x00000001, 0x00000001, 0x00000001, 0x00000001, 0x00000001, 0x00000001, 0x00000001, 0x00000001  /* _iOne */
+        .align 32
+        .long 0x00000555, 0x00000555, 0x00000555, 0x00000555, 0x00000555, 0x00000555, 0x00000555, 0x00000555  /* _i555 */
+        .align 32
+        .long 0x7fffffff, 0x7fffffff, 0x7fffffff, 0x7fffffff, 0x7fffffff, 0x7fffffff, 0x7fffffff, 0x7fffffff  /* _iAbsMask */
+        .align 32
+        .long 0x80800000, 0x80800000, 0x80800000, 0x80800000, 0x80800000, 0x80800000, 0x80800000, 0x80800000  /* _iSubConst */
+        .align 32
+        .long 0xFEFFFFFF, 0xFEFFFFFF, 0xFEFFFFFF, 0xFEFFFFFF, 0xFEFFFFFF, 0xFEFFFFFF, 0xFEFFFFFF, 0xFEFFFFFF  /* _iCmpConst */
+        .align 32
+        .type	__svml_scbrt_data_internal,@object
+        .size	__svml_scbrt_data_internal,.-__svml_scbrt_data_internal
diff --git a/sysdeps/x86_64/fpu/svml_d_cbrt2_core.S b/sysdeps/x86_64/fpu/svml_d_cbrt2_core.S
new file mode 100644
index 0000000000..4bf546564b
--- /dev/null
+++ b/sysdeps/x86_64/fpu/svml_d_cbrt2_core.S
@@ -0,0 +1,29 @@
+/* Function cbrt 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
+   <https://www.gnu.org/licenses/>.  */
+
+#include <sysdep.h>
+#include "svml_d_wrapper_impl.h"
+
+	.text
+ENTRY (_ZGVbN2v_cbrt)
+WRAPPER_IMPL_SSE2 cbrt
+END (_ZGVbN2v_cbrt)
+
+#ifndef USE_MULTIARCH
+ libmvec_hidden_def (_ZGVbN2v_cbrt)
+#endif
diff --git a/sysdeps/x86_64/fpu/svml_d_cbrt4_core.S b/sysdeps/x86_64/fpu/svml_d_cbrt4_core.S
new file mode 100644
index 0000000000..e6d1003e27
--- /dev/null
+++ b/sysdeps/x86_64/fpu/svml_d_cbrt4_core.S
@@ -0,0 +1,29 @@
+/* Function cbrt 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
+   <https://www.gnu.org/licenses/>.  */
+
+#include <sysdep.h>
+#include "svml_d_wrapper_impl.h"
+
+	.text
+ENTRY (_ZGVdN4v_cbrt)
+WRAPPER_IMPL_AVX _ZGVbN2v_cbrt
+END (_ZGVdN4v_cbrt)
+
+#ifndef USE_MULTIARCH
+ libmvec_hidden_def (_ZGVdN4v_cbrt)
+#endif
diff --git a/sysdeps/x86_64/fpu/svml_d_cbrt4_core_avx.S b/sysdeps/x86_64/fpu/svml_d_cbrt4_core_avx.S
new file mode 100644
index 0000000000..70632869ac
--- /dev/null
+++ b/sysdeps/x86_64/fpu/svml_d_cbrt4_core_avx.S
@@ -0,0 +1,25 @@
+/* Function cbrt 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
+   <https://www.gnu.org/licenses/>.  */
+
+#include <sysdep.h>
+#include "svml_d_wrapper_impl.h"
+
+	.text
+ENTRY (_ZGVcN4v_cbrt)
+WRAPPER_IMPL_AVX _ZGVbN2v_cbrt
+END (_ZGVcN4v_cbrt)
diff --git a/sysdeps/x86_64/fpu/svml_d_cbrt8_core.S b/sysdeps/x86_64/fpu/svml_d_cbrt8_core.S
new file mode 100644
index 0000000000..37571673a7
--- /dev/null
+++ b/sysdeps/x86_64/fpu/svml_d_cbrt8_core.S
@@ -0,0 +1,25 @@
+/* Function cbrt 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
+   <https://www.gnu.org/licenses/>.  */
+
+#include <sysdep.h>
+#include "svml_d_wrapper_impl.h"
+
+	.text
+ENTRY (_ZGVeN8v_cbrt)
+WRAPPER_IMPL_AVX512 _ZGVdN4v_cbrt
+END (_ZGVeN8v_cbrt)
diff --git a/sysdeps/x86_64/fpu/svml_s_cbrtf16_core.S b/sysdeps/x86_64/fpu/svml_s_cbrtf16_core.S
new file mode 100644
index 0000000000..1be6294026
--- /dev/null
+++ b/sysdeps/x86_64/fpu/svml_s_cbrtf16_core.S
@@ -0,0 +1,25 @@
+/* Function cbrtf 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
+   <https://www.gnu.org/licenses/>.  */
+
+#include <sysdep.h>
+#include "svml_s_wrapper_impl.h"
+
+	.text
+ENTRY (_ZGVeN16v_cbrtf)
