diff options
Diffstat (limited to 'sysdeps/x86_64/fpu/multiarch/svml_s_atan2f4_core_sse4.S')
| -rw-r--r-- | sysdeps/x86_64/fpu/multiarch/svml_s_atan2f4_core_sse4.S | 384 |
1 files changed, 384 insertions, 0 deletions
diff --git a/sysdeps/x86_64/fpu/multiarch/svml_s_atan2f4_core_sse4.S b/sysdeps/x86_64/fpu/multiarch/svml_s_atan2f4_core_sse4.S new file mode 100644 index 0000000000..e4fbe82501 --- /dev/null +++ b/sysdeps/x86_64/fpu/multiarch/svml_s_atan2f4_core_sse4.S @@ -0,0 +1,384 @@ +/* Function atan2f 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: + * For 0.0 <= x <= 7.0/16.0: atan(x) = atan(0.0) + atan(s), where s=(x-0.0)/(1.0+0.0*x) + * For 7.0/16.0 <= x <= 11.0/16.0: atan(x) = atan(0.5) + atan(s), where s=(x-0.5)/(1.0+0.5*x) + * For 11.0/16.0 <= x <= 19.0/16.0: atan(x) = atan(1.0) + atan(s), where s=(x-1.0)/(1.0+1.0*x) + * For 19.0/16.0 <= x <= 39.0/16.0: atan(x) = atan(1.5) + atan(s), where s=(x-1.5)/(1.0+1.5*x) + * For 39.0/16.0 <= x <= inf : atan(x) = atan(inf) + atan(s), where s=-1.0/x + * Where atan(s) ~= s+s^3*Poly11(s^2) on interval |s|<7.0/0.16. + * + * + */ + +/* Offsets for data table __svml_satan2_data_internal + */ +#define sZERO 0 +#define sSIGN_MASK 16 +#define sABS_MASK 32 +#define sPIO2 48 +#define sPI 64 +#define sPC8 80 +#define sPC7 96 +#define sPC6 112 +#define sPC5 128 +#define sPC4 144 +#define sPC3 160 +#define sPC2 176 +#define sPC1 192 +#define sPC0 208 +#define iCHK_WORK_SUB 224 +#define iCHK_WORK_CMP 240 + +#include <sysdep.h> + + .text + .section .text.sse4,"ax",@progbits +ENTRY(_ZGVbN4vv_atan2f_sse4) + subq $88, %rsp + cfi_def_cfa_offset(96) + movaps %xmm0, %xmm12 + +/* + * #define NO_VECTOR_ZERO_ATAN2_ARGS + * Declarations + * Variables + * Constants + * The end of declarations + * Implementation + * Arguments signs + */ + movups sABS_MASK+__svml_satan2_data_internal(%rip), %xmm10 + movaps %xmm1, %xmm13 + movaps %xmm10, %xmm11 + andps %xmm12, %xmm10 + andps %xmm13, %xmm11 + movaps %xmm10, %xmm7 + cmpltps %xmm11, %xmm7 + +/* + * 1) If y<x then a= y, b=x, PIO2=0 + * 2) If y>x then a=-x, b=y, PIO2=Pi/2 + */ + movups sSIGN_MASK+__svml_satan2_data_internal(%rip), %xmm6 + movaps %xmm7, %xmm0 + orps %xmm11, %xmm6 + movaps %xmm10, %xmm4 + andnps %xmm6, %xmm0 + movaps %xmm7, %xmm6 + movaps %xmm11, %xmm5 + andps %xmm7, %xmm4 + andnps %xmm10, %xmm6 + andps %xmm7, %xmm5 + orps %xmm4, %xmm0 + orps %xmm5, %xmm6 + +/* Division a/b. */ + divps %xmm6, %xmm0 + +/* Testing on working interval. */ + movdqu iCHK_WORK_SUB+__svml_satan2_data_internal(%rip), %xmm14 + movaps %xmm11, %xmm15 + movaps %xmm10, %xmm3 + psubd %xmm14, %xmm15 + psubd %xmm14, %xmm3 + movdqa %xmm15, %xmm1 + movdqu iCHK_WORK_CMP+__svml_satan2_data_internal(%rip), %xmm2 + movdqa %xmm3, %xmm14 + pcmpgtd %xmm2, %xmm1 + pcmpeqd %xmm2, %xmm15 + pcmpgtd %xmm2, %xmm14 + pcmpeqd %xmm2, %xmm3 + por %xmm15, %xmm1 + por %xmm3, %xmm14 + por %xmm14, %xmm1 + +/* Polynomial. */ + movaps %xmm0, %xmm14 + mulps %xmm0, %xmm14 + movaps %xmm13, %xmm4 + movmskps %xmm1, %ecx + movaps %xmm14, %xmm15 + movaps %xmm11, %xmm9 + mulps %xmm14, %xmm15 + pxor %xmm13, %xmm9 + movups sPC8+__svml_satan2_data_internal(%rip), %xmm2 + movaps %xmm10, %xmm8 + mulps %xmm15, %xmm2 + pxor %xmm12, %xmm8 + movups sPC7+__svml_satan2_data_internal(%rip), %xmm3 + xorl %edx, %edx + mulps %xmm15, %xmm3 + addps sPC6+__svml_satan2_data_internal(%rip), %xmm2 + mulps %xmm15, %xmm2 + addps sPC5+__svml_satan2_data_internal(%rip), %xmm3 + mulps %xmm15, %xmm3 + addps sPC4+__svml_satan2_data_internal(%rip), %xmm2 + mulps %xmm15, %xmm2 + addps sPC3+__svml_satan2_data_internal(%rip), %xmm3 + mulps %xmm15, %xmm3 + addps sPC2+__svml_satan2_data_internal(%rip), %xmm2 + mulps %xmm2, %xmm15 + addps sPC1+__svml_satan2_data_internal(%rip), %xmm3 + mulps %xmm3, %xmm14 + addps sPC0+__svml_satan2_data_internal(%rip), %xmm15 + +/* if x<0, sPI = Pi, else sPI =0 */ + movups __svml_satan2_data_internal(%rip), %xmm5 + xorl %eax, %eax + andnps sPIO2+__svml_satan2_data_internal(%rip), %xmm7 + addps %xmm14, %xmm15 + cmpleps %xmm5, %xmm4 + +/* Reconstruction. */ + mulps %xmm15, %xmm0 + andps sPI+__svml_satan2_data_internal(%rip), %xmm4 + addps %xmm7, %xmm0 + orps %xmm9, %xmm0 + addps %xmm4, %xmm0 + orps %xmm8, %xmm0 + +/* Special branch for fast (vector) processing of zero arguments */ + testl %ecx, %ecx + +/* Go to auxilary branch */ + jne L(AUX_BRANCH) + # LOE rbx rbp r12 r13 r14 r15 eax edx xmm0 xmm1 xmm4 xmm5 xmm6 xmm7 xmm8 xmm9 xmm10 xmm11 xmm12 xmm13 + +/* Return from auxilary branch + * for out of main path inputs + */ + +L(AUX_BRANCH_RETURN): +/* + * Special branch for fast (vector) processing of zero arguments + * 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 eax edx xmm0 xmm12 xmm13 + +/* Restore registers + * and exit the function + */ + +L(EXIT): + addq $88, %rsp + cfi_def_cfa_offset(8) + ret + cfi_def_cfa_offset(96) + +/* Branch to process + * special inputs + */ + +L(SPECIAL_VALUES_BRANCH): + movups %xmm12, 32(%rsp) + movups %xmm13, 48(%rsp) + movups %xmm0, 64(%rsp) + # LOE rbx rbp r12 r13 r14 r15 eax edx xmm0 + + 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), %xmm0 + +/* 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 xmm0 + +/* 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 atan2f@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) + cfi_restore(12) + cfi_restore(13) + cfi_restore(14) + # LOE rbx rbp r15 r12d r13d + +/* Auxilary branch + * for out of main path inputs + */ + +L(AUX_BRANCH): +/* Check if both X & Y are not NaNs: iXYnotNAN */ + movaps %xmm13, %xmm3 + movaps %xmm12, %xmm2 + cmpordps %xmm13, %xmm3 + cmpordps %xmm12, %xmm2 + +/* + * Path for zero arguments (at least one of both) + * Check if both args are zeros (den. is zero) + */ + cmpeqps %xmm5, %xmm6 + +/* Check if at least on of Y or Y is zero: iAXAYZERO */ + pcmpeqd %xmm5, %xmm11 + pcmpeqd %xmm5, %xmm10 + andps %xmm2, %xmm3 + por %xmm10, %xmm11 + +/* Check if at least on of Y or Y is zero and not NaN: iAXAYZEROnotNAN */ + andps %xmm3, %xmm11 + +/* Exclude from previous callout mask zero (and not NaN) arguments */ + movaps %xmm11, %xmm10 + pandn %xmm1, %xmm10 + +/* Set sPIO2 to zero if den. is zero */ + movaps %xmm6, %xmm1 + andnps %xmm7, %xmm1 + andps %xmm5, %xmm6 + orps %xmm6, %xmm1 + +/* Res = sign(Y)*(X<0)?