diff options
| author | Sunil K Pandey <skpgkp2@gmail.com> | 2021-12-29 08:35:22 -0800 |
|---|---|---|
| committer | Sunil K Pandey <skpgkp2@gmail.com> | 2021-12-29 11:37:21 -0800 |
| commit | 37475ba88303929e85704693455c7294e50aba77 (patch) | |
| tree | 4defc4412f556258f878588351b27b3b44406fb8 /sysdeps/x86_64/fpu | |
| parent | 11c01de14c879ffc8dbac8ce32242a7552cbd4ad (diff) | |
| download | glibc-37475ba88303929e85704693455c7294e50aba77.zip glibc-37475ba88303929e85704693455c7294e50aba77.tar.gz glibc-37475ba88303929e85704693455c7294e50aba77.tar.bz2 | |
x86-64: Add vector hypot/hypotf implementation to libmvec
Implement vectorized hypot/hypotf containing SSE, AVX, AVX2 and
AVX512 versions for libmvec as per vector ABI. It also contains
accuracy and ABI tests for vector hypot/hypotf with regenerated ulps.
Reviewed-by: H.J. Lu <hjl.tools@gmail.com>
Diffstat (limited to 'sysdeps/x86_64/fpu')
45 files changed, 2123 insertions, 0 deletions
diff --git a/sysdeps/x86_64/fpu/Makeconfig b/sysdeps/x86_64/fpu/Makeconfig index e0eae0b..7769a02 100644 --- a/sysdeps/x86_64/fpu/Makeconfig +++ b/sysdeps/x86_64/fpu/Makeconfig @@ -27,6 +27,7 @@ libmvec-funcs = \ atan \ cos \ exp \ + hypot \ log \ pow \ sin \ diff --git a/sysdeps/x86_64/fpu/Versions b/sysdeps/x86_64/fpu/Versions index 10baf86..e359e5d 100644 --- a/sysdeps/x86_64/fpu/Versions +++ b/sysdeps/x86_64/fpu/Versions @@ -17,8 +17,10 @@ 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; + _ZGVbN2vv_hypot; _ZGVcN4vv_hypot; _ZGVdN4vv_hypot; _ZGVeN8vv_hypot; _ZGVbN4v_acosf; _ZGVcN8v_acosf; _ZGVdN8v_acosf; _ZGVeN16v_acosf; _ZGVbN4v_asinf; _ZGVcN8v_asinf; _ZGVdN8v_asinf; _ZGVeN16v_asinf; _ZGVbN4v_atanf; _ZGVcN8v_atanf; _ZGVdN8v_atanf; _ZGVeN16v_atanf; + _ZGVbN4vv_hypotf; _ZGVcN8vv_hypotf; _ZGVdN8vv_hypotf; _ZGVeN16vv_hypotf; } } diff --git a/sysdeps/x86_64/fpu/libm-test-ulps b/sysdeps/x86_64/fpu/libm-test-ulps index ea0f833..a7513ec 100644 --- a/sysdeps/x86_64/fpu/libm-test-ulps +++ b/sysdeps/x86_64/fpu/libm-test-ulps @@ -1375,6 +1375,26 @@ double: 1 float128: 1 ldouble: 1 +Function: "hypot_vlen16": +float: 1 + +Function: "hypot_vlen2": +double: 1 + +Function: "hypot_vlen4": +double: 1 +float: 1 + +Function: "hypot_vlen4_avx2": +double: 1 + +Function: "hypot_vlen8": +double: 1 +float: 1 + +Function: "hypot_vlen8_avx2": +float: 1 + Function: "j0": double: 3 float: 9 diff --git a/sysdeps/x86_64/fpu/multiarch/svml_d_hypot2_core-sse2.S b/sysdeps/x86_64/fpu/multiarch/svml_d_hypot2_core-sse2.S new file mode 100644 index 0000000..237e384 --- /dev/null +++ b/sysdeps/x86_64/fpu/multiarch/svml_d_hypot2_core-sse2.S @@ -0,0 +1,20 @@ +/* SSE2 version of vectorized hypot. + 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 _ZGVbN2vv_hypot _ZGVbN2vv_hypot_sse2 +#include "../svml_d_hypot2_core.S" diff --git a/sysdeps/x86_64/fpu/multiarch/svml_d_hypot2_core.c b/sysdeps/x86_64/fpu/multiarch/svml_d_hypot2_core.c new file mode 100644 index 0000000..3f0865f --- /dev/null +++ b/sysdeps/x86_64/fpu/multiarch/svml_d_hypot2_core.c @@ -0,0 +1,28 @@ +/* Multiple versions of vectorized hypot, 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 _ZGVbN2vv_hypot +#include "ifunc-mathvec-sse4_1.h" + +libc_ifunc_redirected (REDIRECT_NAME, SYMBOL_NAME, IFUNC_SELECTOR ()); + +#ifdef SHARED +__hidden_ver1 (_ZGVbN2vv_hypot, __GI__ZGVbN2vv_hypot, + __redirect__ZGVbN2vv_hypot) + __attribute__ ((visibility ("hidden"))); +#endif diff --git a/sysdeps/x86_64/fpu/multiarch/svml_d_hypot2_core_sse4.S b/sysdeps/x86_64/fpu/multiarch/svml_d_hypot2_core_sse4.S new file mode 100644 index 0000000..931f34e --- /dev/null +++ b/sysdeps/x86_64/fpu/multiarch/svml_d_hypot2_core_sse4.S @@ -0,0 +1,279 @@ +/* Function hypot 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: + * + * HIGH LEVEL OVERVIEW + * + * Calculate z = (x*x+y*y) + * Calculate reciplicle sqrt (z) + * Calculate error = z*(rsqrt(z)*rsqrt(z)) - 1 + * Calculate fixing part p with polynom + * Fix answer with sqrt(z) = z * rsqrt(z) + error * p * z + * + * ALGORITHM DETAILS + * + * Multiprecision branch for _HA_ only + * Remove sigm from both arguments + * Find maximum (_x) and minimum (_y) (by abs value) between arguments + * Split _x int _a and _b for multiprecision + * If _x >> _y we will we will not split _y for multiprecision + * all _y will be put into lower part (_d) and higher part (_c = 0) + * Fixing _hilo_mask for the case _x >> _y + * Split _y into _c and _d for multiprecision with fixed mask + * + * compute Hi and Lo parts of _z = _x*_x + _y*_y + * + * _zHi = _a*_a + _c*_c + * _zLo = (_x + _a)*_b + _d*_y + _d*_c + * _z = _zHi + _zLo + * + * No multiprecision branch for _LA_ and _EP_ + * _z = _VARG1 * _VARG1 + _VARG2 * _VARG2 + * + * Check _z exponent to be withing borders [3BC ; 441] else goto Callout + * + * _s ~ 1.0/sqrt(_z) + * _s2 ~ 1.0/(sqrt(_z)*sqrt(_z)) ~ 1.0/_z = (1.0/_z + O) + * _e[rror] = (1.0/_z + O) * _z - 1.0 + * calculate fixing part _p + * _p = (((_POLY_C5*_e + _POLY_C4)*_e +_POLY_C3)*_e +_POLY_C2)*_e + _POLY_C1 + * some parts of polynom are skipped for lower flav + * + * result = _z * (1.0/sqrt(_z) + O) + _p * _e[rror] * _z + * + * + */ + +/* Offsets for data table __svml_dhypot_data_internal + */ +#define _dHiLoMask 0 +#define _dAbsMask 16 +#define _dOne 32 +#define _POLY_C5 48 +#define _POLY_C4 64 +#define _POLY_C3 80 +#define _POLY_C2 96 +#define _POLY_C1 112 +#define _LowBoundary 128 +#define _HighBoundary 144 + +#include <sysdep.h> + + .text + .section .text.sse4,"ax",@progbits +ENTRY(_ZGVbN2vv_hypot_sse4) + subq $88, %rsp + cfi_def_cfa_offset(96) + +/* + * Defines + * Implementation + * Multiprecision branch for _HA_ only + * _z = _VARG1 * _VARG1 + _VARG2 * _VARG2 + */ + movaps %xmm0, %xmm10 + movaps %xmm1, %xmm2 + mulpd %xmm0, %xmm10 + mulpd %xmm1, %xmm2 + addpd %xmm2, %xmm10 + +/* + * _s ~ 1.0/sqrt(_z) + * _s2 ~ 1.0/(sqrt(_z)*sqrt(_z)) ~ 1.0/_z + */ + cvtpd2ps %xmm10, %xmm7 + movlhps %xmm7, %xmm7 + rsqrtps %xmm7, %xmm8 + cvtps2pd %xmm8, %xmm11 + movaps %xmm11, %xmm2 + mulpd %xmm11, %xmm2 + +/* _e[rror] ~ (1.0/_z + O) * _z - 1.0 */ + mulpd %xmm10, %xmm2 + subpd _dOne+__svml_dhypot_data_internal(%rip), %xmm2 + +/* + * calculate fixing part _p + * _p = (((_POLY_C5*_e + _POLY_C4)*_e +_POLY_C3)*_e +_POLY_C2)*_e + _POLY_C1 + * some parts of polynom are skipped for lower flav + */ + movups _POLY_C4+__svml_dhypot_data_internal(%rip), %xmm9 + mulpd %xmm2, %xmm9 + addpd _POLY_C3+__svml_dhypot_data_internal(%rip), %xmm9 + mulpd %xmm2, %xmm9 + addpd _POLY_C2+__svml_dhypot_data_internal(%rip), %xmm9 + mulpd %xmm2, %xmm9 + addpd _POLY_C1+__svml_dhypot_data_internal(%rip), %xmm9 + +/* result = _z * (1.0/sqrt(_z) + O) + _p * _e[rror] * _z */ + mulpd %xmm9, %xmm2 + mulpd %xmm11, %xmm2 + mulpd %xmm10, %xmm11 + mulpd %xmm10, %xmm2 + +/* Check _z exponent to be withing borders [3BC ; 441] else goto Callout */ + movq _LowBoundary+__svml_dhypot_data_internal(%rip), %xmm5 + movq _HighBoundary+__svml_dhypot_data_internal(%rip), %xmm3 + pshufd $221, %xmm10, %xmm4 + pcmpgtd %xmm4, %xmm5 + pcmpgtd %xmm3, %xmm4 + por %xmm4, %xmm5 + pshufd $80, %xmm5, %xmm6 + movmskpd %xmm6, %edx + addpd %xmm11, %xmm2 + +/* The end of implementation */ + testl %edx, %edx + +/* Go to special inputs processing branch */ + jne L(SPECIAL_VALUES_BRANCH) + # LOE rbx rbp r12 r13 r14 r15 edx xmm0 xmm1 xmm2 + +/* Restore registers + * and exit the function + */ + +L(EXIT): + movaps %xmm2, %xmm0 + addq $88, %rsp + cfi_def_cfa_offset(8) + ret + cfi_def_cfa_offset(96) + +/* Branch