+WRAPPER_IMPL_AVX512 _ZGVdN8v_cbrtf
+END (_ZGVeN16v_cbrtf)
diff --git a/sysdeps/x86_64/fpu/svml_s_cbrtf4_core.S b/sysdeps/x86_64/fpu/svml_s_cbrtf4_core.S
new file mode 100644
index 0000000000..2469a100f4
--- /dev/null
+++ b/sysdeps/x86_64/fpu/svml_s_cbrtf4_core.S
@@ -0,0 +1,29 @@
+/* Function cbrtf 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
+   <https://www.gnu.org/licenses/>.  */
+
+#include <sysdep.h>
+#include "svml_s_wrapper_impl.h"
+
+	.text
+ENTRY (_ZGVbN4v_cbrtf)
+WRAPPER_IMPL_SSE2 cbrtf
+END (_ZGVbN4v_cbrtf)
+
+#ifndef USE_MULTIARCH
+ libmvec_hidden_def (_ZGVbN4v_cbrtf)
+#endif
diff --git a/sysdeps/x86_64/fpu/svml_s_cbrtf8_core.S b/sysdeps/x86_64/fpu/svml_s_cbrtf8_core.S
new file mode 100644
index 0000000000..efedc22323
--- /dev/null
+++ b/sysdeps/x86_64/fpu/svml_s_cbrtf8_core.S
@@ -0,0 +1,29 @@
+/* Function cbrtf 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
+   <https://www.gnu.org/licenses/>.  */
+
+#include <sysdep.h>
+#include "svml_s_wrapper_impl.h"
+
+	.text
+ENTRY (_ZGVdN8v_cbrtf)
+WRAPPER_IMPL_AVX _ZGVbN4v_cbrtf
+END (_ZGVdN8v_cbrtf)
+
+#ifndef USE_MULTIARCH
+ libmvec_hidden_def (_ZGVdN8v_cbrtf)
+#endif
diff --git a/sysdeps/x86_64/fpu/svml_s_cbrtf8_core_avx.S b/sysdeps/x86_64/fpu/svml_s_cbrtf8_core_avx.S
new file mode 100644
index 0000000000..b5acc62426
--- /dev/null
+++ b/sysdeps/x86_64/fpu/svml_s_cbrtf8_core_avx.S
@@ -0,0 +1,25 @@
+/* Function cbrtf 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
+   <https://www.gnu.org/licenses/>.  */
+
+#include <sysdep.h>
+#include "svml_s_wrapper_impl.h"
+
+        .text
+ENTRY (_ZGVcN8v_cbrtf)
+WRAPPER_IMPL_AVX _ZGVbN4v_cbrtf
+END (_ZGVcN8v_cbrtf)
diff --git a/sysdeps/x86_64/fpu/test-double-libmvec-cbrt-avx.c b/sysdeps/x86_64/fpu/test-double-libmvec-cbrt-avx.c
new file mode 100644
index 0000000000..c8bc643c99
--- /dev/null
+++ b/sysdeps/x86_64/fpu/test-double-libmvec-cbrt-avx.c
@@ -0,0 +1 @@
+#include "test-double-libmvec-cbrt.c"
diff --git a/sysdeps/x86_64/fpu/test-double-libmvec-cbrt-avx2.c b/sysdeps/x86_64/fpu/test-double-libmvec-cbrt-avx2.c
new file mode 100644
index 0000000000..c8bc643c99
--- /dev/null
+++ b/sysdeps/x86_64/fpu/test-double-libmvec-cbrt-avx2.c
@@ -0,0 +1 @@
+#include "test-double-libmvec-cbrt.c"
diff --git a/sysdeps/x86_64/fpu/test-double-libmvec-cbrt-avx512f.c b/sysdeps/x86_64/fpu/test-double-libmvec-cbrt-avx512f.c
new file mode 100644
index 0000000000..c8bc643c99
--- /dev/null
+++ b/sysdeps/x86_64/fpu/test-double-libmvec-cbrt-avx512f.c
@@ -0,0 +1 @@
+#include "test-double-libmvec-cbrt.c"
diff --git a/sysdeps/x86_64/fpu/test-double-libmvec-cbrt.c b/sysdeps/x86_64/fpu/test-double-libmvec-cbrt.c
new file mode 100644
index 0000000000..fb3684b18c
--- /dev/null
+++ b/sysdeps/x86_64/fpu/test-double-libmvec-cbrt.c
@@ -0,0 +1,3 @@
+#define LIBMVEC_TYPE double
+#define LIBMVEC_FUNC cbrt
+#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 db136cc901..b1981ac7e4 100644
--- a/sysdeps/x86_64/fpu/test-double-vlen2-wrappers.c
+++ b/sysdeps/x86_64/fpu/test-double-vlen2-wrappers.c
@@ -36,6 +36,7 @@ VECTOR_WRAPPER (WRAPPER_NAME (exp10), _ZGVbN2v_exp10)
 VECTOR_WRAPPER (WRAPPER_NAME (cosh), _ZGVbN2v_cosh)
 VECTOR_WRAPPER (WRAPPER_NAME (expm1), _ZGVbN2v_expm1)
 VECTOR_WRAPPER (WRAPPER_NAME (sinh), _ZGVbN2v_sinh)
+VECTOR_WRAPPER (WRAPPER_NAME (cbrt), _ZGVbN2v_cbrt)
 