(PIO2+PI):PIO2 */ + pcmpgtd %xmm13, %xmm5 + orps %xmm9, %xmm1 + andps %xmm4, %xmm5 + +/* Merge results from main and spec path */ + movaps %xmm11, %xmm4 + addps %xmm5, %xmm1 + +/* Go to callout */ + movmskps %xmm10, %edx + orps %xmm8, %xmm1 + andnps %xmm0, %xmm4 + andps %xmm11, %xmm1 + movaps %xmm4, %xmm0 + orps %xmm1, %xmm0 + +/* Return to main vector processing path */ + jmp L(AUX_BRANCH_RETURN) + # LOE rbx rbp r12 r13 r14 r15 eax edx xmm0 xmm12 xmm13 +END(_ZGVbN4vv_atan2f_sse4) + + .section .rodata, "a" + .align 16 + +#ifdef __svml_satan2_data_internal_typedef +typedef unsigned int VUINT32; +typedef struct { + __declspec(align(16)) VUINT32 sZERO[4][1]; + __declspec(align(16)) VUINT32 sSIGN_MASK[4][1]; + __declspec(align(16)) VUINT32 sABS_MASK[4][1]; + __declspec(align(16)) VUINT32 sPIO2[4][1]; + __declspec(align(16)) VUINT32 sPI[4][1]; + __declspec(align(16)) VUINT32 sPC8[4][1]; + __declspec(align(16)) VUINT32 sPC7[4][1]; + __declspec(align(16)) VUINT32 sPC6[4][1]; + __declspec(align(16)) VUINT32 sPC5[4][1]; + __declspec(align(16)) VUINT32 sPC4[4][1]; + __declspec(align(16)) VUINT32 sPC3[4][1]; + __declspec(align(16)) VUINT32 sPC2[4][1]; + __declspec(align(16)) VUINT32 sPC1[4][1]; + __declspec(align(16)) VUINT32 sPC0[4][1]; + __declspec(align(16)) VUINT32 iCHK_WORK_SUB[4][1]; + __declspec(align(16)) VUINT32 iCHK_WORK_CMP[4][1]; +} __svml_satan2_data_internal; +#endif +__svml_satan2_data_internal: + .long 0x00000000, 0x00000000, 0x00000000, 0x00000000 // sZERO + .align 16 + .long 0x80000000, 0x80000000, 0x80000000, 0x80000000 // sSIGN_MASK + .align 16 + .long 0x7FFFFFFF, 0x7FFFFFFF, 0x7FFFFFFF, 0x7FFFFFFF // sABS_MASK + .align 16 + .long 0x3FC90FDB, 0x3FC90FDB, 0x3FC90FDB, 0x3FC90FDB // sPIO2 + .align 16 + .long 0x40490FDB, 0x40490FDB, 0x40490FDB, 0x40490FDB // sPI + .align 16 + .long 0x3B322CC0, 0x3B322CC0, 0x3B322CC0, 0x3B322CC0 // sA08 + .align 16 + .long 0xBC7F2631, 0xBC7F2631, 0xBC7F2631, 0xBC7F2631 // sA07 + .align 16 + .long 0x3D2BC384, 0x3D2BC384, 0x3D2BC384, 0x3D2BC384 // sA06 + .align 16 + .long 0xBD987629, 0xBD987629, 0xBD987629, 0xBD987629 // sA05 + .align 16 + .long 0x3DD96474, 0x3DD96474, 0x3DD96474, 0x3DD96474 // sA04 + .align 16 + .long 0xBE1161F8, 0xBE1161F8, 0xBE1161F8, 0xBE1161F8 // sA03 + .align 16 + .long 0x3E4CB79F, 0x3E4CB79F, 0x3E4CB79F, 0x3E4CB79F // sA02 + .align 16 + .long 0xBEAAAA49, 0xBEAAAA49, 0xBEAAAA49, 0xBEAAAA49 // sA01 + .align 16 + .long 0x3F800000, 0x3F800000, 0x3F800000, 0x3F800000 // sA00 + .align 16 + .long 0x81000000, 0x81000000, 0x81000000, 0x81000000 //iCHK_WORK_SUB + .align 16 + .long 0xFC000000, 0xFC000000, 0xFC000000, 0xFC000000 //iCHK_WORK_CMP + .align 16 + .type __svml_satan2_data_internal,@object + .size __svml_satan2_data_internal,.-__svml_satan2_data_internal |