1 /* 2 * CDDL HEADER START 3 * 4 * The contents of this file are subject to the terms of the 5 * Common Development and Distribution License (the "License"). 6 * You may not use this file except in compliance with the License. 7 * 8 * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE 9 * or http://www.opensolaris.org/os/licensing. 10 * See the License for the specific language governing permissions 11 * and limitations under the License. 12 * 13 * When distributing Covered Code, include this CDDL HEADER in each 14 * file and include the License file at usr/src/OPENSOLARIS.LICENSE. 15 * If applicable, add the following below this CDDL HEADER, with the 16 * fields enclosed by brackets "[]" replaced with your own identifying 17 * information: Portions Copyright [yyyy] [name of copyright owner] 18 * 19 * CDDL HEADER END 20 */ 21 22 /* 23 * Copyright 2011 Nexenta Systems, Inc. All rights reserved. 24 */ 25 /* 26 * Copyright 2006 Sun Microsystems, Inc. All rights reserved. 27 * Use is subject to license terms. 28 */ 29 30 #include <sys/isa_defs.h> 31 #include <sys/ccompile.h> 32 33 #ifdef _LITTLE_ENDIAN 34 #define HI(x) *(1+(int*)x) 35 #define LO(x) *(unsigned*)x 36 #else 37 #define HI(x) *(int*)x 38 #define LO(x) *(1+(unsigned*)x) 39 #endif 40 41 #ifdef __RESTRICT 42 #define restrict _Restrict 43 #else 44 #define restrict 45 #endif 46 47 /* 48 * vcos.1.c 49 * 50 * Vector cosine function. Just slight modifications to vsin.8.c, mainly 51 * in the primary range part. 52 * 53 * Modification to primary range processing. If an argument that does not 54 * fall in the primary range is encountered, then processing is continued 55 * in the medium range. 56 * 57 */ 58 59 extern const double __vlibm_TBL_sincos_hi[], __vlibm_TBL_sincos_lo[]; 60 61 static const double 62 half[2] = { 0.5, -0.5 }, 63 one = 1.0, 64 invpio2 = 0.636619772367581343075535, /* 53 bits of pi/2 */ 65 pio2_1 = 1.570796326734125614166, /* first 33 bits of pi/2 */ 66 pio2_2 = 6.077100506303965976596e-11, /* second 33 bits of pi/2 */ 67 pio2_3 = 2.022266248711166455796e-21, /* third 33 bits of pi/2 */ 68 pio2_3t = 8.478427660368899643959e-32, /* pi/2 - pio2_3 */ 69 pp1 = -1.666666666605760465276263943134982554676e-0001, 70 pp2 = 8.333261209690963126718376566146180944442e-0003, 71 qq1 = -4.999999999977710986407023955908711557870e-0001, 72 qq2 = 4.166654863857219350645055881018842089580e-0002, 73 poly1[2]= { -1.666666666666629669805215138920301589656e-0001, 74 -4.999999999999931701464060878888294524481e-0001 }, 75 poly2[2]= { 8.333333332390951295683993455280336376663e-0003, 76 4.166666666394861917535640593963708222319e-0002 }, 77 poly3[2]= { -1.984126237997976692791551778230098403960e-0004, 78 -1.388888552656142867832756687736851681462e-0003 }, 79 poly4[2]= { 2.753403624854277237649987622848330351110e-0006, 80 2.478519423681460796618128289454530524759e-0005 }; 81 82 static const unsigned thresh[2] = { 0x3fc90000, 0x3fc40000 }; 83 84 /* Don't __ the following; acomp will handle it */ 85 extern double fabs(double); 86 extern void __vlibm_vcos_big(int, double *, int, double *, int, int); 87 88 /* 89 * y[i*stridey] := cos( x[i*stridex] ), for i = 0..n. 90 * 91 * Calls __vlibm_vcos_big to handle all elts which have abs >~ 1.647e+06. 