 #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 5fc09ac8c0..47915a7e59 100644
--- a/sysdeps/x86_64/fpu/test-double-vlen4-avx2-wrappers.c
+++ b/sysdeps/x86_64/fpu/test-double-vlen4-avx2-wrappers.c
@@ -39,6 +39,7 @@ VECTOR_WRAPPER (WRAPPER_NAME (exp10), _ZGVdN4v_exp10)
 VECTOR_WRAPPER (WRAPPER_NAME (cosh), _ZGVdN4v_cosh)
 VECTOR_WRAPPER (WRAPPER_NAME (expm1), _ZGVdN4v_expm1)
 VECTOR_WRAPPER (WRAPPER_NAME (sinh), _ZGVdN4v_sinh)
+VECTOR_WRAPPER (WRAPPER_NAME (cbrt), _ZGVdN4v_cbrt)
 
 #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 26ef7fb365..5cd5049807 100644
--- a/sysdeps/x86_64/fpu/test-double-vlen4-wrappers.c
+++ b/sysdeps/x86_64/fpu/test-double-vlen4-wrappers.c
@@ -36,6 +36,7 @@ VECTOR_WRAPPER (WRAPPER_NAME (exp10), _ZGVcN4v_exp10)
 VECTOR_WRAPPER (WRAPPER_NAME (cosh), _ZGVcN4v_cosh)
 VECTOR_WRAPPER (WRAPPER_NAME (expm1), _ZGVcN4v_expm1)
 VECTOR_WRAPPER (WRAPPER_NAME (sinh), _ZGVcN4v_sinh)
+VECTOR_WRAPPER (WRAPPER_NAME (cbrt), _ZGVcN4v_cbrt)
 
 #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 c7055fca76..83970739ab 100644
--- a/sysdeps/x86_64/fpu/test-double-vlen8-wrappers.c
+++ b/sysdeps/x86_64/fpu/test-double-vlen8-wrappers.c
@@ -36,6 +36,7 @@ VECTOR_WRAPPER (WRAPPER_NAME (exp10), _ZGVeN8v_exp10)
 VECTOR_WRAPPER (WRAPPER_NAME (cosh), _ZGVeN8v_cosh)
 VECTOR_WRAPPER (WRAPPER_NAME (expm1), _ZGVeN8v_expm1)
 VECTOR_WRAPPER (WRAPPER_NAME (sinh), _ZGVeN8v_sinh)
+VECTOR_WRAPPER (WRAPPER_NAME (cbrt), _ZGVeN8v_cbrt)
 
