1/* Function cbrt vectorized with AVX2. 2 Copyright (C) 2021-2022 Free Software Foundation, Inc. 3 This file is part of the GNU C Library. 4 5 The GNU C Library is free software; you can redistribute it and/or 6 modify it under the terms of the GNU Lesser General Public 7 License as published by the Free Software Foundation; either 8 version 2.1 of the License, or (at your option) any later version. 9 10 The GNU C Library is distributed in the hope that it will be useful, 11 but WITHOUT ANY WARRANTY; without even the implied warranty of 12 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU 13 Lesser General Public License for more details. 14 15 You should have received a copy of the GNU Lesser General Public 16 License along with the GNU C Library; if not, see 17 https://www.gnu.org/licenses/. */ 18 19/* 20 * ALGORITHM DESCRIPTION: 21 * 22 * x=2^{3*k+j} * 1.b1 b2 ... b5 b6 ... b52 23 * Let r=(x*2^{-3k-j} - 1.b1 b2 ... b5 1)* rcp[b1 b2 ..b5], 24 * where rcp[b1 b2 .. b5]=1/(1.b1 b2 b3 b4 b5 1) in double precision 25 * cbrt(2^j * 1. b1 b2 .. b5 1) is approximated as T[j][b1..b5]+D[j][b1..b5] 26 * (T stores the high 53 bits, D stores the low order bits) 27 * Result=2^k*T+(2^k*T*r)*P+2^k*D 28 * where P=p1+p2*r+..+p8*r^7 29 * 30 */ 31 32/* Offsets for data table __svml_dcbrt_data_internal 33 */ 34#define _dRcp 0 35#define _dCbrtHiLo 256 36#define _dA7 1024 37#define _dA6 1056 38#define _dA5 1088 39#define _dA4 1120 40#define _dA3 1152 41#define _dA2 1184 42#define _dA1 1216 43#define _dNeg65Div64 1248 44#define _dSgnf6Mask 1280 45#define _dNegOne 1312 46#define _dMantissaMask 1344 47#define _lExpHiMask 1376 48#define _lExpLoMask 1408 49#define _l1556 1440 50#define _iRcpIndexMask 1472 51#define _iAbsMask 1504 52#define _iSignMask 1536 53#define _iBias 1568 54#define _iSub 1600 55#define _iCmp 1632 56 57#include <sysdep.h> 58 59 .section .text.avx2, "ax", @progbits 60ENTRY(_ZGVdN4v_cbrt_avx2) 61 pushq %rbp 62 cfi_def_cfa_offset(16) 63 movq %rsp, %rbp 64 cfi_def_cfa(6, 16) 65 cfi_offset(6, -16) 66 andq $-32, %rsp 67 subq $96, %rsp 68 69 /* Load 1/(1+iRcpIndex/32+1/64) reciprocal table value */ 70 lea __svml_dcbrt_data_internal(%rip), %rax 71 vmovapd %ymm0, %ymm5 72 73 /* 74 * Declarations 75 * Load constants 76 * Get iX - high part of argument 77 */ 78 vextractf128 $1, %ymm5, %xmm6 79 80 /* Calculate CbrtIndex */ 81 vpsrlq $52, %ymm5, %ymm15 82 vshufps $221, %xmm6, %xmm5, %xmm4 83 84 /* Calculate Rcp table index */ 85 vandps _iRcpIndexMask+__svml_dcbrt_data_internal(%rip), %xmm4, %xmm10 86 vpsrld $12, %xmm10, %xmm3 87 vmovd %xmm3, %ecx 88 89 /* If the exponent field is zero - go to callout to process denormals */ 90 vandps _iAbsMask+__svml_dcbrt_data_internal(%rip), %xmm4, %xmm7 91 92 /* Compute 2^k */ 93 vpsrld $20, %xmm4, %xmm4 94 