92 * Argument reduction is done here for elts pi/4 < arg < 1.647e+06. 93 * 94 * elts < 2^-27 use the approximation 1.0 ~ cos(x). 95 */ 96 void 97 __vcos(int n, double * restrict x, int stridex, double * restrict y, 98 int stridey) 99 { 100 double x0_or_one[4], x1_or_one[4], x2_or_one[4]; 101 double y0_or_zero[4], y1_or_zero[4], y2_or_zero[4]; 102 double x0, x1, x2, *py0 = 0, *py1 = 0, *py2, *xsave, *ysave; 103 unsigned hx0, hx1, hx2, xsb0, xsb1 = 0, xsb2; 104 int i, biguns, nsave, sxsave, sysave; 105 volatile int v __GNU_UNUSED; 106 nsave = n; 107 xsave = x; 108 sxsave = stridex; 109 ysave = y; 110 sysave = stridey; 111 biguns = 0; 112 113 do /* MAIN LOOP */ 114 { 115 /* Gotos here so _break_ exits MAIN LOOP. */ 116 LOOP0: /* Find first arg in right range. */ 117 xsb0 = HI(x); /* get most significant word */ 118 hx0 = xsb0 & ~0x80000000; /* mask off sign bit */ 119 if (hx0 > 0x3fe921fb) { 120 /* Too big: arg reduction needed, so leave for second part */ 121 biguns = 1; 122 goto MEDIUM; 123 } 124 if (hx0 < 0x3e400000) { 125 /* Too small. cos x ~ 1. */ 126 v = *x; 127 *y = 1.0; 128 x += stridex; 129 y += stridey; 130 i = 0; 131 if (--n <= 0) 132 break; 133 goto LOOP0; 134 } 135 x0 = *x; 136 py0 = y; 137 x += stridex; 138 y += stridey; 139 i = 1; 140 if (--n <= 0) 141 break; 142 143 LOOP1: /* Get second arg, same as above. */ 144 xsb1 = HI(x); 145 hx1 = xsb1 & ~0x80000000; 146 if (hx1 > 0x3fe921fb) 147 { 148 biguns = 2; 149 goto MEDIUM; 150 } 151 if (hx1 < 0x3e400000) 152 { 153 v = *x; 154 *y = 1.0; 155 x += stridex; 156 y += stridey; 157 i = 1; 158 if (--n <= 0) 159 break; 160 goto LOOP1; 161 } 162 x1 = *x; 163 py1 = y; 164 x += stridex; 165 y += stridey; 166 i = 2; 167 if (--n <= 0) 168 break; 169 170 LOOP2: /* Get third arg, same as above. */ 171 xsb2 = HI(x); 172 hx2 = xsb2 & ~0x80000000; 173 if (hx2 > 0x3fe921fb) 174 { 175 biguns = 3; 176 goto MEDIUM; 177 } 178 if (hx2 < 0x3e400000) 179 { 180 v = *x; 181 *y = 1.0; 182 x += stridex; 183 y += stridey; 184 i = 2; 185 if (--n <= 0) 186 break; 187 goto LOOP2; 188 } 189 x2 = *x; 190 py2 = y; 191 192 /* 193 * 0x3fc40000 = 5/32 ~ 0.15625 194 * Get msb after subtraction. Will be 1 only if 195 * hx0 - 5/32 is negative. 196 */ 197 i = (hx0 - 0x3fc40000) >> 31; 198 i |= ((hx1 - 0x3fc40000) >> 30) & 2; 199 i |= ((hx2 - 0x3fc40000) >> 29) & 4; 200 switch (i) 201 { 202 double a0, a1, a2, w0, w1, w2; 203 double t0, t1, t2, z0, z1, z2; 204 unsigned j0, j1, j2; 205 206 case 0: /* All are > 5/32 */ 207 j0 = (xsb0 + 0x4000) & 0xffff8000; 208 j1 = (xsb1 + 0x4000) & 0xffff8000; 209 j2 = (xsb2 + 0x4000) & 0xffff8000; 210 HI(&t0) = j0; 211 HI(&t1) = j1; 212 HI(&t2) = j2; 213 LO(&t0) = 0; 214 LO(&t1) = 0; 215 LO(&t2) = 0; 216 x0 -= t0; 217 x1 -= t1; 218 x2 -= t2; 219 z0 = x0 * x0; 220 z1 = x1 * x1; 221 z2 = x2 * x2; 222 t0 = z0 * (qq1 + z0 * qq2); 223 t1 = z1 * (qq1 + z1 * qq2); 224 t2 = z2 * (qq1 + z2 * qq2); 225 w0 = x0 * (one + z0 * (pp1 + z0 * pp2)); 226 w1 = x1 * (one + z1 * (pp1 + z1 * pp2)); 227 w2 = x2 * (one + z2 * (pp1 + z2 * pp2)); 228 j0 = (((j0 & ~0x80000000) - 0x3fc40000) >> 13) & ~0x3; 229 j1 = (((j1 & ~0x80000000) - 0x3fc40000) >> 13) & ~0x3; 230 j2 = (((j2 & ~0x80000000) - 0x3fc40000) >> 