diff options
Diffstat (limited to 'sysdeps/x86_64/fpu/multiarch/svml_s_asinf4_core_sse4.S')
| -rw-r--r-- | sysdeps/x86_64/fpu/multiarch/svml_s_asinf4_core_sse4.S | 252 |
1 files changed, 252 insertions, 0 deletions
diff --git a/sysdeps/x86_64/fpu/multiarch/svml_s_asinf4_core_sse4.S b/sysdeps/x86_64/fpu/multiarch/svml_s_asinf4_core_sse4.S new file mode 100644 index 0000000000..ddcceeb7b9 --- /dev/null +++ b/sysdeps/x86_64/fpu/multiarch/svml_s_asinf4_core_sse4.S @@ -0,0 +1,252 @@ +/* Function asinf 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: + * + * SelMask = (|x| >= 0.5) ? 1 : 0; + * R = SelMask ? sqrt(0.5 - 0.5*|x|) : |x| + * asin(x) = (SelMask ? (Pi/2 - 2*Poly(R)) : Poly(R))*(-1)^sign(x) + * + * + */ + +/* Offsets for data table __svml_sasin_data_internal + */ +#define AbsMask 0 +#define OneHalf 16 +#define SmallNorm 32 +#define One 48 +#define Two 64 +#define sqrt_coeff 80 +#define poly_coeff 112 +#define Pi2H 192 + +#include <sysdep.h> + + .text + .section .text.sse4,"ax",@progbits +ENTRY(_ZGVbN4v_asinf_sse4) + subq $72, %rsp + cfi_def_cfa_offset(80) + movaps %xmm0, %xmm2 + movups __svml_sasin_data_internal(%rip), %xmm1 + movups OneHalf+__svml_sasin_data_internal(%rip), %xmm5 + +/* x = |arg| */ + movaps %xmm1, %xmm0 + andps %xmm2, %xmm0 + +/* Y = 0.5 - 0.5*x */ + movaps %xmm5, %xmm3 + mulps %xmm0, %xmm3 + movaps %xmm5, %xmm8 + +/* x^2 */ + movaps %xmm0, %xmm14 + movaps %xmm0, %xmm15 + mulps %xmm0, %xmm14 + subps %xmm3, %xmm8 + cmpnltps %xmm5, %xmm15 + +/* SQ ~ -2*sqrt(Y) */ + rsqrtps %xmm8, %xmm6 + minps %xmm8, %xmm14 + movaps %xmm8, %xmm9 + movaps %xmm14, %xmm10 + cmpltps SmallNorm+__svml_sasin_data_internal(%rip), %xmm9 + mulps %xmm14, %xmm10 + addps %xmm8, %xmm8 + andnps %xmm6, %xmm9 + movaps %xmm15, %xmm3 + movaps %xmm9, %xmm7 + andnps %xmm0, %xmm3 + mulps %xmm9, %xmm7 + andnps %xmm2, %xmm1 + mulps %xmm8, %xmm9 + mulps %xmm7, %xmm8 + +/* polynomial */ + movups poly_coeff+__svml_sasin_data_internal(%rip), %xmm11 + mulps %xmm14, %xmm11 + subps Two+__svml_sasin_data_internal(%rip), %xmm8 + movups poly_coeff+32+__svml_sasin_data_internal(%rip), %xmm12 + mulps %xmm14, %xmm12 + addps poly_coeff+16+__svml_sasin_data_internal(%rip), %xmm11 + mulps %xmm10, %xmm11 + addps poly_coeff+48+__svml_sasin_data_internal(%rip), %xmm12 + movups sqrt_coeff+__svml_sasin_data_internal(%rip), %xmm13 + addps %xmm11, %xmm12 + mulps %xmm8, %xmm13 + mulps %xmm9, %xmm8 + mulps %xmm14, %xmm12 + addps sqrt_coeff+16+__svml_sasin_data_internal(%rip), %xmm13 + addps poly_coeff+64+__svml_sasin_data_internal(%rip), %xmm12 + mulps %xmm8, %xmm13 + mulps %xmm12, %xmm14 + subps %xmm9, %xmm13 + andps %xmm15, %xmm13 + orps %xmm13, %xmm3 + mulps %xmm3, %xmm14 + movups One+__svml_sasin_data_internal(%rip), %xmm4 + addps %xmm14, %xmm3 + cmpltps %xmm0, %xmm4 + movups Pi2H+__svml_sasin_data_internal(%rip), %xmm0 + andps %xmm15, %xmm0 + movmskps %xmm4, %edx + addps %xmm3, %xmm0 + pxor %xmm1, %xmm0 + testl %edx, %edx + +/* Go to special inputs processing branch */ + jne L(SPECIAL_VALUES_BRANCH) + # LOE rbx rbp r12 r13 r14 r15 edx xmm0 xmm2 + +/* Restore registers + * and exit the function + */ + +L(EXIT): + addq $72, %rsp + cfi_def_cfa_offset(8) + ret + cfi_def_cfa_offset(80) + +/* Branch to process + * special inputs + */ + +L(SPECIAL_VALUES_BRANCH): + movups %xmm2, 32(%rsp) + movups %xmm0, 48(%rsp) + # LOE rbx rbp r12 r13 r14 r15 edx + + xorl %eax, %eax + movq %r12, 16(%rsp) + cfi_offset(12, -64) + movl %eax, %r12d + movq %r13, 8(%rsp) + cfi_offset(13, -72) + movl %edx, %r13d + movq %r14, (%rsp) + cfi_offset(14, -80) + # LOE rbx rbp r15 r12d r13d + +/* Range mask + * bits check + */ + +L(RANGEMASK_CHECK): + btl %r12d, %r13d + +/* Call scalar math function */ + jc L(SCALAR_MATH_CALL) + # LOE rbx rbp r15 r12d r13d + +/* Special inputs + * processing loop + */ + +L(SPECIAL_VALUES_LOOP): + incl %r12d + cmpl $4, %r12d + +/* Check bits in range mask */ + jl L(RANGEMASK_CHECK) + # LOE rbx rbp r15 r12d r13d + + movq 16(%rsp), %r12 + cfi_restore(12) + movq 8(%rsp), %r13 + cfi_restore(13) + movq (%rsp), %r14 + cfi_restore(14) + movups 48(%rsp), %xmm0 + +/* Go to exit */ + jmp L(EXIT) + cfi_offset(12, -64) + cfi_offset(13, -72) + cfi_offset(14, -80) + # LOE rbx rbp r12 r13 r14 r15 xmm0 + +/* Scalar math fucntion call + * to process special input + */ + +L(SCALAR_MATH_CALL): + movl %r12d, %r14d + movss 32(%rsp,%r14,4), %xmm0 + call asinf@PLT + # LOE rbx rbp r14 r15 r12d r13d xmm0 + + movss %xmm0, 48(%rsp,%r14,4) + +/* Process special inputs in loop */ + jmp L(SPECIAL_VALUES_LOOP) + # LOE rbx rbp r15 r12d r13d +END(_ZGVbN4v_asinf_sse4) + + .section .rodata, "a" + .align 16 + +#ifdef __svml_sasin_data_internal_typedef +typedef unsigned int VUINT32; +typedef struct { + __declspec(align(16)) VUINT32 AbsMask[4][1]; + __declspec(align(16)) VUINT32 OneHalf[4][1]; + __declspec(align(16)) VUINT32 SmallNorm[4][1]; + __declspec(align(16)) VUINT32 One[4][1]; + __declspec(align(16)) VUINT32 Two[4][1]; + __declspec(align(16)) VUINT32 sqrt_coeff[2][4][1]; + __declspec(align(16)) VUINT32 poly_coeff[5][4][1]; + __declspec(align(16)) VUINT32 Pi2H[4][1]; +} __svml_sasin_data_internal; +#endif +__svml_sasin_data_internal: + /*== AbsMask ==*/ + .long 0x7fffffff, 0x7fffffff, 0x7fffffff, 0x7fffffff + /*== OneHalf ==*/ + .align 16 + .long 0x3f000000, 0x3f000000, 0x3f000000, 0x3f000000 + /*== SmallNorm ==*/ + .align 16 + .long 0x2f800000, 0x2f800000, 0x2f800000, 0x2f800000 + /*== One ==*/ + .align 16 + .long 0x3f800000, 0x3f800000, 0x3f800000, 0x3f800000 + /*== Two ==*/ + .align 16 + .long 0x40000000, 0x40000000, 0x40000000, 0x40000000 + /*== sqrt_coeff[2] ==*/ + .align 16 + .long 0xbdC00004, 0xbdC00004, 0xbdC00004, 0xbdC00004 /* sqrt_coeff2 */ + .long 0x3e800001, 0x3e800001, 0x3e800001, 0x3e800001 /* sqrt_coeff1 */ + /*== poly_coeff[5] ==*/ + .align 16 + .long 0x3d2EDC07, 0x3d2EDC07, 0x3d2EDC07, 0x3d2EDC07 /* poly_coeff5 */ + .long 0x3CC32A6B, 0x3CC32A6B, 0x3CC32A6B, 0x3CC32A6B /* poly_coeff4 */ + .long 0x3d3A9AB4, 0x3d3A9AB4, 0x3d3A9AB4, 0x3d3A9AB4 /* poly_coeff3 */ + .long 0x3d997C12, 0x3d997C12, 0x3d997C12, 0x3d997C12 /* poly_coeff2 */ + .long 0x3e2AAAFF, 0x3e2AAAFF, 0x3e2AAAFF, 0x3e2AAAFF /* poly_coeff1 */ + /*== Pi2H ==*/ + .align 16 + .long 0x3fc90FDB, 0x3fc90FDB, 0x3fc90FDB, 0x3fc90FDB + .align 16 + .type __svml_sasin_data_internal,@object + .size __svml_sasin_data_internal,.-__svml_sasin_data_internal |