to process + * special inputs + */ + +L(SPECIAL_VALUES_BRANCH): + movups %xmm0, 32(%rsp) + movups %xmm1, 48(%rsp) + movups %xmm2, 64(%rsp) + # LOE rbx rbp r12 r13 r14 r15 edx + + xorl %eax, %eax + movq %r12, 16(%rsp) + cfi_offset(12, -80) + movl %eax, %r12d + movq %r13, 8(%rsp) + cfi_offset(13, -88) + movl %edx, %r13d + movq %r14, (%rsp) + cfi_offset(14, -96) + # 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 64(%rsp), %xmm2 + +/* Go to exit */ + jmp L(EXIT) + cfi_offset(12, -80) + cfi_offset(13, -88) + cfi_offset(14, -96) + # LOE rbx rbp r12 r13 r14 r15 xmm2 + +/* Scalar math fucntion call + * to process special input + */ + +L(SCALAR_MATH_CALL): + movl %r12d, %r14d + movsd 32(%rsp,%r14,8), %xmm0 + movsd 48(%rsp,%r14,8), %xmm1 + call hypot@PLT + # LOE rbx rbp r14 r15 r12d r13d xmm0 + + movsd %xmm0, 64(%rsp,%r14,8) + +/* Process special inputs in loop */ + jmp L(SPECIAL_VALUES_LOOP) + # LOE rbx rbp r15 r12d r13d +END(_ZGVbN2vv_hypot_sse4) + + .section .rodata, "a" + .align 16 + +#ifdef __svml_dhypot_data_internal_typedef +typedef unsigned int VUINT32; +typedef struct +{ + __declspec(align(16)) VUINT32 _dHiLoMask[2][2]; + __declspec(align(16)) VUINT32 _dAbsMask[2][2]; + __declspec(align(16)) VUINT32 _dOne[2][2]; + __declspec(align(16)) VUINT32 _POLY_C5[2][2]; + __declspec(align(16)) VUINT32 _POLY_C4[2][2]; + __declspec(align(16)) VUINT32 _POLY_C3[2][2]; + __declspec(align(16)) VUINT32 _POLY_C2[2][2]; + __declspec(align(16)) VUINT32 _POLY_C1[2][2]; + __declspec(align(16)) VUINT32 _LowBoundary[4][1]; + __declspec(align(16)) VUINT32 _HighBoundary[4][1]; +} __svml_dhypot_data_internal; +#endif +__svml_dhypot_data_internal: + /* legacy algorithm */ + .quad 0xffffc00000000000, 0xffffc00000000000 /* _dHiLoMask */ + .align 16 + .quad 0x7fffffffffffffff, 0x7fffffffffffffff /* _dAbsMask */ + .align 16 + .quad 0x3FF0000000000000, 0x3FF0000000000000 /* _dOne */ + .align 16 + .quad 0xBFCF800000000000, 0xBFCF800000000000 /* _POLY_C5 */ + .align 16 + .quad 0x3FD1800000000000, 0x3FD1800000000000 /* _POLY_C4 */ + .align 16 + .quad 0xBFD4000000000000, 0xBFD4000000000000 /* _POLY_C3 */ + .align 16 + .quad 0x3FD8000000000000, 0x3FD8000000000000 /* _POLY_C2 */ + .align 16 + .quad 0xBFE0000000000000, 0xBFE0000000000000 /* _POLY_C1 */ + .align 16 + .long 0x3BC00000, 0x3BC00000, 0x3BC00000, 0x3BC00000 /* _LowBoundary */ + .align 16 + .long 0x44100000, 0x44100000, 0x44100000, 0x44100000 /* _HighBoundary */ + .align 16 + .type __svml_dhypot_data_internal,@object + .size __svml_dhypot_data_internal,.-__svml_dhypot_data_internal diff --git a/sysdeps/x86_64/fpu/multiarch/svml_d_hypot4_core-sse.S b/sysdeps/x86_64/fpu/multiarch/svml_d_hypot4_core-sse.S new file mode 100644 index 0000000..5e7c75c --- /dev/null +++ b/sysdeps/x86_64/fpu/multiarch/svml_d_hypot4_core-sse.S @@ -0,0 +1,20 @@ +/* SSE version of vectorized hypot. + 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 _ZGVdN4vv_hypot _ZGVdN4vv_hypot_sse_wrapper +#include "../svml_d_hypot4_core.S" diff --git a/sysdeps/x86_64/fpu/multiarch/svml_d_hypot4_core.c b/sysdeps/x86_64/fpu/multiarch/svml_d_hypot4_core.c new file mode 100644 index 0000000..06f34d3 --- /dev/null +++ b/sysdeps/x86_64/fpu/multiarch/svml_d_hypot4_core.c @@ -0,0 +1,28 @@ +/* Multiple versions of vectorized hypot, 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 _ZGVdN4vv_hypot +#include "ifunc-mathvec-avx2.h" + +libc_ifunc_redirected (REDIRECT_NAME, SYMBOL_NAME, IFUNC_SELECTOR ()); + +#ifdef SHARED +__hidden_ver1 (_ZGVdN4vv_hypot, __GI__ZGVdN4vv_hypot, + __redirect__ZGVdN4vv_hypot) + __attribute__ ((visibility ("hidden"))); +#endif diff --git a/sysdeps/x86_64/fpu/multiarch/svml_d_hypot4_core_avx2.S b/sysdeps/x86_64/fpu/multiarch/svml_d_hypot4_core_avx2.S new file mode 100644 index 0000000..45028ab --- /dev/null +++ b/sysdeps/x86_64/fpu/multiarch/svml_d_hypot4_core_avx2.S @@ -0,0 +1,289 @@ +/* Function hypot 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: + * + * HIGH LEVEL OVERVIEW + * + * Calculate z = (x*x+y*y) + * Calculate reciplicle sqrt (z) + * Calculate error = z*(rsqrt(z)*rsqrt(z)) - 1 + * Calculate fixing part p with polynom + * Fix answer with sqrt(z) = z * rsqrt(z) + error * p * z + * + * ALGORITHM DETAILS + * + * Multiprecision branch for _HA_ only + * Remove sigm from both arguments + * Find maximum (_x) and minimum (_y) (by abs value) between arguments + * Split _x int _a and _b for multiprecision + * If _x >> _y we will we will not split _y for multiprecision + * all _y will be put into lower part (_d) and higher part (_c = 0) + * Fixing _hilo_mask for the case _x >> _y + * Split _y into _c and _d for multiprecision with fixed mask + * + * compute Hi and Lo parts of _z = _x*_x + _y*_y + * + * _zHi = _a*_a + _c*_c + * _zLo = (_x + _a)*_b + _d*_y + _d*_c + * _z = _zHi + _zLo + * + * No multiprecision branch for _LA_ and _EP_ + * _z = _VARG1 * _VARG1 + _VARG2 * _VARG2 + * + * Check _z exponent to be withing borders [3BC ; 441] else goto Callout + * + * _s ~ 1.0/sqrt(_z) + * _s2 ~ 1.0/(sqrt(_z)*sqrt(_z)) ~ 1.0/_z = (1.0/_z + O) + * _e[rror] = (1.0/_z + O) * _z - 1.0 + * calculate fixing part _p + * _p = (((_POLY_C5*_e + _POLY_C4)*_e +_POLY_C3)*_e +_POLY_C2)*_e + _POLY_C1 + * some parts of polynom are skipped for lower flav + * + * result = _z * (1.0/sqrt(_z) + O) + _p * _e[rror] * _z + * + * + */ + +/* Offsets for data table __svml_dhypot_data_internal + */ +#define _dHiLoMask 0 +#define _dAbsMask 32 +#define _dOne 64 +#define _POLY_C5 96 +#define _POLY_C4 128 +#define _POLY_C3 160 +#define _POLY_C2 192 +#define _POLY_C1 224 +#define _LowBoundary 256 +#define _HighBoundary 288 + +#include <sysdep.h> + + .text + .section .text.avx2,"ax",@progbits +ENTRY(_ZGVdN4vv_hypot_avx2) + pushq %rbp + cfi_def_cfa_offset(16) + movq %rsp, %rbp + cfi_def_cfa(6, 16) + cfi_offset(6, -16) + andq $-32, %rsp + subq $128, %rsp + vmovapd %ymm1, %ymm2 + vmovapd %ymm0, %ymm1 + +/* + * Defines + * Implementation + * Multiprecision branch for _HA_ only + * _z = _VARG1 * _VARG1 + _VARG2 * _VARG2 + */ + vmulpd %ymm1, %ymm1, %ymm0 + +/* + * calculate fixing part _p + * _p = (((_POLY_C5*_e + _POLY_C4)*_e +_POLY_C3)*_e +_POLY_C2)*_e + _POLY_C1 + * some parts of polynom are skipped for lower flav + */ + vmovupd _POLY_C4+__svml_dhypot_data_internal(%rip), %ymm15 + vmovups _LowBoundary+__svml_dhypot_data_internal(%rip), %xmm4 + vfmadd231pd %ymm2, %ymm2, %ymm0 + +/* + * _s ~ 1.0/sqrt(_z) + * _s2 ~ 1.0/(sqrt(_z)*sqrt(_z)) ~ 1.0/_z + */ + vcvtpd2ps %ymm0, %xmm12 + +/* Check _z exponent to be withing borders [3BC ; 441] else goto Callout */ + vextractf128 $1, %ymm0, %xmm3 + vrsqrtps %xmm12, %xmm13 + vshufps $221, %xmm3, %xmm0, %xmm5 + vcvtps2pd %xmm13, %ymm3 + vpcmpgtd %xmm5, %xmm4, %xmm6 + vpcmpgtd _HighBoundary+__svml_dhypot_data_internal(%rip), %xmm5, %xmm7 + vpor %xmm7, %xmm6, %xmm9 + vpshufd $80, %xmm9, %xmm8 + vmulpd %ymm3, %ymm3, %ymm14 + vpshufd $250, %xmm9, %xmm10 + +/* _e[rror] ~ (1.0/_z + O) * _z - 1.0 */ + vfmsub213pd _dOne+__svml_dhypot_data_internal(%rip), %ymm0, %ymm14 + vfmadd213pd _POLY_C3+__svml_dhypot_data_internal(%rip), %ymm14, %ymm15 + vfmadd213pd _POLY_C2+__svml_dhypot_data_internal(%rip), %ymm14, %ymm15 + vfmadd213pd _POLY_C1+__svml_dhypot_data_internal(%rip), %ymm14, %ymm15 + +/* result = _z * (1.0/sqrt(_z) + O) + _p * _e[rror] * _z */ + vmulpd %ymm15, %ymm14, %ymm14 + vmulpd %ymm14, %ymm3, %ymm15 + vmulpd %ymm15, %ymm0, %ymm4 + vfmadd213pd %ymm4, %ymm3, %ymm0 + vinsertf128 $1, %xmm10, %ymm8, %ymm11 + vmovmskpd %ymm11, %edx + +/* The end of implementation */ + testl %edx, %edx + +/* Go to special inputs processing branch */ + jne L(SPECIAL_VALUES_BRANCH) + # LOE rbx r12 r13 r14 r15 edx ymm0 ymm1 ymm2 + +/* 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 %ymm1, 32(%rsp) + vmovupd %ymm2, 64(%rsp) + vmovupd %ymm0, 96(%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: -112; DW_OP_plus) */ + .cfi_escape 0x10, 0x0c, 0x0e, 0x38, 0x1c, 0x0d, 0xe0, 0xff, 0xff, 0xff, 0x1a, 0x0d, 0x90, 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: -120; DW_OP_plus) */ + .cfi_escape 0x10, 0x0d, 0x0e, 0x38, 0x1c, 0x0d, 0xe0, 0xff, 0xff, 0xff, 0x1a, 0x0d, 0x88, 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: -128; DW_OP_plus) */ + .cfi_escape 0x10, 0x0e, 0x0e, 0x38, 0x1c, 0x0d, 0xe0, 0xff, 0xff, 0xff, 0x1a, 0x0d, 0x80, 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 96(%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: -112; DW_OP_plus) */ + .cfi_escape 0x10, 0x0c, 0x0e, 0x38, 0x1c, 0x0d, 0xe0, 0xff, 0xff, 0xff, 0x1a, 0x0d, 0x90, 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: -120; DW_OP_plus) */ + .cfi_escape 0x10, 0x0d, 0x0e, 0x38, 0x1c, 0x0d, 0xe0, 0xff, 0xff, 0xff, 0x1a, 0x0d, 0x88, 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: -128; DW_OP_plus) */ + .cfi_escape 0x10, 0x0e, 0x0e, 0x38, 0x1c, 0x0d, 0xe0, 0xff, 0xff, 0xff, 0x1a, 0x0d, 0x80, 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 + movsd 64(%rsp,%r14,8), %xmm1 + call hypot@PLT + # LOE rbx r14 r15 r12d r13d xmm0 + + movsd %xmm0, 96(%rsp,%r14,8) + +/* Process special inputs in loop */ + jmp L(SPECIAL_VALUES_LOOP) + # LOE rbx r15 r12d r13d +END(_ZGVdN4vv_hypot_avx2) + + .section .rodata, "a" + .align 32 + +#ifdef __svml_dhypot_data_internal_typedef +typedef unsigned int VUINT32; +typedef struct +{ + __declspec(align(32)) VUINT32 _dHiLoMask[4][2]; + __declspec(align(32)) VUINT32 _dAbsMask[4][2]; + __declspec(align(32)) VUINT32 _dOne[4][2]; + __declspec(align(32)) VUINT32 _POLY_C5[4][2]; + __declspec(align(32)) VUINT32 _POLY_C4[4][2]; + __declspec(align(32)) VUINT32 _POLY_C3[4][2]; + __declspec(align(32)) VUINT32 _POLY_C2[4][2]; + __declspec(align(32)) VUINT32 _POLY_C1[4][2]; + __declspec(align(32)) VUINT32 _LowBoundary[8][1]; + __declspec(align(32)) VUINT32 _HighBoundary[8][1]; +} __svml_dhypot_data_internal; +#endif +__svml_dhypot_data_internal: + /* legacy algorithm */ + .quad 0xffffc00000000000, 0xffffc00000000000, 0xffffc00000000000, 0xffffc00000000000 /* _dHiLoMask */ + .align 32 + .quad 0x7fffffffffffffff, 0x7fffffffffffffff, 0x7fffffffffffffff, 0x7fffffffffffffff /* _dAbsMask */ + .align 32 + .quad 0x3FF0000000000000, 0x3FF0000000000000, 0x3FF0000000000000, 0x3FF0000000000000 /* _dOne */ + .align 32 + .quad 0xBFCF800000000000, 0xBFCF800000000000, 0xBFCF800000000000, 0xBFCF800000000000 /* _POLY_C5 */ + .align 32 + .quad 0x3FD1800000000000, 0x3FD1800000000000, 0x3FD1800000000000, 0x3FD1800000000000 /* _POLY_C4 */ + .align 32 + .quad 0xBFD4000000000000, 0xBFD4000000000000, 0xBFD4000000000000, 0xBFD4000000000000 /* _POLY_C3 */ + .align 32 + .quad 0x3FD8000000000000, 0x3FD8000000000000, 0x3FD8000000000000, 0x3FD8000000000000 /* _POLY_C2 */ + .align 32 + .quad 0xBFE0000000000000, 0xBFE0000000000000, 0xBFE0000000000000, 0xBFE0000000000000 /* _POLY_C1 */ + .align 32 + .long 0x3BC00000, 0x3BC00000, 0x3BC00000, 0x3BC00000, 0x3BC00000, 0x3BC00000, 0x3BC00000, 0x3BC00000 /* _LowBoundary */ + .align 32 + .long 0x44100000, 0x44100000, 0x44100000, 0x44100000, 0x44100000, 0x44100000, 0x44100000, 0x44100000 /* _HighBoundary */ + .align 32 + .type __svml_dhypot_data_internal,@object + .size __svml_dhypot_data_internal,.-__svml_dhypot_data_internal diff --git a/sysdeps/x86_64/fpu/multiarch/svml_d_hypot8_core-avx2.S b/sysdeps/x86_64/fpu/multiarch/svml_d_hypot8_core-avx2.S new file mode 100644 index 0000000..a53e82c --- /dev/null +++ b/sysdeps/x86_64/fpu/multiarch/svml_d_hypot8_core-avx2.S @@ -0,0 +1,20 @@ +/* AVX2 version of vectorized hypot. + 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 _ZGVeN8vv_hypot _ZGVeN8vv_hypot_avx2_wrapper +#include "../svml_d_hypot8_core.S" diff --git a/sysdeps/x86_64/fpu/multiarch/svml_d_hypot8_core.c b/sysdeps/x86_64/fpu/multiarch/svml_d_hypot8_core.c new file mode 100644 index 0000000..6052c75 --- /dev/null +++ b/sysdeps/x86_64/fpu/multiarch/svml_d_hypot8_core.c @@ -0,0 +1,28 @@ +/* Multiple versions of vectorized hypot, 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 _ZGVeN8vv_hypot +#include "ifunc-mathvec-avx512-skx.h" + +libc_ifunc_redirected (REDIRECT_NAME, SYMBOL_NAME, IFUNC_SELECTOR ()); + +#ifdef SHARED +__hidden_ver1 (_ZGVeN8vv_hypot, __GI__ZGVeN8vv_hypot, + __redirect__ZGVeN8vv_hypot) + __attribute__ ((visibility ("hidden"))); +#endif diff --git a/sysdeps/x86_64/fpu/multiarch/svml_d_hypot8_core_avx512.S b/sysdeps/x86_64/fpu/multiarch/svml_d_hypot8_core_avx512.S new file mode 100644 index 0000000..1e5e716 --- /dev/null +++ b/sysdeps/x86_64/fpu/multiarch/svml_d_hypot8_core_avx512.S @@ -0,0 +1,235 @@ +/* Function hypot 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: + * + * HIGH LEVEL OVERVIEW + * + * Calculate z = (x*x+y*y) + * Calculate reciplicle sqrt (z) + * Calculate error = z*(rsqrt(z)*rsqrt(z)) - 1 + * Calculate fixing part p with polynom + * Fix answer with sqrt(z) = z * rsqrt(z) + error * p * z + * + * ALGORITHM DETAILS + * + * Multiprecision branch for _HA_ only + * Remove sigm from both arguments + * Find maximum (_x) and minimum (_y) (by abs value) between arguments + * Split _x int _a and _b for multiprecision + * If _x >> _y we will we will not split _y for multiprecision + * all _y will be put into lower part (_d) and higher part (_c = 0) + * Fixing _hilo_mask for the case _x >> _y + * Split _y into _c and _d for multiprecision with fixed mask + * + * compute Hi and Lo parts of _z = _x*_x + _y*_y + * + * _zHi = _a*_a + _c*_c + * _zLo = (_x + _a)*_b + _d*_y + _d*_c + * _z = _zHi + _zLo + * + * No multiprecision branch for _LA_ and _EP_ + * _z = _VARG1 * _VARG1 + _VARG2 * _VARG2 + * + * Check _z exponent to be withing borders [3BC ; 441] else goto Callout + * + * _s ~ 1.0/sqrt(_z) + * _s2 ~ 1.0/(sqrt(_z)*sqrt(_z)) ~ 1.0/_z = (1.0/_z + O) + * _e[rror] = (1.0/_z + O) * _z - 1.0 + * calculate fixing part _p + * _p = (((_POLY_C5*_e + _POLY_C4)*_e +_POLY_C3)*_e +_POLY_C2)*_e + _POLY_C1 + * some parts of polynom are skipped for lower flav + * + * result = _z * (1.0/sqrt(_z) + O) + _p * _e[rror] * _z + * + * + */ + +/* Offsets for data table __svml_dhypot_data_internal + */ +#define _dAbsMask 0 +#define _lExpBound_uisa 64 +#define _lExpBound 128 +#define _dHalf 192 + +#include <sysdep.h> + + .text + .section .text.evex512,"ax",@progbits +ENTRY(_ZGVeN8vv_hypot_skx) + pushq %rbp + cfi_def_cfa_offset(16) + movq %rsp, %rbp + cfi_def_cfa(6, 16) + cfi_offset(6, -16) + andq $-64, %rsp + subq $256, %rsp + vgetexppd {sae}, %zmm0, %zmm2 + vgetexppd {sae}, %zmm1, %zmm3 + vmovups _dHalf+__svml_dhypot_data_internal(%rip), %zmm9 + vmaxpd {sae}, %zmm3, %zmm2, %zmm4 + vmulpd {rn-sae}, %zmm0, %zmm0, %zmm2 + vandpd _dAbsMask+__svml_dhypot_data_internal(%rip), %zmm4, %zmm5 + vfmadd231pd {rn-sae}, %zmm1, %zmm1, %zmm2 + +/* Select exponent bound so that no scaling is needed */ + vpcmpq $5, _lExpBound_uisa+__svml_dhypot_data_internal(%rip), %zmm5, %k0 + vrsqrt14pd %zmm2, %zmm6 + kmovw %k0, %edx + vmulpd {rn-sae}, %zmm6, %zmm2, %zmm7 + vmulpd {rn-sae}, %zmm6, %zmm9, %zmm8 + vfnmadd231pd {rn-sae}, %zmm7, %zmm8, %zmm9 + vfmadd231pd {rn-sae}, %zmm9, %zmm8, %zmm8 + vfmadd213pd {rn-sae}, %zmm7, %zmm7, %zmm9 + vfnmadd231pd {rn-sae}, %zmm9, %zmm9, %zmm2 + vfmadd213pd {rn-sae}, %zmm9, %zmm8, %zmm2 + +/* The end of implementation */ + testl %edx, %edx + +/* Go to special inputs processing branch */ + jne L(SPECIAL_VALUES_BRANCH) + # LOE rbx r12 r13 r14 r15 edx zmm0 zmm1 zmm2 + +/* Restore registers + * and exit the function + */ + +L(EXIT): + vmovaps %zmm2, %zmm0 + movq %rbp, %rsp + popq %rbp + cfi_def_cfa(7, 8) + cfi_restore(6) + ret + cfi_def_cfa(6, 16) + cfi_offset(6, -16) + +/* Branch to process + * special inputs + */ + +L(SPECIAL_VALUES_BRANCH): + vmovups %zmm0, 64(%rsp) + vmovups %zmm1, 128(%rsp) + vmovups %zmm2, 192(%rsp) + # LOE rbx r12 r13 r14 r15 edx zmm2 + + xorl %eax, %eax + # LOE rbx r12 r13 r14 r15 eax edx + + vzeroupper + movq %r12, 16(%rsp) + /* DW_CFA_expression: r12 (r12) (DW_OP_lit8; DW_OP_minus; DW_OP_const4s: -64; DW_OP_and; DW_OP_const4s: -240; DW_OP_plus) */ + .cfi_escape 0x10, 0x0c, 0x0e, 0x38, 0x1c, 0x0d, 0xc0, 0xff, 0xff, 0xff, 0x1a, 0x0d, 0x10, 0xff, 0xff, 0xff, 0x22 + movl %eax, %r12d + movq %r13, 8(%rsp) + /* DW_CFA_expression: r13 (r13) (DW_OP_lit8; DW_OP_minus; DW_OP_const4s: -64; DW_OP_and; DW_OP_const4s: -248; DW_OP_plus) */ + .cfi_escape 0x10, 0x0d, 0x0e, 0x38, 0x1c, 0x0d, 0xc0, 0xff, 0xff, 0xff, 0x1a, 0x0d, 0x08, 0xff, 0xff, 0xff, 0x22 + movl %edx, %r13d + movq %r14, (%rsp) + /* DW_CFA_expression: r14 (r14) (DW_OP_lit8; DW_OP_minus; DW_OP_const4s: -64; DW_OP_and; DW_OP_const4s: -256; DW_OP_plus) */ + .cfi_escape 0x10, 0x0e, 0x0e, 0x38, 0x1c, 0x0d, 0xc0, 0xff, 0xff, 0xff, 0x1a, 0x0d, 0x00, 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 192(%rsp), %zmm2 + +/* Go to exit */ + jmp L(EXIT) + /* DW_CFA_expression: r12 (r12) (DW_OP_lit8; DW_OP_minus; DW_OP_const4s: -64; DW_OP_and; DW_OP_const4s: -240; DW_OP_plus) */ + .cfi_escape 0x10, 0x0c, 0x0e, 0x38, 0x1c, 0x0d, 0xc0, 0xff, 0xff, 0xff, 0x1a, 0x0d, 0x10, 0xff, 0xff, 0xff, 0x22 + /* DW_CFA_expression: r13 (r13) (DW_OP_lit8; DW_OP_minus; DW_OP_const4s: -64; DW_OP_and; DW_OP_const4s: -248; DW_OP_plus) */ + .cfi_escape 0x10, 0x0d, 0x0e, 0x38, 0x1c, 0x0d, 0xc0, 0xff, 0xff, 0xff, 0x1a, 0x0d, 0x08, 0xff, 0xff, 0xff, 0x22 + /* DW_CFA_expression: r14 (r14) (DW_OP_lit8; DW_OP_minus; DW_OP_const4s: -64; DW_OP_and; DW_OP_const4s: -256; DW_OP_plus) */ + .cfi_escape 0x10, 0x0e, 0x0e, 0x38, 0x1c, 0x0d, 0xc0, 0xff, 0xff, 0xff, 0x1a, 0x0d, 0x00, 0xff, 0xff, 0xff, 0x22 + # LOE rbx r12 r13 r14 r15 zmm2 + +/* Scalar math fucntion call + * to process special input + */ + +L(SCALAR_MATH_CALL): + movl %r12d, %r14d + movsd 64(%rsp,%r14,8), %xmm0 + movsd 128(%rsp,%r14,8), %xmm1 + call hypot@PLT + # LOE rbx r14 r15 r12d r13d xmm0 + + movsd %xmm0, 192(%rsp,%r14,8) + +/* Process special inputs in loop */ + jmp L(SPECIAL_VALUES_LOOP) + # LOE rbx r15 r12d r13d +END(_ZGVeN8vv_hypot_skx) + + .section .rodata, "a" + .align 64 + +#ifdef __svml_dhypot_data_internal_typedef +typedef unsigned int VUINT32; +typedef struct +{ + __declspec(align(64)) VUINT32 _dAbsMask[8][2]; + __declspec(align(64)) VUINT32 _lExpBound_uisa[8][2]; + __declspec(align(64)) VUINT32 _lExpBound[8][2]; + __declspec(align(64)) VUINT32 _dHalf[8][2]; +} __svml_dhypot_data_internal; +#endif +__svml_dhypot_data_internal: + /* legacy algorithm */ + .quad 0x7fffffffffffffff, 0x7fffffffffffffff, 0x7fffffffffffffff, 0x7fffffffffffffff, 0x7fffffffffffffff, 0x7fffffffffffffff, 0x7fffffffffffffff, 0x7fffffffffffffff /* _dAbsMask */ + /* fma based algorithm*/ + .align 64 + .quad 0x407ff00000000000, 0x407ff00000000000, 0x407ff00000000000, 0x407ff00000000000, 0x407ff00000000000, 0x407ff00000000000, 0x407ff00000000000, 0x407ff00000000000 /* _lExpBound_uisa */ + .align 64 + .quad 0x404f800000000000, 0x404f800000000000, 0x404f800000000000, 0x404f800000000000, 0x404f800000000000, 0x404f800000000000, 0x404f800000000000, 0x404f800000000000 /* _lExpBound */ + .align 64 + .quad 0x3FE0000000000000, 0x3FE0000000000000, 0x3FE0000000000000, 0x3FE0000000000000, 0x3FE0000000000000, 0x3FE0000000000000, 0x3FE0000000000000, 0x3FE0000000000000 /* _dHalf */ + .align 64 + .type __svml_dhypot_data_internal,@object + .size __svml_dhypot_data_internal,.-__svml_dhypot_data_internal diff --git a/sysdeps/x86_64/fpu/multiarch/svml_s_hypotf16_core-avx2.S b/sysdeps/x86_64/fpu/multiarch/svml_s_hypotf16_core-avx2.S new file mode 100644 index 0000000..a6ba40d --- /dev/null +++ b/sysdeps/x86_64/fpu/multiarch/svml_s_hypotf16_core-avx2.S @@ -0,0 +1,20 @@ +/* AVX2 version of vectorized hypotf. + 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 _ZGVeN16vv_hypotf _ZGVeN16vv_hypotf_avx2_wrapper +#include "../svml_s_hypotf16_core.S" diff --git a/sysdeps/x86_64/fpu/multiarch/svml_s_hypotf16_core.c b/sysdeps/x86_64/fpu/multiarch/svml_s_hypotf16_core.c new file mode 100644 index 0000000..0c9eb6a --- /dev/null +++ b/sysdeps/x86_64/fpu/multiarch/svml_s_hypotf16_core.c @@ -0,0 +1,28 @@ +/* Multiple versions of vectorized hypotf, 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 _ZGVeN16vv_hypotf +#include "ifunc-mathvec-avx512-skx.h" + +libc_ifunc_redirected (REDIRECT_NAME, SYMBOL_NAME, IFUNC_SELECTOR ()); + +#ifdef SHARED +__hidden_ver1 (_ZGVeN16vv_hypotf, __GI__ZGVeN16vv_hypotf, + __redirect__ZGVeN16vv_hypotf) + __attribute__ ((visibility ("hidden"))); +#endif diff --git a/sysdeps/x86_64/fpu/multiarch/svml_s_hypotf16_core_avx512.S b/sysdeps/x86_64/fpu/multiarch/svml_s_hypotf16_core_avx512.S new file mode 100644 index 0000000..46a156d --- /dev/null +++ b/sysdeps/x86_64/fpu/multiarch/svml_s_hypotf16_core_avx512.S @@ -0,0 +1,239 @@ +/* Function hypotf 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: + * + * HIGH LEVEL OVERVIEW + * + * Calculate z = (x*x+y*y) + * Calculate reciplicle sqrt (z) + * Calculate make two NR iterations + * + * ALGORITHM DETAILS + * + * Multiprecision branch for _HA_ only + * Remove sigm from both arguments + * Find maximum (_x) and minimum (_y) (by abs value) between arguments + * Split _x int _a and _b for multiprecision + * If _x >> _y we will we will not split _y for multiprecision + * all _y will be put into lower part (_d) and higher part (_c = 0) + * Fixing _hilo_mask for the case _x >> _y + * Split _y into _c and _d for multiprecision with fixed mask + * + * compute Hi and Lo parts of _z = _x*_x + _y*_y + * + * _zHi = _a*_a + _c*_c + * _zLo = (_x + _a)*_b + _d*_y + _d*_c + * _z = _zHi + _zLo + * + * No multiprecision branch for _LA_ and _EP_ + * _z = _VARG1 * _VARG1 + _VARG2 * _VARG2 + * + * Check _z exponent to be withing borders [1E3 ; 60A] else goto Callout + * + * Compute resciplicle sqrt s0 ~ 1.0/sqrt(_z), + * that multiplied by _z, is final result for _EP_ version. + * + * First iteration (or zero iteration): + * s = z * s0 + * h = .5 * s0 + * d = s * h - .5 + * + * Second iteration: + * h = d * h + h + * s = s * d + s + * d = s * s - z (in multiprecision for _HA_) + * + * result = s - h * d + * + * EP version of the function can be implemented as y[i]=sqrt(a[i]^2+b[i]^2) + * with all intermediate operations done in target precision for i=1,..,n. + * It can return result y[i]=0 in case a[i]^2 and b[i]^2 underflow in target + * precision (for some i). It can return result y[i]=NAN in case + * a[i]^2+b[i]^2 overflow in target precision, for some i. It can return + * result y[i]=NAN in case a[i] or b[i] is infinite, for some i. + * + * + */ + +/* Offsets for data table __svml_shypot_data_internal + */ +#define _sAbsMask 0 +#define _sHalf 64 +#define _iExpBound 128 + +#include <sysdep.h> + + .text + .section .text.exex512,"ax",@progbits +ENTRY(_ZGVeN16vv_hypotf_skx) + pushq %rbp + cfi_def_cfa_offset(16) + movq %rsp, %rbp + cfi_def_cfa(6, 16) + cfi_offset(6, -16) + andq $-64, %rsp + subq $256, %rsp + vgetexpps {sae}, %zmm0, %zmm2 + vgetexpps {sae}, %zmm1, %zmm3 + vmovups _sHalf+__svml_shypot_data_internal(%rip), %zmm6 + vmaxps {sae}, %zmm3, %zmm2, %zmm4 + vmulps {rn-sae}, %zmm0, %zmm0, %zmm2 + vandps _sAbsMask+__svml_shypot_data_internal(%rip), %zmm4, %zmm5 + vfmadd231ps {rn-sae}, %zmm1, %zmm1, %zmm2 + vpcmpd $5, _iExpBound+__svml_shypot_data_internal(%rip), %zmm5, %k0 + vrsqrt14ps %zmm2, %zmm7 + kmovw %k0, %edx + vmulps {rn-sae}, %zmm7, %zmm2, %zmm9 + vmulps {rn-sae}, %zmm7, %zmm6, %zmm8 + vfnmadd231ps {rn-sae}, %zmm9, %zmm9, %zmm2 + vfmadd213ps {rn-sae}, %zmm9, %zmm8, %zmm2 + +/* + * VSCALEF( S, _VRES1, _VRES1, sExp ); + * The end of implementation + */ + testl %edx, %edx + +/* Go to special inputs processing branch */ + jne L(SPECIAL_VALUES_BRANCH) + # LOE rbx r12 r13 r14 r15 edx zmm0 zmm1 zmm2 + +/* Restore registers + * and exit the function + */ + +L(EXIT): + vmovaps %zmm2, %zmm0 + movq %rbp, %rsp + popq %rbp + cfi_def_cfa(7, 8) + cfi_restore(6) + ret + cfi_def_cfa(6, 16) + cfi_offset(6, -16) + +/* Branch to process + * special inputs + */ + +L(SPECIAL_VALUES_BRANCH): + vmovups %zmm0, 64(%rsp) + vmovups %zmm1, 128(%rsp) + vmovups %zmm2, 192(%rsp) + # LOE rbx r12 r13 r14 r15 edx zmm2 + + xorl %eax, %eax + # LOE rbx r12 r13 r14 r15 eax edx + + vzeroupper + movq %r12, 16(%rsp) + /* DW_CFA_expression: r12 (r12) (DW_OP_lit8; DW_OP_minus; DW_OP_const4s: -64; DW_OP_and; DW_OP_const4s: -240; DW_OP_plus) */ + .cfi_escape 0x10, 0x0c, 0x0e, 0x38, 0x1c, 0x0d, 0xc0, 0xff, 0xff, 0xff, 0x1a, 0x0d, 0x10, 0xff, 0xff, 0xff, 0x22 + movl %eax, %r12d + movq %r13, 8(%rsp) + /* DW_CFA_expression: r13 (r13) (DW_OP_lit8; DW_OP_minus; DW_OP_const4s: -64; DW_OP_and; DW_OP_const4s: -248; DW_OP_plus) */ + .cfi_escape 0x10, 0x0d, 0x0e, 0x38, 0x1c, 0x0d, 0xc0, 0xff, 0xff, 0xff, 0x1a, 0x0d, 0x08, 0xff, 0xff, 0xff, 0x22 + movl %edx, %r13d + movq %r14, (%rsp) + /* DW_CFA_expression: r14 (r14) (DW_OP_lit8; DW_OP_minus; DW_OP_const4s: -64; DW_OP_and; DW_OP_const4s: -256; DW_OP_plus) */ + .cfi_escape 0x10, 0x0e, 0x0e, 0x38, 0x1c, 0x0d, 0xc0, 0xff, 0xff, 0xff, 0x1a, 0x0d, 0x00, 0xff, 0xff, 0xff, 0x22 + # LOE rbx r15 r12d r13d + +/* Range mask + * bits check + */ + +L(RANGEMASK_CHECK): + btl %r12d, %r13d + +/* Call scalar math function */ + jc L(SCALAR_MATH_CALL) + # LOE rbx r15 r12d r13d + +/* Special inputs + * processing loop + */ + +L(SPECIAL_VALUES_LOOP): + incl %r12d + cmpl $16, %r12d + +/* Check bits in range mask */ + jl L(RANGEMASK_CHECK) + # LOE rbx r15 r12d r13d + + movq 16(%rsp), %r12 + cfi_restore(12) + movq 8(%rsp), %r13 + cfi_restore(13) + movq (%rsp), %r14 + cfi_restore(14) + vmovups 192(%rsp), %zmm2 + +/* Go to exit */ + jmp L(EXIT) + /* DW_CFA_expression: r12 (r12) (DW_OP_lit8; DW_OP_minus; DW_OP_const4s: -64; DW_OP_and; DW_OP_const4s: -240; DW_OP_plus) */ + .cfi_escape 0x10, 0x0c, 0x0e, 0x38, 0x1c, 0x0d, 0xc0, 0xff, 0xff, 0xff, 0x1a, 0x0d, 0x10, 0xff, 0xff, 0xff, 0x22 + /* DW_CFA_expression: r13 (r13) (DW_OP_lit8; DW_OP_minus; DW_OP_const4s: -64; DW_OP_and; DW_OP_const4s: -248; DW_OP_plus) */ + .cfi_escape 0x10, 0x0d, 0x0e, 0x38, 0x1c, 0x0d, 0xc0, 0xff, 0xff, 0xff, 0x1a, 0x0d, 0x08, 0xff, 0xff, 0xff, 0x22 + /* DW_CFA_expression: r14 (r14) (DW_OP_lit8; DW_OP_minus; DW_OP_const4s: -64; DW_OP_and; DW_OP_const4s: -256; DW_OP_plus) */ + .cfi_escape 0x10, 0x0e, 0x0e, 0x38, 0x1c, 0x0d, 0xc0, 0xff, 0xff, 0xff, 0x1a, 0x0d, 0x00, 0xff, 0xff, 0xff, 0x22 + # LOE rbx r12 r13 r14 r15 zmm2 + +/* Scalar math fucntion call + * to process special input + */ + +L(SCALAR_MATH_CALL): + movl %r12d, %r14d + movss 64(%rsp,%r14,4), %xmm0 + movss 128(%rsp,%r14,4), %xmm1 + call hypotf@PLT + # LOE rbx r14 r15 r12d r13d xmm0 + + movss %xmm0, 192(%rsp,%r14,4) + +/* Process special inputs in loop */ + jmp L(SPECIAL_VALUES_LOOP) + # LOE rbx r15 r12d r13d +END(_ZGVeN16vv_hypotf_skx) + + .section .rodata, "a" + .align 64 + +#ifdef __svml_shypot_data_internal_typedef +typedef unsigned int VUINT32; +typedef struct +{ + __declspec(align(64)) VUINT32 _sAbsMask[16][1]; + __declspec(align(64)) VUINT32 _sHalf[16][1]; + __declspec(align(64)) VUINT32 _iExpBound[16][1]; +} __svml_shypot_data_internal; +#endif +__svml_shypot_data_internal: + .long 0x7fffffff, 0x7fffffff, 0x7fffffff, 0x7fffffff, 0x7fffffff, 0x7fffffff, 0x7fffffff, 0x7fffffff, 0x7fffffff, 0x7fffffff, 0x7fffffff, 0x7fffffff, 0x7fffffff, 0x7fffffff, 0x7fffffff, 0x7fffffff /* _sAbsMask */ + .align 64 + .long 0x3f000000, 0x3f000000, 0x3f000000, 0x3f000000, 0x3f000000, 0x3f000000, 0x3f000000, 0x3f000000, 0x3f000000, 0x3f000000, 0x3f000000, 0x3f000000, 0x3f000000, 0x3f000000, 0x3f000000, 0x3f000000 /* _sHalf */ + /* fma based algorithm*/ + .align 64 + .long 0x427C0000, 0x427C0000, 0x427C0000, 0x427C0000, 0x427C0000, 0x427C0000, 0x427C0000, 0x427C0000, 0x427C0000, 0x427C0000, 0x427C0000, 0x427C0000, 0x427C0000, 0x427C0000, 0x427C0000, 0x427C0000 /* _iExpBound */ + .align 64 + .type __svml_shypot_data_internal,@object + .size __svml_shypot_data_internal,.-__svml_shypot_data_internal diff --git a/sysdeps/x86_64/fpu/multiarch/svml_s_hypotf4_core-sse2.S b/sysdeps/x86_64/fpu/multiarch/svml_s_hypotf4_core-sse2.S new file mode 100644 index 0000000..5e9dd22 --- /dev/null +++ b/sysdeps/x86_64/fpu/multiarch/svml_s_hypotf4_core-sse2.S @@ -0,0 +1,20 @@ +/* SSE2 version of vectorized hypotf. + 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 _ZGVbN4vv_hypotf _ZGVbN4vv_hypotf_sse2 +#include "../svml_s_hypotf4_core.S" diff --git a/sysdeps/x86_64/fpu/multiarch/svml_s_hypotf4_core.c b/sysdeps/x86_64/fpu/multiarch/svml_s_hypotf4_core.c new file mode 100644 index 0000000..91c9f5c --- /dev/null +++ b/sysdeps/x86_64/fpu/multiarch/svml_s_hypotf4_core.c @@ -0,0 +1,28 @@ +/* Multiple versions of vectorized hypotf, 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 _ZGVbN4vv_hypotf +#include "ifunc-mathvec-sse4_1.h" + +libc_ifunc_redirected (REDIRECT_NAME, SYMBOL_NAME, IFUNC_SELECTOR ()); + +#ifdef SHARED +__hidden_ver1 (_ZGVbN4vv_hypotf, __GI__ZGVbN4vv_hypotf, + __redirect__ZGVbN4vv_hypotf) + __attribute__ ((visibility ("hidden"))); +#endif diff --git a/sysdeps/x86_64/fpu/multiarch/svml_s_hypotf4_core_sse4.S b/sysdeps/x86_64/fpu/multiarch/svml_s_hypotf4_core_sse4.S new file mode 100644 index 0000000..a3f6d21 --- /dev/null +++ b/sysdeps/x86_64/fpu/multiarch/svml_s_hypotf4_core_sse4.S @@ -0,0 +1,265 @@ +/* Function hypotf 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: + * + * HIGH LEVEL OVERVIEW + * + * Calculate z = (x*x+y*y) + * Calculate reciplicle sqrt (z) + * Calculate make two NR iterations + * + * ALGORITHM DETAILS + * + * Multiprecision branch for _HA_ only + * Remove sigm from both arguments + * Find maximum (_x) and minimum (_y) (by abs value) between arguments + * Split _x int _a and _b for multiprecision + * If _x >> _y we will we will not split _y for multiprecision + * all _y will be put into lower part (_d) and higher part (_c = 0) + * Fixing _hilo_mask for the case _x >> _y + * Split _y into _c and _d for multiprecision with fixed mask + * + * compute Hi and Lo parts of _z = _x*_x + _y*_y + * + * _zHi = _a*_a + _c*_c + * _zLo = (_x + _a)*_b + _d*_y + _d*_c + * _z = _zHi + _zLo + * + * No multiprecision branch for _LA_ and _EP_ + * _z = _VARG1 * _VARG1 + _VARG2 * _VARG2 + * + * Check _z exponent to be withing borders [1E3 ; 60A] else goto Callout + * + * Compute resciplicle sqrt s0 ~ 1.0/sqrt(_z), + * that multiplied by _z, is final result for _EP_ version. + * + * First iteration (or zero iteration): + * s = z * s0 + * h = .5 * s0 + * d = s * h - .5 + * + * Second iteration: + * h = d * h + h + * s = s * d + s + * d = s * s - z (in multiprecision for _HA_) + * + * result = s - h * d + * + * EP version of the function can be implemented as y[i]=sqrt(a[i]^2+b[i]^2) + * with all intermediate operations done in target precision for i=1,..,n. + * It can return result y[i]=0 in case a[i]^2 and b[i]^2 underflow in target + * precision (for some i). It can return result y[i]=NAN in case + * a[i]^2+b[i]^2 overflow in target precision, for some i. It can return + * result y[i]=NAN in case a[i] or b[i] is infinite, for some i. + * + * + */ + +/* Offsets for data table __svml_shypot_data_internal + */ +#define _sHiLoMask 0 +#define _sAbsMask 16 +#define _sHalf 32 +#define _LowBoundary 48 +#define _HighBoundary 64 + +#include <sysdep.h> + + .text + .section .text.sse4,"ax",@progbits +ENTRY(_ZGVbN4vv_hypotf_sse4) + subq $88, %rsp + cfi_def_cfa_offset(96) + +/* + * Implementation + * Multiprecision branch for _HA_ only + * No multiprecision branch for _LA_ + * _z = _VARG1 * _VARG1 + _VARG2 * _VARG2 + */ + movaps %xmm0, %xmm8 + movaps %xmm1, %xmm2 + mulps %xmm0, %xmm8 + mulps %xmm1, %xmm2 + +/* + * Variables + * Defines + * Constants loading + */ + movups _sHalf+__svml_shypot_data_internal(%rip), %xmm5 + addps %xmm2, %xmm8 + +/* _s0 ~ 1.0/sqrt(_z) */ + rsqrtps %xmm8, %xmm10 + +/* First iteration */ + movaps %xmm10, %xmm2 + movaps %xmm8, %xmm3 + mulps %xmm8, %xmm2 + mulps %xmm5, %xmm10 + movaps %xmm2, %xmm6 + mulps %xmm10, %xmm6 + +/* Check _z exponent to be withing borders [1E3 ; 60A] else goto Callout */ + movdqu _LowBoundary+__svml_shypot_data_internal(%rip), %xmm4 + subps %xmm6, %xmm5 + +/* Second iteration */ + movaps %xmm5, %xmm7 + pcmpgtd %xmm8, %xmm4 + mulps %xmm2, %xmm5 + mulps %xmm10, %xmm7 + addps %xmm5, %xmm2 + addps %xmm7, %xmm10 + +/* Finish second iteration in native precision for _LA_ */ + movaps %xmm2, %xmm9 + mulps %xmm2, %xmm9 + pcmpgtd _HighBoundary+__svml_shypot_data_internal(%rip), %xmm3 + subps %xmm8, %xmm9 + mulps %xmm9, %xmm10 + por %xmm3, %xmm4 + movmskps %xmm4, %edx + subps %xmm10, %xmm2 + +/* The end of implementation */ + testl %edx, %edx + +/* Go to special inputs processing branch */ + jne L(SPECIAL_VALUES_BRANCH) + # LOE rbx rbp r12 r13 r14 r15 edx xmm0 xmm1 xmm2 + +/* Restore registers + * and exit the function + */ + +L(EXIT): + movaps %xmm2, %xmm0 + addq $88, %rsp + cfi_def_cfa_offset(8) + ret + cfi_def_cfa_offset(96) + +/* Branch to process + * special inputs + */ + +L(SPECIAL_VALUES_BRANCH): + movups %xmm0, 32(%rsp) + movups %xmm1, 48(%rsp) + movups %xmm2, 64(%rsp) + # LOE rbx rbp r12 r13 r14 r15 edx + + xorl %eax, %eax + movq %r12, 16(%rsp) + cfi_offset(12, -80) + movl %eax, %r12d + movq %r13, 8(%rsp) + cfi_offset(13, -88) + movl %edx, %r13d + movq %r14, (%rsp) + cfi_offset(14, -96) + # 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 64(%rsp), %xmm2 + +/* Go to exit */ + jmp L(EXIT) + cfi_offset(12, -80) + cfi_offset(13, -88) + cfi_offset(14, -96) + # LOE rbx rbp r12 r13 r14 r15 xmm2 + +/* Scalar math fucntion call + * to process special input + */ + +L(SCALAR_MATH_CALL): + movl %r12d, %r14d + movss 32(%rsp,%r14,4), %xmm0 + movss 48(%rsp,%r14,4), %xmm1 + call hypotf@PLT + # LOE rbx rbp r14 r15 r12d r13d xmm0 + + movss %xmm0, 64(%rsp,%r14,4) + +/* Process special inputs in loop */ + jmp L(SPECIAL_VALUES_LOOP) + # LOE rbx rbp r15 r12d r13d +END(_ZGVbN4vv_hypotf_sse4) + + .section .rodata, "a" + .align 16 + +#ifdef __svml_shypot_data_internal_typedef +typedef unsigned int VUINT32; +typedef struct +{ + __declspec(align(16)) VUINT32 _sHiLoMask[4][1]; + __declspec(align(16)) VUINT32 _sAbsMask[4][1]; + __declspec(align(16)) VUINT32 _sHalf[4][1]; + __declspec(align(16)) VUINT32 _LowBoundary[4][1]; + __declspec(align(16)) VUINT32 _HighBoundary[4][1]; +} __svml_shypot_data_internal; +#endif +__svml_shypot_data_internal: + /* legacy algorithm */ + .long 0xFFF80000, 0xFFF80000, 0xFFF80000, 0xFFF80000 /* _sHiLoMask */ + .align 16 + .long 0x7fffffff, 0x7fffffff, 0x7fffffff, 0x7fffffff /* _sAbsMask */ + .align 16 + .long 0x3f000000, 0x3f000000, 0x3f000000, 0x3f000000 /* _sHalf */ + .align 16 + .long 0x1E300000, 0x1E300000, 0x1E300000, 0x1E300000 /* _LowBoundary */ + .align 16 + .long 0x60A00000, 0x60A00000, 0x60A00000, 0x60A00000 /* _HighBoundary */ + .align 16 + .type __svml_shypot_data_internal,@object + .size __svml_shypot_data_internal,.-__svml_shypot_data_internal diff --git a/sysdeps/x86_64/fpu/multiarch/svml_s_hypotf8_core-sse.S b/sysdeps/x86_64/fpu/multiarch/svml_s_hypotf8_core-sse.S new file mode 100644 index 0000000..d37556e --- /dev/null +++ b/sysdeps/x86_64/fpu/multiarch/svml_s_hypotf8_core-sse.S @@ -0,0 +1,20 @@ +/* SSE version of vectorized hypotf. + 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 _ZGVdN8vv_hypotf _ZGVdN8vv_hypotf_sse_wrapper +#include "../svml_s_hypotf8_core.S" diff --git a/sysdeps/x86_64/fpu/multiarch/svml_s_hypotf8_core.c b/sysdeps/x86_64/fpu/multiarch/svml_s_hypotf8_core.c new file mode 100644 index 0000000..6cc497e --- /dev/null +++ b/sysdeps/x86_64/fpu/multiarch/svml_s_hypotf8_core.c @@ -0,0 +1,28 @@ +/* Multiple versions of vectorized sinf, 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 _ZGVdN8vv_hypotf +#include "ifunc-mathvec-avx2.h" + +libc_ifunc_redirected (REDIRECT_NAME, SYMBOL_NAME, IFUNC_SELECTOR ()); + +#ifdef SHARED +__hidden_ver1 (_ZGVdN8vv_hypotf, __GI__ZGVdN8vv_hypotf, + __redirect__ZGVdN8vv_hypotf) + __attribute__ ((visibility ("hidden"))); +#endif diff --git a/sysdeps/x86_64/fpu/multiarch/svml_s_hypotf8_core_avx2.S b/sysdeps/x86_64/fpu/multiarch/svml_s_hypotf8_core_avx2.S new file mode 100644 index 0000000..733022f --- /dev/null +++ b/sysdeps/x86_64/fpu/multiarch/svml_s_hypotf8_core_avx2.S @@ -0,0 +1,269 @@ +/* Function hypotf 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: + * + * HIGH LEVEL OVERVIEW + * + * Calculate z = (x*x+y*y) + * Calculate reciplicle sqrt (z) + * Calculate make two NR iterations + * + * ALGORITHM DETAILS + * + * Multiprecision branch for _HA_ only + * Remove sigm from both arguments + * Find maximum (_x) and minimum (_y) (by abs value) between arguments + * Split _x int _a and _b for multiprecision + * If _x >> _y we will we will not split _y for multiprecision + * all _y will be put into lower part (_d) and higher part (_c = 0) + * Fixing _hilo_mask for the case _x >> _y + * Split _y into _c and _d for multiprecision with fixed mask + * + * compute Hi and Lo parts of _z = _x*_x + _y*_y + * + * _zHi = _a*_a + _c*_c + * _zLo = (_x + _a)*_b + _d*_y + _d*_c + * _z = _zHi + _zLo + * + * No multiprecision branch for _LA_ and _EP_ + * _z = _VARG1 * _VARG1 + _VARG2 * _VARG2 + * + * Check _z exponent to be withing borders [1E3 ; 60A] else goto Callout + * + * Compute resciplicle sqrt s0 ~ 1.0/sqrt(_z), + * that multiplied by _z, is final result for _EP_ version. + * + * First iteration (or zero iteration): + * s = z * s0 + * h = .5 * s0 + * d = s * h - .5 + * + * Second iteration: + * h = d * h + h + * s = s * d + s + * d = s * s - z (in multiprecision for _HA_) + * + * result = s - h * d + * + * EP version of the function can be implemented as y[i]=sqrt(a[i]^2+b[i]^2) + * with all intermediate operations done in target precision for i=1,..,n. + * It can return result y[i]=0 in case a[i]^2 and b[i]^2 underflow in target + * precision (for some i). It can return result y[i]=NAN in case + * a[i]^2+b[i]^2 overflow in target precision, for some i. It can return + * result y[i]=NAN in case a[i] or b[i] is infinite, for some i. + * + * + */ + +/* Offsets for data table __svml_shypot_data_internal + */ +#define _sHiLoMask 0 +#define _sAbsMask 32 +#define _sHalf 64 +#define _LowBoundary 96 +#define _HighBoundary 128 + +#include <sysdep.h> + + .text + .section .text.avx2,"ax",@progbits +ENTRY(_ZGVdN8vv_hypotf_avx2) + pushq %rbp + cfi_def_cfa_offset(16) + movq %rsp, %rbp + cfi_def_cfa(6, 16) + cfi_offset(6, -16) + andq $-32, %rsp + subq $128, %rsp + +/* + * Implementation + * Multiprecision branch for _HA_ only + * No multiprecision branch for _LA_ + * _z = _VARG1 * _VARG1 + _VARG2 * _VARG2 + */ + vmulps %ymm0, %ymm0, %ymm8 + +/* + * Variables + * Defines + * Constants loading + */ + vmovups _sHalf+__svml_shypot_data_internal(%rip), %ymm7 + +/* Check _z exponent to be withing borders [1E3 ; 60A] else goto Callout */ + vmovups _LowBoundary+__svml_shypot_data_internal(%rip), %ymm2 + vfmadd231ps %ymm1, %ymm1, %ymm8 + +/* _s0 ~ 1.0/sqrt(_z) */ + vrsqrtps %ymm8, %ymm6 + vpcmpgtd %ymm8, %ymm2, %ymm3 + +/* First iteration */ + vmulps %ymm8, %ymm6, %ymm9 + vmulps %ymm7, %ymm6, %ymm2 + vfnmadd231ps %ymm9, %ymm2, %ymm7 + vfmadd213ps %ymm9, %ymm7, %ymm9 + +/* Second iteration */ + vfmadd132ps %ymm7, %ymm2, %ymm2 + vpcmpgtd _HighBoundary+__svml_shypot_data_internal(%rip), %ymm8, %ymm4 + vpor %ymm4, %ymm3, %ymm5 + +/* Finish second iteration in native precision for _LA_ */ + vfmsub231ps %ymm9, %ymm9, %ymm8 + vmovmskps %ymm5, %edx + vfnmadd213ps %ymm9, %ymm8, %ymm2 + +/* The end of implementation */ + testl %edx, %edx + +/* Go to special inputs processing branch */ + jne L(SPECIAL_VALUES_BRANCH) + # LOE rbx r12 r13 r14 r15 edx ymm0 ymm1 ymm2 + +/* Restore registers + * and exit the function + */ + +L(EXIT): + vmovaps %ymm2, %ymm0 + 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 %ymm0, 32(%rsp) + vmovups %ymm1, 64(%rsp) + vmovups %ymm2, 96(%rsp) + # LOE rbx r12 r13 r14 r15 edx ymm2 + + 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: -112; DW_OP_plus) */ + .cfi_escape 0x10, 0x0c, 0x0e, 0x38, 0x1c, 0x0d, 0xe0, 0xff, 0xff, 0xff, 0x1a, 0x0d, 0x90, 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: -120; DW_OP_plus) */ + .cfi_escape 0x10, 0x0d, 0x0e, 0x38, 0x1c, 0x0d, 0xe0, 0xff, 0xff, 0xff, 0x1a, 0x0d, 0x88, 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: -128; DW_OP_plus) */ + .cfi_escape 0x10, 0x0e, 0x0e, 0x38, 0x1c, 0x0d, 0xe0, 0xff, 0xff, 0xff, 0x1a, 0x0d, 0x80, 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 96(%rsp), %ymm2 + +/* 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: -112; DW_OP_plus) */ + .cfi_escape 0x10, 0x0c, 0x0e, 0x38, 0x1c, 0x0d, 0xe0, 0xff, 0xff, 0xff, 0x1a, 0x0d, 0x90, 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: -120; DW_OP_plus) */ + .cfi_escape 0x10, 0x0d, 0x0e, 0x38, 0x1c, 0x0d, 0xe0, 0xff, 0xff, 0xff, 0x1a, 0x0d, 0x88, 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: -128; DW_OP_plus) */ + .cfi_escape 0x10, 0x0e, 0x0e, 0x38, 0x1c, 0x0d, 0xe0, 0xff, 0xff, 0xff, 0x1a, 0x0d, 0x80, 0xff, 0xff, 0xff, 0x22 + # LOE rbx r12 r13 r14 r15 ymm2 + +/* Scalar math fucntion call + * to process special input + */ + +L(SCALAR_MATH_CALL): + movl %r12d, %r14d + movss 32(%rsp,%r14,4), %xmm0 + movss 64(%rsp,%r14,4), %xmm1 + call hypotf@PLT + # LOE rbx r14 r15 r12d r13d xmm0 + + movss %xmm0, 96(%rsp,%r14,4) + +/* Process special inputs in loop */ + jmp L(SPECIAL_VALUES_LOOP) + # LOE rbx r15 r12d r13d +END(_ZGVdN8vv_hypotf_avx2) + + .section .rodata, "a" + .align 32 + +#ifdef __svml_shypot_data_internal_typedef +typedef unsigned int VUINT32; +typedef struct +{ + __declspec(align(32)) VUINT32 _sHiLoMask[8][1]; + __declspec(align(32)) VUINT32 _sAbsMask[8][1]; + __declspec(align(32)) VUINT32 _sHalf[8][1]; + __declspec(align(32)) VUINT32 _LowBoundary[8][1]; + __declspec(align(32)) VUINT32 _HighBoundary[8][1]; +} __svml_shypot_data_internal; +#endif +__svml_shypot_data_internal: + /* legacy algorithm */ + .long 0xFFF80000, 0xFFF80000, 0xFFF80000, 0xFFF80000, 0xFFF80000, 0xFFF80000, 0xFFF80000, 0xFFF80000 /* _sHiLoMask */ + .align 32 + .long 0x7fffffff, 0x7fffffff, 0x7fffffff, 0x7fffffff, 0x7fffffff, 0x7fffffff, 0x7fffffff, 0x7fffffff /* _sAbsMask */ + .align 32 + .long 0x3f000000, 0x3f000000, 0x3f000000, 0x3f000000, 0x3f000000, 0x3f000000, 0x3f000000, 0x3f000000 /* _sHalf */ + .align 32 + .long 0x1E300000, 0x1E300000, 0x1E300000, 0x1E300000, 0x1E300000, 0x1E300000, 0x1E300000, 0x1E300000 /* _LowBoundary */ + .align 32 + .long 0x60A00000, 0x60A00000, 0x60A00000, 0x60A00000, 0x60A00000, 0x60A00000, 0x60A00000, 0x60A00000 /* _HighBoundary */ + .align 32 + .type __svml_shypot_data_internal,@object + .size __svml_shypot_data_internal,.-__svml_shypot_data_internal diff --git a/sysdeps/x86_64/fpu/svml_d_hypot2_core.S b/sysdeps/x86_64/fpu/svml_d_hypot2_core.S new file mode 100644 index 0000000..ea98f36 --- /dev/null +++ b/sysdeps/x86_64/fpu/svml_d_hypot2_core.S @@ -0,0 +1,29 @@ +/* Function hypot 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 (_ZGVbN2vv_hypot) +WRAPPER_IMPL_SSE2_ff hypot +END (_ZGVbN2vv_hypot) + +#ifndef USE_MULTIARCH + libmvec_hidden_def (_ZGVbN2vv_hypot) +#endif diff --git a/sysdeps/x86_64/fpu/svml_d_hypot4_core.S b/sysdeps/x86_64/fpu/svml_d_hypot4_core.S new file mode 100644 index 0000000..cedbbff --- /dev/null +++ b/sysdeps/x86_64/fpu/svml_d_hypot4_core.S @@ -0,0 +1,29 @@ +/* Function hypot 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 (_ZGVdN4vv_hypot) +WRAPPER_IMPL_AVX_ff _ZGVbN2vv_hypot +END (_ZGVdN4vv_hypot) + +#ifndef USE_MULTIARCH + libmvec_hidden_def (_ZGVdN4vv_hypot) +#endif diff --git a/sysdeps/x86_64/fpu/svml_d_hypot4_core_avx.S b/sysdeps/x86_64/fpu/svml_d_hypot4_core_avx.S new file mode 100644 index 0000000..e0fef52 --- /dev/null +++ b/sysdeps/x86_64/fpu/svml_d_hypot4_core_avx.S @@ -0,0 +1,25 @@ +/* Function hypot 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 (_ZGVcN4vv_hypot) +WRAPPER_IMPL_AVX_ff _ZGVbN2vv_hypot +END (_ZGVcN4vv_hypot) diff --git a/sysdeps/x86_64/fpu/svml_d_hypot8_core.S b/sysdeps/x86_64/fpu/svml_d_hypot8_core.S new file mode 100644 index 0000000..7588e44 --- /dev/null +++ b/sysdeps/x86_64/fpu/svml_d_hypot8_core.S @@ -0,0 +1,25 @@ +/* Function hypot 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_d_wrapper_impl.h" + + .text +ENTRY (_ZGVeN8vv_hypot) +WRAPPER_IMPL_AVX512_ff _ZGVdN4vv_hypot +END (_ZGVeN8vv_hypot) diff --git a/sysdeps/x86_64/fpu/svml_s_hypotf16_core.S b/sysdeps/x86_64/fpu/svml_s_hypotf16_core.S new file mode 100644 index 0000000..06d421a --- /dev/null +++ b/sysdeps/x86_64/fpu/svml_s_hypotf16_core.S @@ -0,0 +1,25 @@ +/* Function hypotf 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 (_ZGVeN16vv_hypotf) +WRAPPER_IMPL_AVX512_ff _ZGVdN8vv_hypotf +END (_ZGVeN16vv_hypotf) diff --git a/sysdeps/x86_64/fpu/svml_s_hypotf4_core.S b/sysdeps/x86_64/fpu/svml_s_hypotf4_core.S new file mode 100644 index 0000000..7e8553c --- /dev/null +++ b/sysdeps/x86_64/fpu/svml_s_hypotf4_core.S @@ -0,0 +1,29 @@ +/* Function hypotf 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_s_wrapper_impl.h" + + .text +ENTRY (_ZGVbN4vv_hypotf) +WRAPPER_IMPL_SSE2_ff hypotf +END (_ZGVbN4vv_hypotf) + +#ifndef USE_MULTIARCH + libmvec_hidden_def (_ZGVbN4vv_hypotf) +#endif diff --git a/sysdeps/x86_64/fpu/svml_s_hypotf8_core.S b/sysdeps/x86_64/fpu/svml_s_hypotf8_core.S new file mode 100644 index 0000000..a9bf273 --- /dev/null +++ b/sysdeps/x86_64/fpu/svml_s_hypotf8_core.S @@ -0,0 +1,29 @@ +/* Function hypotf 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 (_ZGVdN8vv_hypotf) +WRAPPER_IMPL_AVX_ff _ZGVbN4vv_hypotf +END (_ZGVdN8vv_hypotf) + +#ifndef USE_MULTIARCH + libmvec_hidden_def (_ZGVdN8vv_hypotf) +#endif diff --git a/sysdeps/x86_64/fpu/svml_s_hypotf8_core_avx.S b/sysdeps/x86_64/fpu/svml_s_hypotf8_core_avx.S new file mode 100644 index 0000000..8b8008a --- /dev/null +++ b/sysdeps/x86_64/fpu/svml_s_hypotf8_core_avx.S @@ -0,0 +1,25 @@ +/* Function hypotf 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(_ZGVcN8vv_hypotf) +WRAPPER_IMPL_AVX_ff _ZGVbN4vv_hypotf +END(_ZGVcN8vv_hypotf) diff --git a/sysdeps/x86_64/fpu/test-double-libmvec-hypot-avx.c b/sysdeps/x86_64/fpu/test-double-libmvec-hypot-avx.c new file mode 100644 index 0000000..c6a26a6 --- /dev/null +++ b/sysdeps/x86_64/fpu/test-double-libmvec-hypot-avx.c @@ -0,0 +1 @@ +#include "test-double-libmvec-hypot.c" diff --git a/sysdeps/x86_64/fpu/test-double-libmvec-hypot-avx2.c b/sysdeps/x86_64/fpu/test-double-libmvec-hypot-avx2.c new file mode 100644 index 0000000..c6a26a6 --- /dev/null +++ b/sysdeps/x86_64/fpu/test-double-libmvec-hypot-avx2.c @@ -0,0 +1 @@ +#include "test-double-libmvec-hypot.c" diff --git a/sysdeps/x86_64/fpu/test-double-libmvec-hypot-avx512f.c b/sysdeps/x86_64/fpu/test-double-libmvec-hypot-avx512f.c new file mode 100644 index 0000000..c6a26a6 --- /dev/null +++ b/sysdeps/x86_64/fpu/test-double-libmvec-hypot-avx512f.c @@ -0,0 +1 @@ +#include "test-double-libmvec-hypot.c" diff --git a/sysdeps/x86_64/fpu/test-double-libmvec-hypot.c b/sysdeps/x86_64/fpu/test-double-libmvec-hypot.c new file mode 100644 index 0000000..c0f600a --- /dev/null +++ b/sysdeps/x86_64/fpu/test-double-libmvec-hypot.c @@ -0,0 +1,3 @@ +#define LIBMVEC_TYPE double +#define LIBMVEC_FUNC hypot +#include "test-vector-abi-arg2.h" diff --git a/sysdeps/x86_64/fpu/test-double-vlen2-wrappers.c b/sysdeps/x86_64/fpu/test-double-vlen2-wrappers.c index 5746bb5..9bc9d1d 100644 --- a/sysdeps/x86_64/fpu/test-double-vlen2-wrappers.c +++ b/sysdeps/x86_64/fpu/test-double-vlen2-wrappers.c @@ -30,6 +30,7 @@ VECTOR_WRAPPER_ff (WRAPPER_NAME (pow), _ZGVbN2vv_pow) VECTOR_WRAPPER (WRAPPER_NAME (acos), _ZGVbN2v_acos) VECTOR_WRAPPER (WRAPPER_NAME (atan), _ZGVbN2v_atan) VECTOR_WRAPPER (WRAPPER_NAME (asin), _ZGVbN2v_asin) +VECTOR_WRAPPER_ff (WRAPPER_NAME (hypot), _ZGVbN2vv_hypot) #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 8d3d549..c41994d 100644 --- a/sysdeps/x86_64/fpu/test-double-vlen4-avx2-wrappers.c +++ b/sysdeps/x86_64/fpu/test-double-vlen4-avx2-wrappers.c @@ -33,6 +33,7 @@ VECTOR_WRAPPER_ff (WRAPPER_NAME (pow), _ZGVdN4vv_pow) VECTOR_WRAPPER (WRAPPER_NAME (acos), _ZGVdN4v_acos) VECTOR_WRAPPER (WRAPPER_NAME (atan), _ZGVdN4v_atan) VECTOR_WRAPPER (WRAPPER_NAME (asin), _ZGVdN4v_asin) +VECTOR_WRAPPER_ff (WRAPPER_NAME (hypot), _ZGVdN4vv_hypot) #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 f43328f..881f6c8 100644 --- a/sysdeps/x86_64/fpu/test-double-vlen4-wrappers.c +++ b/sysdeps/x86_64/fpu/test-double-vlen4-wrappers.c @@ -30,6 +30,7 @@ VECTOR_WRAPPER_ff (WRAPPER_NAME (pow), _ZGVcN4vv_pow) VECTOR_WRAPPER (WRAPPER_NAME (acos), _ZGVcN4v_acos) VECTOR_WRAPPER (WRAPPER_NAME (atan), _ZGVcN4v_atan) VECTOR_WRAPPER (WRAPPER_NAME (asin), _ZGVcN4v_asin) +VECTOR_WRAPPER_ff (WRAPPER_NAME (hypot), _ZGVcN4vv_hypot) #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 8b566c1..6fd106f 100644 --- a/sysdeps/x86_64/fpu/test-double-vlen8-wrappers.c +++ b/sysdeps/x86_64/fpu/test-double-vlen8-wrappers.c @@ -30,6 +30,7 @@ VECTOR_WRAPPER_ff (WRAPPER_NAME (pow), _ZGVeN8vv_pow) VECTOR_WRAPPER (WRAPPER_NAME (acos), _ZGVeN8v_acos) VECTOR_WRAPPER (WRAPPER_NAME (atan), _ZGVeN8v_atan) VECTOR_WRAPPER (WRAPPER_NAME (asin), _ZGVeN8v_asin) +VECTOR_WRAPPER_ff (WRAPPER_NAME (hypot), _ZGVeN8vv_hypot) #ifndef __ILP32__ # define VEC_INT_TYPE __m512i diff --git a/sysdeps/x86_64/fpu/test-float-libmvec-hypotf-avx.c b/sysdeps/x86_64/fpu/test-float-libmvec-hypotf-avx.c new file mode 100644 index 0000000..97d11ad --- /dev/null +++ b/sysdeps/x86_64/fpu/test-float-libmvec-hypotf-avx.c @@ -0,0 +1 @@ +#include "test-float-libmvec-hypotf.c" diff --git a/sysdeps/x86_64/fpu/test-float-libmvec-hypotf-avx2.c b/sysdeps/x86_64/fpu/test-float-libmvec-hypotf-avx2.c new file mode 100644 index 0000000..97d11ad --- /dev/null +++ b/sysdeps/x86_64/fpu/test-float-libmvec-hypotf-avx2.c @@ -0,0 +1 @@ +#include "test-float-libmvec-hypotf.c" diff --git a/sysdeps/x86_64/fpu/test-float-libmvec-hypotf-avx512f.c b/sysdeps/x86_64/fpu/test-float-libmvec-hypotf-avx512f.c new file mode 100644 index 0000000..97d11ad --- /dev/null +++ b/sysdeps/x86_64/fpu/test-float-libmvec-hypotf-avx512f.c @@ -0,0 +1 @@ +#include "test-float-libmvec-hypotf.c" diff --git a/sysdeps/x86_64/fpu/test-float-libmvec-hypotf.c b/sysdeps/x86_64/fpu/test-float-libmvec-hypotf.c new file mode 100644 index 0000000..38776fa --- /dev/null +++ b/sysdeps/x86_64/fpu/test-float-libmvec-hypotf.c @@ -0,0 +1,3 @@ +#define LIBMVEC_TYPE float +#define LIBMVEC_FUNC hypotf +#include "test-vector-abi-arg2.h" diff --git a/sysdeps/x86_64/fpu/test-float-vlen16-wrappers.c b/sysdeps/x86_64/fpu/test-float-vlen16-wrappers.c index 3d3218a..4c2ea6d 100644 --- a/sysdeps/x86_64/fpu/test-float-vlen16-wrappers.c +++ b/sysdeps/x86_64/fpu/test-float-vlen16-wrappers.c @@ -30,6 +30,7 @@ VECTOR_WRAPPER_ff (WRAPPER_NAME (powf), _ZGVeN16vv_powf) VECTOR_WRAPPER (WRAPPER_NAME (acosf), _ZGVeN16v_acosf) VECTOR_WRAPPER (WRAPPER_NAME (atanf), _ZGVeN16v_atanf) VECTOR_WRAPPER (WRAPPER_NAME (asinf), _ZGVeN16v_asinf) +VECTOR_WRAPPER_ff (WRAPPER_NAME (hypotf), _ZGVeN16vv_hypotf) #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 7d75b9f..1d5d952 100644 --- a/sysdeps/x86_64/fpu/test-float-vlen4-wrappers.c +++ b/sysdeps/x86_64/fpu/test-float-vlen4-wrappers.c @@ -30,6 +30,7 @@ VECTOR_WRAPPER_ff (WRAPPER_NAME (powf), _ZGVbN4vv_powf) VECTOR_WRAPPER (WRAPPER_NAME (acosf), _ZGVbN4v_acosf) VECTOR_WRAPPER (WRAPPER_NAME (atanf), _ZGVbN4v_atanf) VECTOR_WRAPPER (WRAPPER_NAME (asinf), _ZGVbN4v_asinf) +VECTOR_WRAPPER_ff (WRAPPER_NAME (hypotf), _ZGVbN4vv_hypotf) #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 405dde4..7a750f3 100644 --- a/sysdeps/x86_64/fpu/test-float-vlen8-avx2-wrappers.c +++ b/sysdeps/x86_64/fpu/test-float-vlen8-avx2-wrappers.c @@ -33,6 +33,7 @@ VECTOR_WRAPPER_ff (WRAPPER_NAME (powf), _ZGVdN8vv_powf) VECTOR_WRAPPER (WRAPPER_NAME (acosf), _ZGVdN8v_acosf) VECTOR_WRAPPER (WRAPPER_NAME (atanf), _ZGVdN8v_atanf) VECTOR_WRAPPER (WRAPPER_NAME (asinf), _ZGVdN8v_asinf) +VECTOR_WRAPPER_ff (WRAPPER_NAME (hypotf), _ZGVdN8vv_hypotf) /* 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 7558443..af816a7 100644 --- a/sysdeps/x86_64/fpu/test-float-vlen8-wrappers.c +++ b/sysdeps/x86_64/fpu/test-float-vlen8-wrappers.c @@ -30,6 +30,7 @@ VECTOR_WRAPPER_ff (WRAPPER_NAME (powf), _ZGVcN8vv_powf) VECTOR_WRAPPER (WRAPPER_NAME (acosf), _ZGVcN8v_acosf) VECTOR_WRAPPER (WRAPPER_NAME (atanf), _ZGVcN8v_atanf) VECTOR_WRAPPER (WRAPPER_NAME (asinf), _ZGVcN8v_asinf) +VECTOR_WRAPPER_ff (WRAPPER_NAME (hypotf), _ZGVcN8vv_hypotf) #define VEC_INT_TYPE __m128i |