13) & ~0x3; 231 xsb0 = (xsb0 >> 30) & 2; 232 xsb1 = (xsb1 >> 30) & 2; 233 xsb2 = (xsb2 >> 30) & 2; 234 a0 = __vlibm_TBL_sincos_hi[j0+1]; /* cos_hi(t) */ 235 a1 = __vlibm_TBL_sincos_hi[j1+1]; 236 a2 = __vlibm_TBL_sincos_hi[j2+1]; 237 /* cos_lo(t) sin_hi(t) */ 238 t0 = __vlibm_TBL_sincos_lo[j0+1] - (__vlibm_TBL_sincos_hi[j0+xsb0]*w0 - a0*t0); 239 t1 = __vlibm_TBL_sincos_lo[j1+1] - (__vlibm_TBL_sincos_hi[j1+xsb1]*w1 - a1*t1); 240 t2 = __vlibm_TBL_sincos_lo[j2+1] - (__vlibm_TBL_sincos_hi[j2+xsb2]*w2 - a2*t2); 241 242 *py0 = a0 + t0; 243 *py1 = a1 + t1; 244 *py2 = a2 + t2; 245 break; 246 247 case 1: 248 j1 = (xsb1 + 0x4000) & 0xffff8000; 249 j2 = (xsb2 + 0x4000) & 0xffff8000; 250 HI(&t1) = j1; 251 HI(&t2) = j2; 252 LO(&t1) = 0; 253 LO(&t2) = 0; 254 x1 -= t1; 255 x2 -= t2; 256 z0 = x0 * x0; 257 z1 = x1 * x1; 258 z2 = x2 * x2; 259 t0 = z0 * (poly3[1] + z0 * poly4[1]); 260 t1 = z1 * (qq1 + z1 * qq2); 261 t2 = z2 * (qq1 + z2 * qq2); 262 t0 = z0 * (poly1[1] + z0 * (poly2[1] + t0)); 263 w1 = x1 * (one + z1 * (pp1 + z1 * pp2)); 264 w2 = x2 * (one + z2 * (pp1 + z2 * pp2)); 265 j1 = (((j1 & ~0x80000000) - 0x3fc40000) >> 13) & ~0x3; 266 j2 = (((j2 & ~0x80000000) - 0x3fc40000) >> 13) & ~0x3; 267 xsb1 = (xsb1 >> 30) & 2; 268 xsb2 = (xsb2 >> 30) & 2; 269 a1 = __vlibm_TBL_sincos_hi[j1+1]; 270 a2 = __vlibm_TBL_sincos_hi[j2+1]; 271 t1 = __vlibm_TBL_sincos_lo[j1+1] - (__vlibm_TBL_sincos_hi[j1+xsb1]*w1 - a1*t1); 272 t2 = __vlibm_TBL_sincos_lo[j2+1] - (__vlibm_TBL_sincos_hi[j2+xsb2]*w2 - a2*t2); 273 *py0 = one + t0; 274 *py1 = a1 + t1; 275 *py2 = a2 + t2; 276 break; 277 278 case 2: 279 j0 = (xsb0 + 0x4000) & 0xffff8000; 280 j2 = (xsb2 + 0x4000) & 0xffff8000; 281 HI(&t0) = j0; 282 HI(&t2) = j2; 283 LO(&t0) = 0; 284 LO(&t2) = 0; 285 x0 -= t0; 286 x2 -= t2; 287 z0 = x0 * x0; 288 z1 = x1 * x1; 289 z2 = x2 * x2; 290 t0 = z0 * (qq1 + z0 * qq2); 291 t1 = z1 * (poly3[1] + z1 * poly4[1]); 292 t2 = z2 * (qq1 + z2 * qq2); 293 w0 = x0 * (one + z0 * (pp1 + z0 * pp2)); 294 t1 = z1 * (poly1[1] + z1 * (poly2[1] + t1)); 295 w2 = x2 * (one + z2 * (pp1 + z2 * pp2)); 296 j0 = (((j0 & ~0x80000000) - 0x3fc40000) >> 13) & ~0x3; 297 j2 = (((j2 & ~0x80000000) - 0x3fc40000) >> 13) & ~0x3; 298 xsb0 = (xsb0 >> 30) & 2; 299 xsb2 = (xsb2 >> 30) & 2; 300 a0 = __vlibm_TBL_sincos_hi[j0+1]; 301 a2 = __vlibm_TBL_sincos_hi[j2+1]; 302 t0 = __vlibm_TBL_sincos_lo[j0+1] - (__vlibm_TBL_sincos_hi[j0+xsb0]*w0 - a0*t0); 303 t2 = __vlibm_TBL_sincos_lo[j2+1] - (__vlibm_TBL_sincos_hi[j2+xsb2]*w2 - a2*t2); 304 *py0 = a0 + t0; 305 *py1 = one + t1; 306 *py2 = a2 + t2; 307 break; 308 309 case 3: 310 j2 = (xsb2 + 0x4000) & 0xffff8000; 311 HI(&t2) = j2; 312 LO(&t2) = 0; 313 x2 -= t2; 314 z0 = x0 * x0; 315 z1 = x1 * x1; 316 z2 = x2 * x2; 317 t0 = z0 * (poly3[1] + z0 * poly4[1]); 318 t1 = z1 * (poly3[1] + z1 * poly4[1]); 319 t2 = z2 * (qq1 + z2 * qq2); 320 t0 = z0 * (poly1[1] + z0 * (poly2[1] + t0)); 321 t1 = z1 * (poly1[1] + z1 * (poly2[1] + t1)); 322 w2 = x2 * (one + z2 * (pp1 + z2 * pp2)); 323 j2 = (((j2 & ~0x80000000) - 0x3fc40000) >> 13) & ~0x3; 324 xsb2 = (xsb2 >> 30) & 2; 325 a2 = __vlibm_TBL_sincos_hi[j2+1]; 326 t2 = __vlibm_TBL_sincos_lo[j2+1] - (__vlibm_TBL_sincos_hi[j2+xsb2]*w2 - a2*t2); 327 *py0 = one + t0; 328 *py1 = one + t1; 329 *py2 = a2 + t2; 330 break; 331 332 case 4: 333 j0 = (xsb0 + 0x4000) & 0xffff8000; 334 j1 = (xsb1 + 0x4000) & 0xffff8000; 335 HI(&t0) = j0; 336 HI(&t1) = j1; 337 LO(&t0) = 0; 338 LO(&t1) = 0; 339 x0 -= t0; 340 x1 -= t1; 341 z0 = x0 * x0; 342 z1 = x1 * x1; 343 z2 = x2 * x2; 344 t0 = z0 * (qq1 + z0 * qq2); 345 t1 = z1 * (qq1 + z1 * qq2); 346 t2 = z2 * (poly3[1] + z2 * poly4[1]); 347 w0 = x0 * (one + z0 * (pp1 + z0 * pp2)); 348 w1 = x1 * (one + z1 * (pp1 + z1 * pp2)); 349 t2 = z2 * (poly1[1] + z2 * (poly2[1] + t2)); 350 j0 = (((j0 & ~0x80000000) - 0x3fc40000) >> 13) & ~0x3; 351 j1 = (((j1 & ~0x80000000) - 0x3fc40000) >> 13) & ~0x3; 352 xsb0 = (xsb0 >> 30) & 2; 353 xsb1 = (xsb1 >> 30) & 2; 354 a0 = __vlibm_TBL_sincos_hi[j0+1]; 355 a1 = __vlibm_TBL_sincos_hi[j1+1]; 356 t0 = __vlibm_TBL_sincos_lo[j0+1] - (__vlibm_TBL_sincos_hi[j0+xsb0]*w0 - a0*t0); 357 t1 = __vlibm_TBL_sincos_lo[j1+1] - (__vlibm_TBL_sincos_hi[j1+xsb1]*w1 - a1*t1); 358 *py0 = a0 + t0; 359 *py1 = a1 + t1; 360 *py2 = one + t2; 361 break; 362 363 case 5: 364 j1 = (xsb1 + 0x4000) & 0xffff8000; 365 HI(&t1) = j1; 366 LO(&t1) = 0; 367 x1 -= t1; 368 z0 = x0 * x0; 369 z1 = x1 * x1; 370 z2 = x2 * x2; 371 t0 = z0 * (poly3[1] + z0 * poly4[1]); 372 t1 = z1 * (qq1 + z1 * qq2); 373 t2 = z2 * (poly3[1] + z2 * poly4[1]); 374 t0 = z0 * (poly1[1] + z0 * (poly2[1] + t0)); 375 w1 = x1 * (one + z1 * (pp1 + z1 * pp2)); 376 t2 = z2 * (poly1[1] + z2 * (poly2[1] + t2)); 377 j1 = (((j1 & ~0x80000000) - 0x3fc40000) >> 13) & ~0x3; 378 xsb1 = (xsb1 >> 30) & 2; 379 a1 = __vlibm_TBL_sincos_hi[j1+1]; 380 t1 = __vlibm_TBL_sincos_lo[j1+1] - (__vlibm_TBL_sincos_hi[j1+xsb1]*w1 - a1*t1); 381 *py0 = one + t0; 382 *py1 = a1 + t1; 383 *py2 = one + t2; 384 break; 385 386 case 6: 387 j0 = (xsb0 + 0x4000) & 0xffff8000; 388 HI(&t0) = j0; 389 LO(&t0) = 0; 390 x0 -= t0; 391 z0 = x0 * x0; 392 z1 = x1 * x1; 393 z2 = x2 * x2; 394 t0 = z0 * (qq1 + z0 * qq2); 395 t1 = z1 * (poly3[1] + z1 * poly4[1]); 396 t2 = z2 * (poly3[1] + z2 * poly4[1]); 397 w0 = x0 * (one + z0 * (pp1 + z0 * pp2)); 398 t1 = z1 * (poly1[1] + z1 * (poly2[1] + t1)); 399 t2 = z2 * (poly1[1] + z2 * (poly2[1] + t2)); 400 j0 = (((j0 & ~0x80000000) - 0x3fc40000) >> 13) & ~0x3; 401 xsb0 = (xsb0 >> 30) & 2; 402 a0 = __vlibm_TBL_sincos_hi[j0+1]; 403 t0 = __vlibm_TBL_sincos_lo[j0+1] - (__vlibm_TBL_sincos_hi[j0+xsb0]*w0 - a0*t0); 404 *py0 = a0 + t0; 405 *py1 = one + t1; 406 *py2 = one + t2; 407 break; 408 409 case 7: /* All are < 5/32 */ 410 z0 = x0 * x0; 411 z1 = x1 * x1; 412 z2 = x2 * x2; 413 t0 = z0 * (poly3[1] + z0 * poly4[1]); 414 t1 = z1 * (poly3[1] + z1 * poly4[1]); 415 t2 = z2 * (poly3[1] + z2 * poly4[1]); 416 t0 = z0 * (poly1[1] + z0 * (poly2[1] + t0)); 417 t1 = z1 * (poly1[1] + z1 * (poly2[1] + t1)); 418 t2 = z2 * (poly1[1] + z2 * (poly2[1] + t2)); 419 *py0 = one + t0; 420 *py1 = one + t1; 421 *py2 = one + t2; 422 break; 423 } 424 425 x += stridex; 426 y += stridey; 427 i = 0; 428 } while (--n > 0); /* END MAIN LOOP */ 429 430 /* 431 * CLEAN UP last 0, 1, or 2 elts. 432 */ 433 if (i > 0) /* Clean up elts at tail. i < 3. */ 434 { 435 double a0, a1, w0, w1; 436 double t0, t1, z0, z1; 437 unsigned j0, j1; 438 439 if (i > 1) 440 { 441 if (hx1 < 0x3fc40000) 442 { 443 z1 = x1 * x1; 444 t1 = z1 * (poly3[1] + z1 * poly4[1]); 445 t1 = z1 * (poly1[1] + z1 * (poly2[1] + t1)); 446 t1 = one + t1; 447 *py1 = t1; 448 } 449 else 450 { 451 j1 = (xsb1 + 0x4000) & 0xffff8000; 452 HI(&t1) = j1; 453 LO(&t1) = 0; 454 x1 -= t1; 455 z1 = x1 * x1; 456 t1 = z1 * (qq1 + z1 * qq2); 457 w1 = x1 * (one + z1 * (pp1 + z1 * pp2)); 458 j1 = (((j1 & ~0x80000000) - 0x3fc40000) >> 13) & ~0x3; 459 xsb1 = (xsb1 >> 30) & 2; 460 a1 = __vlibm_TBL_sincos_hi[j1+1]; 461 t1 = __vlibm_TBL_sincos_lo[j1+1] 462 - (__vlibm_TBL_sincos_hi[j1+xsb1]*w1 - a1*t1); 463 *py1 = a1 + t1; 464 } 465 } 466 if (hx0 < 0x3fc40000) 467 { 468 z0 = x0 * x0; 469 t0 = z0 * (poly3[1] + z0 * poly4[1]); 470 t0 = z0 * (poly1[1] + z0 * (poly2[1] + t0)); 471 t0 = one + t0; 472 *py0 = t0; 473 } 474 else 475 { 476 j0 = (xsb0 + 0x4000) & 0xffff8000; 477 HI(&t0) = j0; 478 LO(&t0) = 0; 479 x0 -= t0; 480 z0 = x0 * x0; 481 t0 = z0 * (qq1 + z0 * qq2); 482 w0 = x0 * (one + z0 * (pp1 + z0 * pp2)); 483 j0 = (((j0 & ~0x80000000) - 0x3fc40000) >> 13) & ~0x3; 484 xsb0 = (xsb0 >> 30) & 2; 485 a0 = __vlibm_TBL_sincos_hi[j0+1]; 486 t0 = __vlibm_TBL_sincos_lo[j0+1] - (__vlibm_TBL_sincos_hi[j0+xsb0]*w0 - a0*t0); 487 *py0 = a0 + t0; 488 } 489 } /* END CLEAN UP */ 490 491 return; 492 493 /* 494 * Take care of BIGUNS. 495 * 496 * We have jumped here in the middle of processing after having 497 * encountered a medium range argument. Therefore things are in a 498 * bit of a tizzy. 499 */ 500 501 MEDIUM: 502 503 x0_or_one[1] = 1.0; 504 x1_or_one[1] = 1.0; 505 x2_or_one[1] = 1.0; 506 x0_or_one[3] = -1.0; 507 x1_or_one[3] = -1.0; 508 x2_or_one[3] = -1.0; 509 y0_or_zero[1] = 0.0; 510 y1_or_zero[1] = 0.0; 511 y2_or_zero[1] = 0.0; 512 y0_or_zero[3] = 0.0; 513 y1_or_zero[3] = 0.0; 514 y2_or_zero[3] = 0.0; 515 516 if (biguns == 3) 517 { 518 biguns = 0; 519 xsb0 = xsb0 >> 31; 520 xsb1 = xsb1 >> 31; 521 goto loop2; 522 } 523 else if (biguns == 2) 524 { 525 xsb0 = xsb0 >> 31; 526 biguns = 0; 527 goto loop1; 528 } 529 biguns = 0; 530 531 do 532 { 533 double fn0, fn1, fn2, a0, a1, a2, w0, w1, w2, y0, y1, y2; 534 unsigned hx; 535 int n0, n1, n2; 536 537 /* 538 * Find 3 more to work on: Not already done, not too big. 539 */ 540 541 loop0: 542 hx = HI(x); 543 xsb0 = hx >> 31; 544 hx &= ~0x80000000; 545 if (hx > 0x413921fb) /* (1.6471e+06) Too big: leave it. */ 546 { 547 if (hx >= 0x7ff00000) /* Inf or NaN */ 548 { 549 x0 = *x; 550 *y = x0 - x0; 551 } 552 else 553 biguns = 1; 554 x += stridex; 555 y += stridey; 556 i = 0; 557 if (--n <= 0) 558 break; 559 goto loop0; 560 } 561 x0 = *x; 562 py0 = y; 563 x += stridex; 564 y += stridey; 565 i = 1; 566 if (--n <= 0) 567 break; 568 569 loop1: 570 hx = HI(x); 571 xsb1 = hx >> 31; 572 hx &= ~0x80000000; 573 if (hx > 0x413921fb) 574 { 575 if (hx >= 0x7ff00000) 576 { 577 x1 = *x; 578 *y = x1 - x1; 579 } 580 else 581 biguns = 1; 582 x += stridex; 583 y += stridey; 584 i = 1; 585 if (--n <= 0) 586 break; 587 goto loop1; 588 } 589 x1 = *x; 590 py1 = y; 591 x += stridex; 592 y += stridey; 593 i = 2; 594 if (--n <= 0) 595 break; 596 597 loop2: 598 hx = HI(x); 599 xsb2 = hx >> 31; 600 hx &= ~0x80000000; 601 if (hx > 0x413921fb) 602 { 603 if (hx >= 0x7ff00000) 604 { 605 x2 = *x; 606 *y = x2 - x2; 607 } 608 else 609 biguns = 1; 610 x += stridex; 611 y += stridey; 612 i = 2; 613 if (--n <= 0) 614 break; 615 goto loop2; 616 } 617 x2 = *x; 618 py2 = y; 619 620 n0 = (int) (x0 * invpio2 + half[xsb0]); 621 n1 = (int) (x1 * invpio2 + half[xsb1]); 622 n2 = (int) (x2 * invpio2 + half[xsb2]); 623 fn0 = (double) n0; 624 fn1 = (double) n1; 625 fn2 = (double) n2; 626 n0 = (n0 + 1) & 3; /* Add 1 (before the mod) to make sin into cos */ 627 n1 = (n1 + 1) & 3; 628 n2 = (n2 + 1) & 3; 629 a0 = x0 - fn0 * pio2_1; 630 a1 = x1 - fn1 * pio2_1; 631 a2 = x2 - fn2 * pio2_1; 632 w0 = fn0 * pio2_2; 633 w1 = fn1 * pio2_2; 634 w2 = fn2 * pio2_2; 635 x0 = a0 - w0; 636 x1 = a1 - w1; 637 x2 = a2 - w2; 638 y0 = (a0 - x0) - w0; 639 y1 = (a1 - x1) - w1; 640 y2 = (a2 - x2) - w2; 641 a0 = x0; 642 a1 = x1; 643 a2 = x2; 644 w0 = fn0 * pio2_3 - y0; 645 w1 = fn1 * pio2_3 - y1; 646 w2 = fn2 * pio2_3 - y2; 647 x0 = a0 - w0; 648 x1 = a1 - w1; 649 x2 = a2 - w2; 650 y0 = (a0 - x0) - w0; 651 y1 = (a1 - x1) - w1; 652 y2 = (a2 - x2) - w2; 653 a0 = x0; 654 a1 = x1; 655 a2 = x2; 656 w0 = fn0 * pio2_3t - y0; 657 w1 = fn1 * pio2_3t - y1; 658 w2 = fn2 * pio2_3t - y2; 659 x0 = a0 - w0; 660 x1 = a1 - w1; 661 x2 = a2 - w2; 662 y0 = (a0 - x0) - w0; 663 y1 = (a1 - x1) - w1; 664 y2 = (a2 - x2) - w2; 665 xsb0 = HI(&x0); 666 i = ((xsb0 & ~0x80000000) - thresh[n0&1]) >> 31; 667 xsb1 = HI(&x1); 668 i |= (((xsb1 & ~0x80000000) - thresh[n1&1]) >> 30) & 2; 669 xsb2 = HI(&x2); 670 i |= (((xsb2 & ~0x80000000) - thresh[n2&1]) >> 29) & 4; 671 switch (i) 672 { 673 double t0, t1, t2, z0, z1, z2; 674 unsigned j0, j1, j2; 675 676 case 0: 677 j0 = (xsb0 + 0x4000) & 0xffff8000; 678 j1 = (xsb1 + 0x4000) & 0xffff8000; 679 j2 = (xsb2 + 0x4000) & 0xffff8000; 680 HI(&t0) = j0; 681 HI(&t1) = j1; 682 HI(&t2) = j2; 683 LO(&t0) = 0; 684 LO(&t1) = 0; 685 LO(&t2) = 0; 686 x0 = (x0 - t0) + y0; 687 x1 = (x1 - t1) + y1; 688 x2 = (x2 - t2) + y2; 689 z0 = x0 * x0; 690 z1 = x1 * x1; 691 z2 = x2 * x2; 692 t0 = z0 * (qq1 + z0 * qq2); 693 t1 = z1 * (qq1 + z1 * qq2); 694 t2 = z2 * (qq1 + z2 * qq2); 695 w0 = x0 * (one + z0 * (pp1 + z0 * pp2)); 696 w1 = x1 * (one + z1 * (pp1 + z1 * pp2)); 697 w2 = x2 * (one + z2 * (pp1 + z2 * pp2)); 698 j0 = (((j0 & ~0x80000000) - 0x3fc40000) >> 13) & ~0x3; 699 j1 = (((j1 & ~0x80000000) - 0x3fc40000) >> 13) & ~0x3; 700 j2 = (((j2 & ~0x80000000) - 0x3fc40000) >> 13) & ~0x3; 701 xsb0 = (xsb0 >> 30) & 2; 702 xsb1 = (xsb1 >> 30) & 2; 703 xsb2 = (xsb2 >> 30) & 2; 704 n0 ^= (xsb0 & ~(n0 << 1)); 705 n1 ^= (xsb1 & ~(n1 << 1)); 706 n2 ^= (xsb2 & ~(n2 << 1)); 707 xsb0 |= 1; 708 xsb1 |= 1; 709 xsb2 |= 1; 710 a0 = __vlibm_TBL_sincos_hi[j0+n0]; 711 a1 = __vlibm_TBL_sincos_hi[j1+n1]; 712 a2 = __vlibm_TBL_sincos_hi[j2+n2]; 713 t0 = (__vlibm_TBL_sincos_hi[j0+((n0+xsb0)&3)] * w0 + a0 * t0) + __vlibm_TBL_sincos_lo[j0+n0]; 714 t1 = (__vlibm_TBL_sincos_hi[j1+((n1+xsb1)&3)] * w1 + a1 * t1) + __vlibm_TBL_sincos_lo[j1+n1]; 715 t2 = (__vlibm_TBL_sincos_hi[j2+((n2+xsb2)&3)] * w2 + a2 * t2) + __vlibm_TBL_sincos_lo[j2+n2]; 716 *py0 = ( a0 + t0 ); 717 *py1 = ( a1 + t1 ); 718 *py2 = ( a2 + t2 ); 719 break; 720 721 case 1: 722 j0 = n0 & 1; 723 j1 = (xsb1 + 0x4000) & 0xffff8000; 724 j2 = (xsb2 + 0x4000) & 0xffff8000; 725 HI(&t1) = j1; 726 HI(&t2) = j2; 727 LO(&t1) = 0; 728 