 #ifndef __ILP32__
 # define VEC_INT_TYPE __m512i
diff --git a/sysdeps/x86_64/fpu/test-float-libmvec-cbrtf-avx.c b/sysdeps/x86_64/fpu/test-float-libmvec-cbrtf-avx.c
new file mode 100644
index 0000000000..59b8d77f71
--- /dev/null
+++ b/sysdeps/x86_64/fpu/test-float-libmvec-cbrtf-avx.c
@@ -0,0 +1 @@
+#include "test-float-libmvec-cbrtf.c"
diff --git a/sysdeps/x86_64/fpu/test-float-libmvec-cbrtf-avx2.c b/sysdeps/x86_64/fpu/test-float-libmvec-cbrtf-avx2.c
new file mode 100644
index 0000000000..59b8d77f71
--- /dev/null
+++ b/sysdeps/x86_64/fpu/test-float-libmvec-cbrtf-avx2.c
@@ -0,0 +1 @@
+#include "test-float-libmvec-cbrtf.c"
diff --git a/sysdeps/x86_64/fpu/test-float-libmvec-cbrtf-avx512f.c b/sysdeps/x86_64/fpu/test-float-libmvec-cbrtf-avx512f.c
new file mode 100644
index 0000000000..59b8d77f71
--- /dev/null
+++ b/sysdeps/x86_64/fpu/test-float-libmvec-cbrtf-avx512f.c
@@ -0,0 +1 @@
+#include "test-float-libmvec-cbrtf.c"
diff --git a/sysdeps/x86_64/fpu/test-float-libmvec-cbrtf.c b/sysdeps/x86_64/fpu/test-float-libmvec-cbrtf.c
new file mode 100644
index 0000000000..3a06ba79e0
--- /dev/null
+++ b/sysdeps/x86_64/fpu/test-float-libmvec-cbrtf.c
@@ -0,0 +1,3 @@
+#define LIBMVEC_TYPE float
+#define LIBMVEC_FUNC cbrtf
+#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 d353bcb0f2..0420f11c28 100644
--- a/sysdeps/x86_64/fpu/test-float-vlen16-wrappers.c
+++ b/sysdeps/x86_64/fpu/test-float-vlen16-wrappers.c
@@ -36,6 +36,7 @@ VECTOR_WRAPPER (WRAPPER_NAME (exp10f), _ZGVeN16v_exp10f)
 VECTOR_WRAPPER (WRAPPER_NAME (coshf), _ZGVeN16v_coshf)
 VECTOR_WRAPPER (WRAPPER_NAME (expm1f), _ZGVeN16v_expm1f)
 VECTOR_WRAPPER (WRAPPER_NAME (sinhf), _ZGVeN16v_sinhf)
+VECTOR_WRAPPER (WRAPPER_NAME (cbrtf), _ZGVeN16v_cbrtf)
 
 #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 5e59117626..c8f7580265 100644
--- a/sysdeps/x86_64/fpu/test-float-vlen4-wrappers.c
+++ b/sysdeps/x86_64/fpu/test-float-vlen4-wrappers.c
@@ -36,6 +36,7 @@ VECTOR_WRAPPER (WRAPPER_NAME (exp10f), _ZGVbN4v_exp10f)
 VECTOR_WRAPPER (WRAPPER_NAME (coshf), _ZGVbN4v_coshf)
 VECTOR_WRAPPER (WRAPPER_NAME (expm1f), _ZGVbN4v_expm1f)
 VECTOR_WRAPPER (WRAPPER_NAME (sinhf), _ZGVbN4v_sinhf)
+VECTOR_WRAPPER (WRAPPER_NAME (cbrtf), _ZGVbN4v_cbrtf)
 
 #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 e884a5f4df..b581796b88 100644
--- a/sysdeps/x86_64/fpu/test-float-vlen8-avx2-wrappers.c
+++ b/sysdeps/x86_64/fpu/test-float-vlen8-avx2-wrappers.c
@@ -39,6 +39,7 @@ VECTOR_WRAPPER (WRAPPER_NAME (exp10f), _ZGVdN8v_exp10f)
 VECTOR_WRAPPER (WRAPPER_NAME (coshf), _ZGVdN8v_coshf)
 VECTOR_WRAPPER (WRAPPER_NAME (expm1f), _ZGVdN8v_expm1f)
 VECTOR_WRAPPER (WRAPPER_NAME (sinhf), _ZGVdN8v_sinhf)
+VECTOR_WRAPPER (WRAPPER_NAME (cbrtf), _ZGVdN8v_cbrtf)
 
 /* 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 95910d39e9..f16789e5ff 100644
--- a/sysdeps/x86_64/fpu/test-float-vlen8-wrappers.c
+++ b/sysdeps/x86_64/fpu/test-float-vlen8-wrappers.c
@@ -36,6 +36,7 @@ VECTOR_WRAPPER (WRAPPER_NAME (exp10f), _ZGVcN8v_exp10f)
 VECTOR_WRAPPER (WRAPPER_NAME (coshf), _ZGVcN8v_coshf)
 VECTOR_WRAPPER (WRAPPER_NAME (expm1f), _ZGVcN8v_expm1f)
 VECTOR_WRAPPER (WRAPPER_NAME (sinhf), _ZGVcN8v_sinhf)
+VECTOR_WRAPPER (WRAPPER_NAME (cbrtf), _ZGVcN8v_cbrtf)
 
 #define VEC_INT_TYPE __m128i