vpsubd _iSub+__svml_dcbrt_data_internal(%rip), %xmm7, %xmm8 95 vandps _lExpLoMask+__svml_dcbrt_data_internal(%rip), %ymm15, %ymm0 96 vpmuludq _l1556+__svml_dcbrt_data_internal(%rip), %ymm0, %ymm6 97 vpextrd $2, %xmm3, %edi 98 movslq %ecx, %rcx 99 vpextrd $1, %xmm3, %esi 100 movslq %edi, %rdi 101 vpextrd $3, %xmm3, %r8d 102 movslq %esi, %rsi 103 movslq %r8d, %r8 104 vpcmpgtd _iCmp+__svml_dcbrt_data_internal(%rip), %xmm8, %xmm9 105 vmovsd (%rax, %rcx), %xmm11 106 vmovmskps %xmm9, %edx 107 vmovsd (%rax, %rdi), %xmm13 108 vmovhpd (%rax, %rsi), %xmm11, %xmm12 109 vmovhpd (%rax, %r8), %xmm13, %xmm14 110 vextractf128 $1, %ymm6, %xmm7 111 vshufps $136, %xmm7, %xmm6, %xmm8 112 vmovups __VUNPACK_ODD_ind1.613.0.1(%rip), %ymm7 113 vextractf128 $1, %ymm0, %xmm1 114 vshufps $136, %xmm1, %xmm0, %xmm9 115 vpsrld $14, %xmm8, %xmm1 116 vpsubd %xmm1, %xmm9, %xmm10 117 vpaddd %xmm1, %xmm1, %xmm11 118 119 /* 120 * VAND( L, l2k, = l2k, lExpHiMask ); 121 * Argument reduction Z 122 */ 123 vandpd _dMantissaMask+__svml_dcbrt_data_internal(%rip), %ymm5, %ymm9 124 vinsertf128 $1, %xmm14, %ymm12, %ymm2 125 vpsubd %xmm11, %xmm10, %xmm12 126 vpslld $8, %xmm12, %xmm13 127 vpaddd %xmm13, %xmm3, %xmm15 128 129 /* Load cbrt(2^j*(1+iRcpIndex/32+1/64)) Hi & Lo values */ 130 vmovd %xmm15, %r9d 131 vpextrd $2, %xmm15, %r11d 132 movslq %r9d, %r9 133 vpextrd $1, %xmm15, %r10d 134 movslq %r11d, %r11 135 vpextrd $3, %xmm15, %ecx 136 movslq %r10d, %r10 137 movslq %ecx, %rcx 138 vmovsd 256(%rax, %r9), %xmm3 139 vmovsd 256(%rax, %r11), %xmm0 140 vandpd _dSgnf6Mask+__svml_dcbrt_data_internal(%rip), %ymm5, %ymm10 141 vmovhpd 256(%rax, %r10), %xmm3, %xmm14 142 vmovhpd 256(%rax, %rcx), %xmm0, %xmm3 143 vorpd _dNegOne+__svml_dcbrt_data_internal(%rip), %ymm9, %ymm11 144 vorpd _dNeg65Div64+__svml_dcbrt_data_internal(%rip), %ymm10, %ymm12 145 vsubpd %ymm12, %ymm11, %ymm13 146 vmulpd %ymm13, %ymm2, %ymm2 147 vinsertf128 $1, %xmm3, %ymm14, %ymm0 148 vpand _iSignMask+__svml_dcbrt_data_internal(%rip), %xmm4, %xmm3 149 vpor _iBias+__svml_dcbrt_data_internal(%rip), %xmm3, %xmm4 150 vpaddd %xmm1, %xmm4, %xmm1 151 vpslld $20, %xmm1, %xmm6 152 153 /* Polynomial */ 154 vmovupd _dA7+__svml_dcbrt_data_internal(%rip), %ymm1 155 vfmadd213pd _dA6+__svml_dcbrt_data_internal(%rip), %ymm2, %ymm1 156 vfmadd213pd _dA5+__svml_dcbrt_data_internal(%rip), %ymm2, %ymm1 157 vfmadd213pd _dA4+__svml_dcbrt_data_internal(%rip), %ymm2, %ymm1 158 vfmadd213pd _dA3+__svml_dcbrt_data_internal(%rip), %ymm2, %ymm1 159 vfmadd213pd _dA2+__svml_dcbrt_data_internal(%rip), %ymm2, %ymm1 160 vfmadd213pd _dA1+__svml_dcbrt_data_internal(%rip), %ymm2, %ymm1 161 vpermps %ymm6, %ymm7, %ymm8 162 vandps __VUNPACK_ODD_mask.613.0.1(%rip), %ymm8, %ymm14 163 164 /* THi*2^k, TLo*2^k */ 165 vmulpd %ymm14, %ymm0, %ymm0 166 167 /* THi*2^k*Z */ 168 vmulpd %ymm0, %ymm2, %ymm2 169 170 /* Final reconstruction */ 171 vmulpd %ymm2, %ymm1, %ymm3 172 vaddpd %ymm3, %ymm0, %ymm0 173 testl %edx, %edx 174 175 /* Go to special inputs processing branch */ 176 jne L(SPECIAL_VALUES_BRANCH) 177 # LOE rbx r12 r13 r14 r15 edx ymm0 ymm5 178 179 /* Restore registers 180 * and exit the function 181 */ 182 183L(EXIT): 184 movq %rbp, %rsp 185 popq %rbp 186 cfi_def_cfa(7, 8) 187 cfi_restore(6) 188 ret 189 cfi_def_cfa(6, 16) 190 cfi_offset(6, -16) 191 192 /* Branch to process 193 * special inputs 194 */ 195 196L(SPECIAL_VALUES_BRANCH): 197 vmovupd %ymm5, 32(%rsp) 198 vmovupd %ymm0, 64(%rsp) 199 # LOE rbx r12 r13 r14 r15 edx ymm0 200 201 xorl %eax, %eax 202 # LOE rbx r12 r13 r14 r15 eax edx 203 204 vzeroupper 205 movq %r12, 16(%rsp) 206 /* 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) */ 207 .cfi_escape 0x10, 0x0c, 0x0e, 0x38, 0x1c, 0x0d, 0xe0, 0xff, 0xff, 0xff, 0x1a, 0x0d, 0xb0, 0xff, 0xff, 0xff, 0x22 208 movl %eax, %r12d 209 movq %r13, 8(%rsp) 210 /* 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) */ 211 .cfi_escape 0x10, 0x0d, 0x0e, 0x38, 0x1c, 0x0d, 0xe0, 0xff, 0xff, 0xff, 0x1a, 0x0d, 0xa8, 0xff, 0xff, 0xff, 0x22 212 movl %edx, %r13d 213 movq %r14, (%rsp) 214 /* 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) */ 215 .cfi_escape 0x10, 0x0e, 0x0e, 0x38, 0x1c, 0x0d, 0xe0, 0xff, 0xff, 0xff, 0x1a, 0x0d, 0xa0, 0xff, 0xff, 0xff, 0x22 216 # LOE rbx r15 r12d r13d 217 218 /* Range mask 219 * bits check 220 */ 221 222L(RANGEMASK_CHECK): 223 btl %r12d, %r13d 224 225 /* Call scalar math function */ 226 jc L(SCALAR_MATH_CALL) 227 # LOE rbx r15 r12d r13d 228 229 /* Special inputs 230 * processing loop 231 */ 232 233L(SPECIAL_VALUES_LOOP): 234 incl %r12d 235 cmpl $4, %r12d 236 237 /* Check bits in range mask */ 238 jl L(RANGEMASK_CHECK) 239 # LOE rbx r15 r12d r13d 240 241 movq 16(%rsp), %r12 242 cfi_restore(12) 243 movq 8(%rsp), %r13 244 cfi_restore(13) 245 movq (%rsp), %r14 246 cfi_restore(14) 247 vmovupd 64(%rsp), %ymm0 248 249 /* Go to exit */ 250 jmp L(EXIT) 251 /* 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) */ 252 .cfi_escape 0x10, 0x0c, 0x0e, 0x38, 0x1c, 0x0d, 0xe0, 0xff, 0xff, 0xff, 0x1a, 0x0d, 0xb0, 0xff, 0xff, 0xff, 0x22 253 /* 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) */ 254 .cfi_escape 0x10, 0x0d, 0x0e, 0x38, 0x1c, 0x0d, 0xe0, 0xff, 0xff, 0xff, 0x1a, 0x0d, 0xa8, 0xff, 0xff, 0xff, 0x22 255 /* 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) */ 256 .cfi_escape 0x10, 0x0e, 0x0e, 0x38, 0x1c, 0x0d, 0xe0, 0xff, 0xff, 0xff, 0x1a, 0x0d, 0xa0, 0xff, 0xff, 0xff, 0x22 257 # LOE rbx r12 r13 r14 r15 ymm0 258 259 /* Scalar math fucntion call 260 * to process special input 261 */ 262 263L(SCALAR_MATH_CALL): 264 movl %r12d, %r14d 265 vmovsd 32(%rsp, %r14, 