diff options
Diffstat (limited to 'sysdeps/x86_64/fpu/multiarch/svml_d_acos4_core_avx2.S')
-rw-r--r-- | sysdeps/x86_64/fpu/multiarch/svml_d_acos4_core_avx2.S | 285 |
1 files changed, 285 insertions, 0 deletions
diff --git a/sysdeps/x86_64/fpu/multiarch/svml_d_acos4_core_avx2.S b/sysdeps/x86_64/fpu/multiarch/svml_d_acos4_core_avx2.S new file mode 100644 index 0000000000..bf85bdbd37 --- /dev/null +++ b/sysdeps/x86_64/fpu/multiarch/svml_d_acos4_core_avx2.S @@ -0,0 +1,285 @@ +/* Function acos 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: + * + * SelMask = (|x| >= 0.5) ? 1 : 0; + * R = SelMask ? sqrt(0.5 - 0.5*|x|) : |x| + * acos(|x|) = SelMask ? 2*Poly(R) : (Pi/2 - Poly(R)) + * acos(x) = sign(x) ? (Pi - acos(|x|)) : acos(|x|) + * + */ + +/* Offsets for data table __svml_dacos_data_internal + */ +#define SgnBit 0 +#define OneHalf 32 +#define SmallNorm 64 +#define MOne 96 +#define Two 128 +#define sqrt_coeff 160 +#define poly_coeff 288 +#define PiH 672 +#define Pi2H 704 + +#include <sysdep.h> + + .text + .section .text.avx2,"ax",@progbits +ENTRY(_ZGVdN4v_acos_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 + vmovupd __svml_dacos_data_internal(%rip), %ymm6 + vmovupd OneHalf+__svml_dacos_data_internal(%rip), %ymm7 + vmovapd %ymm0, %ymm5 + +/* x = -|arg| */ + vorpd %ymm5, %ymm6, %ymm4 + +/* Y = 0.5 + 0.5*(-x) */ + vfmadd231pd %ymm4, %ymm7, %ymm7 + +/* x^2 */ + vmulpd %ymm4, %ymm4, %ymm8 + +/* S ~ 2*sqrt(Y) */ + vmovupd sqrt_coeff+__svml_dacos_data_internal(%rip), %ymm0 + vcmplt_oqpd SmallNorm+__svml_dacos_data_internal(%rip), %ymm7, %ymm12 + vminpd %ymm7, %ymm8, %ymm2 + +/* NaN processed in special branch (so wind test passed) */ + vcmpnge_uqpd MOne+__svml_dacos_data_internal(%rip), %ymm4, %ymm9 + vcvtpd2ps %ymm7, %xmm10 + vmovupd poly_coeff+64+__svml_dacos_data_internal(%rip), %ymm8 + vcmpnlt_uqpd %ymm7, %ymm2, %ymm1 + vrsqrtps %xmm10, %xmm11 + vfmadd213pd poly_coeff+96+__svml_dacos_data_internal(%rip), %ymm2, %ymm8 + vcvtps2pd %xmm11, %ymm13 + vmovupd poly_coeff+128+__svml_dacos_data_internal(%rip), %ymm11 + vandnpd %ymm13, %ymm12, %ymm14 + vmulpd %ymm14, %ymm14, %ymm15 + vfmadd213pd poly_coeff+160+__svml_dacos_data_internal(%rip), %ymm2, %ymm11 + vmulpd %ymm2, %ymm2, %ymm13 + vmovupd poly_coeff+256+__svml_dacos_data_internal(%rip), %ymm12 + vmulpd %ymm13, %ymm13, %ymm10 + vfmadd213pd poly_coeff+288+__svml_dacos_data_internal(%rip), %ymm2, %ymm12 + vandpd %ymm5, %ymm6, %ymm3 + vaddpd %ymm7, %ymm7, %ymm6 + vmulpd %ymm6, %ymm14, %ymm7 + vfmsub213pd Two+__svml_dacos_data_internal(%rip), %ymm15, %ymm6 + vmovupd poly_coeff+320+__svml_dacos_data_internal(%rip), %ymm14 + vfmadd213pd sqrt_coeff+32+__svml_dacos_data_internal(%rip), %ymm6, %ymm0 + vmulpd %ymm6, %ymm7, %ymm15 + vfmadd213pd poly_coeff+352+__svml_dacos_data_internal(%rip), %ymm2, %ymm14 + vfmadd213pd sqrt_coeff+64+__svml_dacos_data_internal(%rip), %ymm6, %ymm0 + vfmadd213pd sqrt_coeff+96+__svml_dacos_data_internal(%rip), %ymm6, %ymm0 + +/* polynomial */ + vmovupd poly_coeff+__svml_dacos_data_internal(%rip), %ymm6 + vfnmadd213pd %ymm7, %ymm15, %ymm0 + vfmadd213pd poly_coeff+32+__svml_dacos_data_internal(%rip), %ymm2, %ymm6 + vblendvpd %ymm1, %ymm0, %ymm4, %ymm0 + vfmadd213pd %ymm8, %ymm13, %ymm6 + vmovmskpd %ymm9, %edx + vmovupd poly_coeff+192+__svml_dacos_data_internal(%rip), %ymm9 + vfmadd213pd poly_coeff+224+__svml_dacos_data_internal(%rip), %ymm2, %ymm9 + vfmadd213pd %ymm9, %ymm13, %ymm11 + vfmadd213pd %ymm11, %ymm10, %ymm6 + vfmadd213pd %ymm12, %ymm13, %ymm6 + vfmadd213pd %ymm14, %ymm13, %ymm6 + vmulpd %ymm6, %ymm2, %ymm9 + +/* X<X^2 iff X<0 */ + vcmplt_oqpd %ymm2, %ymm5, %ymm6 + vandpd PiH+__svml_dacos_data_internal(%rip), %ymm1, %ymm2 + vandnpd Pi2H+__svml_dacos_data_internal(%rip), %ymm1, %ymm7 + vxorpd %ymm3, %ymm0, %ymm1 + vfmadd213pd %ymm1, %ymm1, %ymm9 + vandpd %ymm6, %ymm2, %ymm2 + vaddpd %ymm7, %ymm2, %ymm8 + vaddpd %ymm9, %ymm8, %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 acos@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_acos_avx2) + + .section .rodata, "a" + .align 32 + +#ifdef __svml_dacos_data_internal_typedef +typedef unsigned int VUINT32; +typedef struct { + __declspec(align(32)) VUINT32 SgnBit[4][2]; + __declspec(align(32)) VUINT32 OneHalf[4][2]; + __declspec(align(32)) VUINT32 SmallNorm[4][2]; + __declspec(align(32)) VUINT32 MOne[4][2]; + __declspec(align(32)) VUINT32 Two[4][2]; + __declspec(align(32)) VUINT32 sqrt_coeff[4][4][2]; + __declspec(align(32)) VUINT32 poly_coeff[12][4][2]; + __declspec(align(32)) VUINT32 PiH[4][2]; + __declspec(align(32)) VUINT32 Pi2H[4][2]; +} __svml_dacos_data_internal; +#endif +__svml_dacos_data_internal: + /*== SgnBit ==*/ + .quad 0x8000000000000000, 0x8000000000000000, 0x8000000000000000, 0x8000000000000000 + /*== OneHalf ==*/ + .align 32 + .quad 0x3fe0000000000000, 0x3fe0000000000000, 0x3fe0000000000000, 0x3fe0000000000000 + /*== SmallNorm ==*/ + .align 32 + .quad 0x3000000000000000, 0x3000000000000000, 0x3000000000000000, 0x3000000000000000 + /*== MOne ==*/ + .align 32 + .quad 0xbff0000000000000, 0xbff0000000000000, 0xbff0000000000000, 0xbff0000000000000 + /*== Two ==*/ + .align 32 + .quad 0x4000000000000000, 0x4000000000000000, 0x4000000000000000, 0x4000000000000000 + /*== sqrt_coeff[4] ==*/ + .align 32 + .quad 0xbf918000993B24C3, 0xbf918000993B24C3, 0xbf918000993B24C3, 0xbf918000993B24C3 /* sqrt_coeff4 */ + .quad 0x3fa400006F70D42D, 0x3fa400006F70D42D, 0x3fa400006F70D42D, 0x3fa400006F70D42D /* sqrt_coeff3 */ + .quad 0xbfb7FFFFFFFFFE97, 0xbfb7FFFFFFFFFE97, 0xbfb7FFFFFFFFFE97, 0xbfb7FFFFFFFFFE97 /* sqrt_coeff2 */ + .quad 0x3fcFFFFFFFFFFF9D, 0x3fcFFFFFFFFFFF9D, 0x3fcFFFFFFFFFFF9D, 0x3fcFFFFFFFFFFF9D /* sqrt_coeff1 */ + /*== poly_coeff[12] ==*/ + .align 32 + .quad 0x3fa07520C70EB909, 0x3fa07520C70EB909, 0x3fa07520C70EB909, 0x3fa07520C70EB909 /* poly_coeff12 */ + .quad 0xbf90FB17F7DBB0ED, 0xbf90FB17F7DBB0ED, 0xbf90FB17F7DBB0ED, 0xbf90FB17F7DBB0ED /* poly_coeff11 */ + .quad 0x3f943F44BFBC3BAE, 0x3f943F44BFBC3BAE, 0x3f943F44BFBC3BAE, 0x3f943F44BFBC3BAE /* poly_coeff10 */ + .quad 0x3f7A583395D45ED5, 0x3f7A583395D45ED5, 0x3f7A583395D45ED5, 0x3f7A583395D45ED5 /* poly_coeff9 */ + .quad 0x3f88F8DC2AFCCAD6, 0x3f88F8DC2AFCCAD6, 0x3f88F8DC2AFCCAD6, 0x3f88F8DC2AFCCAD6 /* poly_coeff8 */ + .quad 0x3f8C6DBBCB88BD57, 0x3f8C6DBBCB88BD57, 0x3f8C6DBBCB88BD57, 0x3f8C6DBBCB88BD57 /* poly_coeff7 */ + .quad 0x3f91C6DCF538AD2E, 0x3f91C6DCF538AD2E, 0x3f91C6DCF538AD2E, 0x3f91C6DCF538AD2E /* poly_coeff6 */ + .quad 0x3f96E89CEBDEFadd, 0x3f96E89CEBDEFadd, 0x3f96E89CEBDEFadd, 0x3f96E89CEBDEFadd /* poly_coeff5 */ + .quad 0x3f9F1C72E13AD8BE, 0x3f9F1C72E13AD8BE, 0x3f9F1C72E13AD8BE, 0x3f9F1C72E13AD8BE /* poly_coeff4 */ + .quad 0x3fa6DB6DB3B445F8, 0x3fa6DB6DB3B445F8, 0x3fa6DB6DB3B445F8, 0x3fa6DB6DB3B445F8 /* poly_coeff3 */ + .quad 0x3fb333333337E0DE, 0x3fb333333337E0DE, 0x3fb333333337E0DE, 0x3fb333333337E0DE /* poly_coeff2 */ + .quad 0x3fc555555555529C, 0x3fc555555555529C, 0x3fc555555555529C, 0x3fc555555555529C /* poly_coeff1 */ + /*== PiH ==*/ + .align 32 + .quad 0x400921fb54442d18, 0x400921fb54442d18, 0x400921fb54442d18, 0x400921fb54442d18 + /*== Pi2H ==*/ + .align 32 + .quad 0x3ff921fb54442d18, 0x3ff921fb54442d18, 0x3ff921fb54442d18, 0x3ff921fb54442d18 + .align 32 + .type __svml_dacos_data_internal,@object + .size __svml_dacos_data_internal,.-__svml_dacos_data_internal |