8), %xmm0 266 call cbrt@PLT 267 # LOE rbx r14 r15 r12d r13d xmm0 268 269 vmovsd %xmm0, 64(%rsp, %r14, 8) 270 271 /* Process special inputs in loop */ 272 jmp L(SPECIAL_VALUES_LOOP) 273 # LOE rbx r15 r12d r13d 274END(_ZGVdN4v_cbrt_avx2) 275 .section .rodata, "a" 276 .align 32 277 278__VUNPACK_ODD_ind1.613.0.1: 279 .rept 3 280 .long 0 281 .endr 282 .long 1 283 .long 0 284 .long 2 285 .long 0 286 .long 3 287 .align 32 288 289__VUNPACK_ODD_mask.613.0.1: 290 .long 0 291 .long -1 292 .long 0 293 .long -1 294 .long 0 295 .long -1 296 .long 0 297 .long -1 298 299 .section .rodata, "a" 300 .align 32 301 302#ifdef __svml_dcbrt_data_internal_typedef 303typedef unsigned int VUINT32; 304typedef struct { 305 __declspec(align(32)) VUINT32 _dRcp[32][2]; 306 __declspec(align(32)) VUINT32 _dCbrtHiLo[96][2]; 307 __declspec(align(32)) VUINT32 _dA7[4][2]; 308 __declspec(align(32)) VUINT32 _dA6[4][2]; 309 __declspec(align(32)) VUINT32 _dA5[4][2]; 310 __declspec(align(32)) VUINT32 _dA4[4][2]; 311 __declspec(align(32)) VUINT32 _dA3[4][2]; 312 __declspec(align(32)) VUINT32 _dA2[4][2]; 313 __declspec(align(32)) VUINT32 _dA1[4][2]; 314 __declspec(align(32)) VUINT32 _dNeg65Div64[4][2]; 315 __declspec(align(32)) VUINT32 _dSgnf6Mask[4][2]; 316 __declspec(align(32)) VUINT32 _dNegOne[4][2]; 317 __declspec(align(32)) VUINT32 _dMantissaMask[4][2]; 318 __declspec(align(32)) VUINT32 _lExpHiMask[4][2]; 319 __declspec(align(32)) VUINT32 _lExpLoMask[4][2]; 320 __declspec(align(32)) VUINT32 _l1556[4][2]; 321 __declspec(align(32)) VUINT32 _iRcpIndexMask[8][1]; 322 __declspec(align(32)) VUINT32 _iAbsMask[8][1]; 323 __declspec(align(32)) VUINT32 _iSignMask[8][1]; 324 __declspec(align(32)) VUINT32 _iBias[8][1]; 325 __declspec(align(32)) VUINT32 _iSub[8][1]; 326 __declspec(align(32)) VUINT32 _iCmp[8][1]; 327} __svml_dcbrt_data_internal; 328#endif 329__svml_dcbrt_data_internal: 330 /* _dRcp */ 331 .quad 0xBFEF81F81F81F820 /* (1/(1+0/32+1/64)) = -.984615 */ 332 .quad 0xBFEE9131ABF0B767 /* (1/(1+1/32+1/64)) = -.955224 */ 333 .quad 0xBFEDAE6076B981DB /* (1/(1+2/32+1/64)) = -.927536 */ 334 .quad 0xBFECD85689039B0B /* (1/(1+3/32+1/64)) = -.901408 */ 335 .quad 0xBFEC0E070381C0E0 /* (1/(1+4/32+1/64)) = -.876712 */ 336 .quad 0xBFEB4E81B4E81B4F /* (1/(1+5/32+1/64)) = -.853333 */ 337 .quad 0xBFEA98EF606A63BE /* (1/(1+6/32+1/64)) = -.831169 */ 338 .quad 0xBFE9EC8E951033D9 /* (1/(1+7/32+1/64)) = -.810127 */ 339 .quad 0xBFE948B0FCD6E9E0 /* (1/(1+8/32+1/64)) = -.790123 */ 340 .quad 0xBFE8ACB90F6BF3AA /* (1/(1+9/32+1/64)) = -.771084 */ 341 .quad 0xBFE8181818181818 /* (1/(1+10/32+1/64)) = -.752941 */ 342 .quad 0xBFE78A4C8178A4C8 /* (1/(1+11/32+1/64)) = -.735632 */ 343 .quad 0xBFE702E05C0B8170 /* (1/(1+12/32+1/64)) = -.719101 */ 344 .quad 0xBFE6816816816817 /* (1/(1+13/32+1/64)) = -.703297 */ 345 .quad 0xBFE6058160581606 /* (1/(1+14/32+1/64)) = -.688172 */ 346 .quad 0xBFE58ED2308158ED /* (1/(1+15/32+1/64)) = -.673684 */ 347 .quad 0xBFE51D07EAE2F815 /* (1/(1+16/32+1/64)) = -.659794 */ 348 .quad 0xBFE4AFD6A052BF5B /* (1/(1+17/32+1/64)) = -.646465 */ 349 .quad 0xBFE446F86562D9FB /* (1/(1+18/32+1/64)) = -.633663 */ 350 .quad 0xBFE3E22CBCE4A902 /* (1/(1+19/32+1/64)) = -.621359 */ 351 .quad 0xBFE3813813813814 /* (1/(1+20/32+1/64)) = -.609524 */ 352 .quad 0xBFE323E34A2B10BF /* (1/(1+21/32+1/64)) = -.598131 */ 353 .quad 0xBFE2C9FB4D812CA0 /* (1/(1+22/32+1/64)) = -.587156 */ 354 .quad 0xBFE27350B8812735 /* (1/(1+23/32+1/64)) = -.576577 */ 355 .quad 0xBFE21FB78121FB78 /* (1/(1+24/32+1/64)) = -.566372 */ 356 .quad 0xBFE1CF06ADA2811D /* (1/(1+25/32+1/64)) = -.556522 */ 357 .quad 0xBFE1811811811812 /* (1/(1+26/32+1/64)) = -.547009 */ 358 .quad 0xBFE135C81135C811 /* (1/(1+27/32+1/64)) = -.537815 */ 359 .quad 0xBFE0ECF56BE69C90 /* (1/(1+28/32+1/64)) = -.528926 */ 360 .quad 0xBFE0A6810A6810A7 /* (1/(1+29/32+1/64)) = -.520325 */ 361 .quad 0xBFE0624DD2F1A9FC /* (1/(1+30/32+1/64)) = -.512 */ 362 .quad 0xBFE0204081020408 /* (1/(1+31/32+1/64)) = -.503937 */ 363 /* _dCbrtHiLo */ 364 .align 32 365 .quad 0x3FF01539221D4C97 /* HI((2^0*(1+0/32+1/64))^(1/3)) = 1.005181 */ 366 .quad 0x3FF03F06771A2E33 /* HI((2^0*(1+1/32+1/64))^(1/3)) = 1.015387 */ 367 .quad 0x3FF06800E629D671 /* HI((2^0*(1+2/32+1/64))^(1/3)) = 1.025391 */ 368 .quad 0x3FF090328731DEB2 /* HI((2^0*(1+3/32+1/64))^(1/3)) = 1.035204 */ 369 .quad 0x3FF0B7A4B1BD64AC /* HI((2^0*(1+4/32+1/64))^(1/3)) = 1.044835 */ 370 .quad 0x3FF0DE601024FB87 /* HI((2^0*(1+5/32+1/64))^(1/3)) = 1.054291 */ 371 .quad 0x3FF1046CB0597000 /* HI((2^0*(1+6/32+1/64))^(1/3)) = 1.06358 */ 372 .quad 0x3FF129D212A9BA9B /* HI((2^0*(1+7/32+1/64))^(1/3)) = 1.07271 */ 373 .quad 0x3FF14E9736CDAF38 /* HI((2^0*(1+8/32+1/64))^(1/3)) = 1.081687 */ 374 .quad 0x3FF172C2A772F507 /* HI((2^0*(1+9/32+1/64))^(1/3)) = 1.090518 */ 375 .quad 0x3FF1965A848001D3 /* HI((2^0*(1+10/32+1/64))^(1/3)) = 1.099207 */ 376 .quad 0x3FF1B9648C38C55D /* HI((2^0*(1+11/32+1/64))^(1/3)) = 1.107762 */ 377 .quad 0x3FF1DBE6236A0C45 /* HI((2^0*(1+12/32+1/64))^(1/3)) = 1.116186 */ 378 .quad 0x3FF1FDE45CBB1F9F /* HI((2^0*(1+13/32+1/64))^(1/3)) = 1.124485 */ 379 .quad 0x3FF21F63FF409042 /* HI((2^0*(1+14/32+1/64))^(1/3)) = 1.132664 */ 380 .quad 0x3FF240698C6746E5 /* HI((2^0*(1+15/32+1/64))^(1/3)) = 1.140726 */ 381 .quad 0x3FF260F9454BB99B /* HI((2^0*(1+16/32+1/64))^(1/3)) = 1.148675 */ 382 .quad 0x3FF281172F8E7073 /* HI((2^0*(1+17/32+1/64))^(1/3)) = 1.156516 */ 383 .quad 0x3FF2A0C719B4B6D0 /* HI((2^0*(1+18/32+1/64))^(1/3)) = 1.164252 */ 384 .quad 0x3FF2C00C9F2263EC /* HI((2^0*(1+19/32+1/64))^(1/3)) = 1.171887 */ 385 .quad 0x3FF2DEEB2BB7FB78 /* HI((2^0*(1+20/32+1/64))^(1/3)) = 1.179423 */ 386 .quad 0x3FF2FD65FF1EFBBC /* HI((2^0*(1+21/32+1/64))^(1/3)) = 1.186865 */ 387 .quad 0x3FF31B802FCCF6A2 /* HI((2^0*(1+22/32+1/64))^(1/3)) = 1.194214 */ 388 .quad 0x3FF3393CADC50708 /* HI((2^0*(1+23/32+1/64))^(1/3)) = 1.201474 */ 389 .quad 0x3FF3569E451E4C2A /* HI((2^0*(1+24/32+1/64))^(1/3)) = 1.208647 */ 390 .quad 0x3FF373A7A0554CDE /* HI((2^0*(1+25/32+1/64))^(1/3)) = 1.215736 */ 391 .quad 0x3FF3905B4A6D76CE /* HI((2^0*(1+26/32+1/64))^(1/3)) = 1.222743 */ 392 .quad 0x3FF3ACBBB0E756B6 /* HI((2^0*(1+27/32+1/64))^(1/3)) = 1.229671 */ 393 .quad 0x3FF3C8CB258FA340 /* HI((2^0*(1+28/32+1/64))^(1/3)) = 1.236522 */ 394 .quad 0x3FF3E48BE02AC0CE /* HI((2^0*(1+29/32+1/64))^(1/3)) = 1.243297 */ 395 .quad 0x3FF4000000000000 /* HI((2^0*(1+30/32+1/64))^(1/3)) = 1.25 */ 396 .quad 0x3FF41B298D47800E /* HI((2^0*(1+31/32+1/64))^(1/3)) = 1.256631 */ 397 .quad 0x3FF443604B34D9B2 /* HI((2^1*(1+0/32+1/64))^(1/3)) = 1.266449 */ 398 .quad 0x3FF4780B20906571 /* HI((2^1*(1+1/32+1/64))^(1/3)) = 1.279307 */ 399 .quad 0x3FF4ABAC3EE06706 /* HI((2^1*(1+2/32+1/64))^(1/3)) = 1.291912 */ 400 .quad 0x3FF4DE505DA66B8D /* HI((2^1*(1+3/32+1/64))^(1/3)) = 1.304276 */ 401 .quad 0x3FF51003420A5C07 /* HI((2^1*(1+4/32+1/64))^(1/3)) = 1.316409 */ 402 .quad 0x3FF540CFD6FD11C1 /* HI((2^1*(1+5/32+1/64))^(1/3)) = 1.328323 */ 403 .quad 0x3FF570C04260716B /* HI((2^1*(1+6/32+1/64))^(1/3)) = 1.340027 */ 404 .quad 0x3FF59FDDF7A45F38 /* HI((2^1*(1+7/32+1/64))^(1/3)) = 1.35153 */ 405 .quad 0x3FF5CE31C83539DF /* HI((2^1*(1+8/32+1/64))^(1/3)) = 1.36284 */ 406 .quad 0x3FF5FBC3F20966A4 /* HI((2^1*(1+9/32+1/64))^(1/3)) = 1.373966 */ 407 .quad 0x3FF6289C2C8F1B70 /* HI((2^1*(1+10/32+1/64))^(1/3)) = 1.384915 */ 408 .quad 0x3FF654C1B4316DCF /* HI((2^1*(1+11/32+1/64))^(1/3)) = 1.395693 */ 409 .quad 0x3FF6803B54A34E44 /* HI((2^1*(1+12/32+1/64))^(1/3)) = 1.406307 */ 410 .quad 0x3FF6AB0F72182659 /* HI((2^1*(1+13/32+1/64))^(1/3)) = 1.416763 */ 411 .quad 0x3FF6D544118C08BC /* HI((2^1*(1+14/32+1/64))^(1/3)) = 1.427067 */ 412 .quad 0x3FF6FEDEE0388D4A /* HI((2^1*(1+15/32+1/64))^(1/3)) = 1.437224 */ 413 .quad 0x3FF727E53A4F645E /* HI((2^1*(1+16/32+1/64))^(1/3)) = 1.44724 */ 414 .quad 0x3FF7505C31104114 /* HI((2^1*(1+17/32+1/64))^(1/3)) = 1.457119 */ 415 .quad 0x3FF77848904CD549 /* HI((2^1*(1+18/32+1/64))^(1/3)) = 1.466866 */ 416 .quad 0x3FF79FAEE36B2534 /* HI((2^1*(1+19/32+1/64))^(1/3)) = 1.476485 */ 417 .quad 0x3FF7C69379F4605B /* HI((2^1*(1+20/32+1/64))^(1/3)) = 1.48598 */ 418 .quad 0x3FF7ECFA6BBCA391 /* HI((2^1*(1+21/32+1/64))^(1/3)) = 1.495356 */ 419 .quad 0x3FF812E79CAE7EB9 /* HI((2^1*(1+22/32+1/64))^(1/3)) = 1.504615 */ 420 .quad 0x3FF8385EC043C71D /* HI((2^1*(1+23/32+1/64))^(1/3)) = 1.513762 */ 421 .quad 0x3FF85D635CB41B9D /* HI((2^1*(1+24/32+1/64))^(1/3)) = 1.5228 */ 422 .quad 0x3FF881F8CDE083DB /* HI((2^1*(1+25/32+1/64))^(1/3)) = 1.531731 */ 423 .quad 0x3FF8A6224802B8A8 /* HI((2^1*(1+26/32+1/64))^(1/3)) = 1.54056 */ 424 .quad 0x3FF8C9E2DA25E5E4 /* HI((2^1*(1+27/32+1/64))^(1/3)) = 1.549289 */ 425 .quad 0x3FF8ED3D706E1010 /* HI((2^1*(1+28/32+1/64))^(1/3)) = 1.55792 */ 426 .quad 0x3FF91034D632B6DF /* HI((2^1*(1+29/32+1/64))^(1/3)) = 1.566457 */ 427 .quad 0x3FF932CBB7F0CF2D /* HI((2^1*(1+30/32+1/64))^(1/3)) = 1.574901 */ 428 .quad 0x3FF95504A517BF3A /* HI((2^1*(1+31/32+1/64))^(1/3)) = 1.583256 */ 429 .quad 0x3FF987AF34F8BB19 /* HI((2^2*(1+0/32+1/64))^(1/3)) = 1.595626 */ 430 .quad 0x3FF9CA0A8337B317 /* HI((2^2*(1+1/32+1/64))^(1/3)) = 1.611826 */ 431 .quad 0x3FFA0B1709CC13D5 /* HI((2^2*(1+2/32+1/64))^(1/3)) = 1.627708 */ 432 .quad 0x3FFA4AE4CE6419ED /* HI((2^2*(1+3/32+1/64))^(1/3)) = 1.643285 */ 433 .quad 0x3FFA8982A5567031 /* HI((2^2*(1+4/32+1/64))^(1/3)) = 1.658572 */ 434 .quad 0x3FFAC6FE500AB570 /* HI((2^2*(1+5/32+1/64))^(1/3)) = 1.673582 */ 435 .quad 0x3FFB036497A15A17 /* HI((2^2*(1+6/32+1/64))^(1/3)) = 1.688328 */ 436 .quad 0x3FFB3EC164671755 /* HI((2^2*(1+7/32+1/64))^(1/3)) = 1.702821 */ 437 .quad 0x3FFB791FD288C46F /* HI((2^2*(1+8/32+1/64))^(1/3)) = 1.717071 */ 438 .quad 0x3FFBB28A44693BE4 /* HI((2^2*(1+9/32+1/64))^(1/3)) = 1.731089 */ 439 .quad 0x3FFBEB0A72EB6E31 /* HI((2^2*(1+10/32+1/64))^(1/3)) = 1.744883 */ 440 .quad 0x3FFC22A97BF5F697 /* HI((2^2*(1+11/32+1/64))^(1/3)) = 1.758462 */ 441 .quad 0x3FFC596FEF6AF983 /* HI((2^2*(1+12/32+1/64))^(1/3)) = 1.771835 */ 442 .quad 0x3FFC8F65DAC655A3 /* HI((2^2*(1+13/32+1/64))^(1/3)) = 1.785009 */ 443 .quad 0x3FFCC492D38CE8D9 /* HI((2^2*(1+14/32+1/64))^(1/3)) = 1.797992 */ 444 .quad 0x3FFCF8FE00B19367 /* HI((2^2*(1+15/32+1/64))^(1/3)) = 1.810789 */ 445 .quad 0x3FFD2CAE230F8709 /* HI((2^2*(1+16/32+1/64))^(1/3)) = 1.823408 */ 446 .quad 0x3FFD5FA99D15208F /* HI((2^2*(1+17/32+1/64))^(1/3)) = 1.835855 */ 447 .quad 0x3FFD91F679B6E505 /* HI((2^2*(1+18/32+1/64))^(1/3)) = 1.848135 */ 448 .quad 0x3FFDC39A72BF2302 /* HI((2^2*(1+19/32+1/64))^(1/3)) = 1.860255 */ 449 .quad 0x3FFDF49AF68C1570 /* HI((2^2*(1+20/32+1/64))^(1/3)) = 1.872218 */ 450 .quad 0x3FFE24FD2D4C23B8 /* HI((2^2*(1+21/32+1/64))^(1/3)) = 1.884031 */ 451 .quad 0x3FFE54C5FDC5EC73 /* HI((2^2*(1+22/32+1/64))^(1/3)) = 1.895697 */ 452 .quad 0x3FFE83FA11B81DBB /* HI((2^2*(1+23/32+1/64))^(1/3)) = 1.907221 */ 453 .quad 0x3FFEB29DD9DBAF25 /* HI((2^2*(1+24/32+1/64))^(1/3)) = 1.918608 */ 454 .quad 0x3FFEE0B59191D374 /* HI((2^2*(1+25/32+1/64))^(1/3)) = 1.929861 */ 455 .quad 0x3FFF0E454245E4BF /* HI((2^2*(1+26/32+1/64))^(1/3)) = 1.940984 */ 456 .quad 0x3FFF3B50C68A9DD3 /* HI((2^2*(1+27/32+1/64))^(1/3)) = 1.951981 */ 457 .quad 0x3FFF67DBCCF922DC /* HI((2^2*(1+28/32+1/64))^(1/3)) = 1.962856 */ 458 .quad 0x3FFF93E9DAD7A4A6 /* HI((2^2*(1+29/32+1/64))^(1/3)) = 1.973612 */ 459 .quad 0x3FFFBF7E4E8CC9CB /* HI((2^2*(1+30/32+1/64))^(1/3)) = 1.984251 */ 460 .quad 0x3FFFEA9C61E47CD3 /* HI((2^2*(1+31/32+1/64))^(1/3)) = 1.994778 */ 461 .align 32 462 .quad 0x3F93750AD588F115, 0x3F93750AD588F115, 0x3F93750AD588F115, 0x3F93750AD588F115 /* _dA7 */ 463 .align 32 464 .quad 0xBF98090D6221A247, 0xBF98090D6221A247, 0xBF98090D6221A247, 0xBF98090D6221A247 /* _dA6 */ 465 .align 32 466 .quad 0x3F9EE7113506AC12, 0x3F9EE7113506AC12, 0x3F9EE7113506AC12, 0x3F9EE7113506AC12 /* _dA5 */ 467 .align 32 468 .quad 0xBFA511E8D2B3183B, 0xBFA511E8D2B3183B, 0xBFA511E8D2B3183B, 0xBFA511E8D2B3183B /* _dA4 */ 469 .align 32 470 .quad 0x3FAF9ADD3C0CA458, 0x3FAF9ADD3C0CA458, 0x3FAF9ADD3C0CA458, 0x3FAF9ADD3C0CA458 /* _dA3 */ 471 .align 32 472 .quad 0xBFBC71C71C71C71C, 0xBFBC71C71C71C71C, 0xBFBC71C71C71C71C, 0xBFBC71C71C71C71C /* _dA2 */ 473 .align 32 474 .quad 0x3FD5555555555555, 0x3FD5555555555555, 0x3FD5555555555555, 0x3FD5555555555555 /* _dA1 */ 475 .align 32 476 .quad 0xBFF0400000000000, 0xBFF0400000000000, 0xBFF0400000000000, 0xBFF0400000000000 /* _dNeg65Div64 */ 477 .align 32 478 .quad 0x000FC00000000000, 0x000FC00000000000, 0x000FC00000000000, 0x000FC00000000000 /* _dSgnf6Mask */ 479 .align 32 480 .quad 0xBFF0000000000000, 0xBFF0000000000000, 0xBFF0000000000000, 0xBFF0000000000000 /* _dNegOne */ 481 .align 32 482 .quad 0x000FFFFFFFFFFFFF, 0x000FFFFFFFFFFFFF, 0x000FFFFFFFFFFFFF, 0x000FFFFFFFFFFFFF /* _dMantissaMask */ 483 .align 32 484 .quad 0xFFF0000000000000, 0xFFF0000000000000, 0xFFF0000000000000, 0xFFF0000000000000 /* _lExpHiMask */ 485 .align 32 486 .quad 0x00000000000007FF, 0x00000000000007FF, 0x00000000000007FF, 0x00000000000007FF /* _lExpLoMask */ 487 .align 32 488 .quad 0x0000000000001556, 0x0000000000001556, 0x0000000000001556, 0x0000000000001556 /* _l1556 */ 489 .align 32 490 .long 0x000F8000, 0x000F8000, 0x000F8000, 0x000F8000, 0x000F8000, 0x000F8000, 0x000F8000, 0x000F8000 /* _iRcpIndexMask */ 491 .align 32 492 .long 0x7FFFFFFF, 0x7FFFFFFF, 0x7FFFFFFF, 0x7FFFFFFF, 0x7FFFFFFF, 0x7FFFFFFF, 0x7FFFFFFF, 0x7FFFFFFF /* _iAbsMask */ 493 .align 32 494 .long 0x00000800, 0x00000800, 0x00000800, 0x00000800, 0x00000800, 0x00000800, 0x00000800, 0x00000800 /* _iSignMask */ 495 .align 32 496 .long 0x000002AA, 0x000002AA, 0x000002AA, 0x000002AA, 0x000002AA, 0x000002AA, 0x000002AA, 0x000002AA /* _iBias */ 497 .align 32 498 .long 0x80100000, 0x80100000, 0x80100000, 0x80100000, 0x80100000, 0x80100000, 0x80100000, 0x80100000 /* _iSub */ 499 .align 32 500 .long 0xffdfffff, 0xffdfffff, 0xffdfffff, 0xffdfffff, 0xffdfffff, 0xffdfffff, 0xffdfffff, 0xffdfffff /* _iCmp */ 501 .align 32 502 .type __svml_dcbrt_data_internal, @object 503 .size __svml_dcbrt_data_internal, .-__svml_dcbrt_data_internal 504