LO(&t2) = 0; 729 x0_or_one[0] = x0; 730 x0_or_one[2] = -x0; 731 y0_or_zero[0] = y0; 732 y0_or_zero[2] = -y0; 733 x1 = (x1 - t1) + y1; 734 x2 = (x2 - t2) + y2; 735 z0 = x0 * x0; 736 z1 = x1 * x1; 737 z2 = x2 * x2; 738 t0 = z0 * (poly3[j0] + z0 * poly4[j0]); 739 t1 = z1 * (qq1 + z1 * qq2); 740 t2 = z2 * (qq1 + z2 * qq2); 741 t0 = z0 * (poly1[j0] + z0 * (poly2[j0] + t0)); 742 w1 = x1 * (one + z1 * (pp1 + z1 * pp2)); 743 w2 = x2 * (one + z2 * (pp1 + z2 * pp2)); 744 j1 = (((j1 & ~0x80000000) - 0x3fc40000) >> 13) & ~0x3; 745 j2 = (((j2 & ~0x80000000) - 0x3fc40000) >> 13) & ~0x3; 746 xsb1 = (xsb1 >> 30) & 2; 747 xsb2 = (xsb2 >> 30) & 2; 748 n1 ^= (xsb1 & ~(n1 << 1)); 749 n2 ^= (xsb2 & ~(n2 << 1)); 750 xsb1 |= 1; 751 xsb2 |= 1; 752 a1 = __vlibm_TBL_sincos_hi[j1+n1]; 753 a2 = __vlibm_TBL_sincos_hi[j2+n2]; 754 t0 = x0_or_one[n0] + (y0_or_zero[n0] + x0_or_one[n0] * t0); 755 t1 = (__vlibm_TBL_sincos_hi[j1+((n1+xsb1)&3)] * w1 + a1 * t1) + __vlibm_TBL_sincos_lo[j1+n1]; 756 t2 = (__vlibm_TBL_sincos_hi[j2+((n2+xsb2)&3)] * w2 + a2 * t2) + __vlibm_TBL_sincos_lo[j2+n2]; 757 *py0 = t0; 758 *py1 = ( a1 + t1 ); 759 *py2 = ( a2 + t2 ); 760 break; 761 762 case 2: 763 j0 = (xsb0 + 0x4000) & 0xffff8000; 764 j1 = n1 & 1; 765 j2 = (xsb2 + 0x4000) & 0xffff8000; 766 HI(&t0) = j0; 767 HI(&t2) = j2; 768 LO(&t0) = 0; 769 LO(&t2) = 0; 770 x1_or_one[0] = x1; 771 x1_or_one[2] = -x1; 772 x0 = (x0 - t0) + y0; 773 y1_or_zero[0] = y1; 774 y1_or_zero[2] = -y1; 775 x2 = (x2 - t2) + y2; 776 z0 = x0 * x0; 777 z1 = x1 * x1; 778 z2 = x2 * x2; 779 t0 = z0 * (qq1 + z0 * qq2); 780 t1 = z1 * (poly3[j1] + z1 * poly4[j1]); 781 t2 = z2 * (qq1 + z2 * qq2); 782 w0 = x0 * (one + z0 * (pp1 + z0 * pp2)); 783 t1 = z1 * (poly1[j1] + z1 * (poly2[j1] + t1)); 784 w2 = x2 * (one + z2 * (pp1 + z2 * pp2)); 785 j0 = (((j0 & ~0x80000000) - 0x3fc40000) >> 13) & ~0x3; 786 j2 = (((j2 & ~0x80000000) - 0x3fc40000) >> 13) & ~0x3; 787 xsb0 = (xsb0 >> 30) & 2; 788 xsb2 = (xsb2 >> 30) & 2; 789 n0 ^= (xsb0 & ~(n0 << 1)); 790 n2 ^= (xsb2 & ~(n2 << 1)); 791 xsb0 |= 1; 792 xsb2 |= 1; 793 a0 = __vlibm_TBL_sincos_hi[j0+n0]; 794 a2 = __vlibm_TBL_sincos_hi[j2+n2]; 795 t0 = (__vlibm_TBL_sincos_hi[j0+((n0+xsb0)&3)] * w0 + a0 * t0) + __vlibm_TBL_sincos_lo[j0+n0]; 796 t1 = x1_or_one[n1] + (y1_or_zero[n1] + x1_or_one[n1] * t1); 797 t2 = (__vlibm_TBL_sincos_hi[j2+((n2+xsb2)&3)] * w2 + a2 * t2) + __vlibm_TBL_sincos_lo[j2+n2]; 798 *py0 = ( a0 + t0 ); 799 *py1 = t1; 800 *py2 = ( a2 + t2 ); 801 break; 802 803 case 3: 804 j0 = n0 & 1; 805 j1 = n1 & 1; 806 j2 = (xsb2 + 0x4000) & 0xffff8000; 807 HI(&t2) = j2; 808 LO(&t2) = 0; 809 x0_or_one[0] = x0; 810 x0_or_one[2] = -x0; 811 x1_or_one[0] = x1; 812 x1_or_one[2] = -x1; 813 y0_or_zero[0] = y0; 814 y0_or_zero[2] = -y0; 815 y1_or_zero[0] = y1; 816 y1_or_zero[2] = -y1; 817 x2 = (x2 - t2) + y2; 818 z0 = x0 * x0; 819 z1 = x1 * x1; 820 z2 = x2 * x2; 821 t0 = z0 * (poly3[j0] + z0 * poly4[j0]); 822 t1 = z1 * (poly3[j1] + z1 * poly4[j1]); 823 t2 = z2 * (qq1 + z2 * qq2); 824 t0 = z0 * (poly1[j0] + z0 * (poly2[j0] + t0)); 825 t1 = z1 * (poly1[j1] + z1 * (poly2[j1] + t1)); 826 w2 = x2 * (one + z2 * (pp1 + z2 * pp2)); 827 j2 = (((j2 & ~0x80000000) - 0x3fc40000) >> 13) & ~0x3; 828 xsb2 = (xsb2 >> 30) & 2; 829 n2 ^= (xsb2 & ~(n2 << 1)); 830 xsb2 |= 1; 831 a2 = __vlibm_TBL_sincos_hi[j2+n2]; 832 t0 = x0_or_one[n0] + (y0_or_zero[n0] + x0_or_one[n0] * t0); 833 t1 = x1_or_one[n1] + (y1_or_zero[n1] + x1_or_one[n1] * t1); 834 t2 = (__vlibm_TBL_sincos_hi[j2+((n2+xsb2)&3)] * w2 + a2 * t2) + __vlibm_TBL_sincos_lo[j2+n2]; 835 *py0 = t0; 836 *py1 = t1; 837 *py2 = ( a2 + t2 ); 838 break; 839 840 case 4: 841 j0 = (xsb0 + 0x4000) & 0xffff8000; 842 j1 = (xsb1 + 0x4000) & 0xffff8000; 843 j2 = n2 & 1; 844 HI(&t0) = j0; 845 HI(&t1) = j1; 846 LO(&t0) = 0; 847 LO(&t1) = 0; 848 x2_or_one[0] = x2; 849 x2_or_one[2] = -x2; 850 x0 = (x0 - t0) + y0; 851 x1 = (x1 - t1) + y1; 852 y2_or_zero[0] = y2; 853 y2_or_zero[2] = -y2; 854 z0 = x0 * x0; 855 z1 = x1 * x1; 856 z2 = x2 * x2; 857 t0 = z0 * (qq1 + z0 * qq2); 858 t1 = z1 * (qq1 + z1 * qq2); 859 t2 = z2 * (poly3[j2] + z2 * poly4[j2]); 860 w0 = x0 * (one + z0 * (pp1 + z0 * pp2)); 861 w1 = x1 * (one + z1 * (pp1 + z1 * pp2)); 862 t2 = z2 * (poly1[j2] + z2 * (poly2[j2] + t2)); 863 j0 = (((j0 & ~0x80000000) - 0x3fc40000) >> 13) & ~0x3; 864 j1 = (((j1 & ~0x80000000) - 0x3fc40000) >> 13) & ~0x3; 865 xsb0 = (xsb0 >> 30) & 2; 866 xsb1 = (xsb1 >> 30) & 2; 867 n0 ^= (xsb0 & ~(n0 << 1)); 868 n1 ^= (xsb1 & ~(n1 << 1)); 869 xsb0 |= 1; 870 xsb1 |= 1; 871 a0 = __vlibm_TBL_sincos_hi[j0+n0]; 872 a1 = __vlibm_TBL_sincos_hi[j1+n1]; 873 t0 = (__vlibm_TBL_sincos_hi[j0+((n0+xsb0)&3)] * w0 + a0 * t0) + __vlibm_TBL_sincos_lo[j0+n0]; 874 t1 = (__vlibm_TBL_sincos_hi[j1+((n1+xsb1)&3)] * w1 + a1 * t1) + __vlibm_TBL_sincos_lo[j1+n1]; 875 t2 = x2_or_one[n2] + (y2_or_zero[n2] + x2_or_one[n2] * t2); 876 *py0 = ( a0 + t0 ); 877 *py1 = ( a1 + t1 ); 878 *py2 = t2; 879 break; 880 881 case 5: 882 j0 = n0 & 1; 883 j1 = (xsb1 + 0x4000) & 0xffff8000; 884 j2 = n2 & 1; 885 HI(&t1) = j1; 886 LO(&t1) = 0; 887 x0_or_one[0] = x0; 888 x0_or_one[2] = -x0; 889 x2_or_one[0] = x2; 890 x2_or_one[2] = -x2; 891 y0_or_zero[0] = y0; 892 y0_or_zero[2] = -y0; 893 x1 = (x1 - t1) + y1; 894 y2_or_zero[0] = y2; 895 y2_or_zero[2] = -y2; 896 z0 = x0 * x0; 897 z1 = x1 * x1; 898 z2 = x2 * x2; 899 t0 = z0 * (poly3[j0] + z0 * poly4[j0]); 900 t1 = z1 * (qq1 + z1 * qq2); 901 t2 = z2 * (poly3[j2] + z2 * poly4[j2]); 902 t0 = z0 * (poly1[j0] + z0 * (poly2[j0] + t0)); 903 w1 = x1 * (one + z1 * (pp1 + z1 * pp2)); 904 t2 = z2 * (poly1[j2] + z2 * (poly2[j2] + t2)); 905 j1 = (((j1 & ~0x80000000) - 0x3fc40000) >> 13) & ~0x3; 906 xsb1 = (xsb1 >> 30) & 2; 907 n1 ^= (xsb1 & ~(n1 << 1)); 908 xsb1 |= 1; 909 a1 = __vlibm_TBL_sincos_hi[j1+n1]; 910 t0 = x0_or_one[n0] + (y0_or_zero[n0] + x0_or_one[n0] * t0); 911 t1 = (__vlibm_TBL_sincos_hi[j1+((n1+xsb1)&3)] * w1 + a1 * t1) + __vlibm_TBL_sincos_lo[j1+n1]; 912 t2 = x2_or_one[n2] + (y2_or_zero[n2] + x2_or_one[n2] * t2); 913 *py0 = t0; 914 *py1 = ( a1 + t1 ); 915 *py2 = t2; 916 break; 917 918 case 6: 919 j0 = (xsb0 + 0x4000) & 0xffff8000; 920 j1 = n1 & 1; 921 j2 = n2 & 1; 922 HI(&t0) = j0; 923 LO(&t0) = 0; 924 x1_or_one[0] = x1; 925 x1_or_one[2] = -x1; 926 x2_or_one[0] = x2; 927 x2_or_one[2] = -x2; 928 x0 = (x0 - t0) + y0; 929 y1_or_zero[0] = y1; 930 y1_or_zero[2] = -y1; 931 y2_or_zero[0] = y2; 932 y2_or_zero[2] = -y2; 933 z0 = x0 * x0; 934 z1 = x1 * x1; 935 z2 = x2 * x2; 936 t0 = z0 * (qq1 + z0 * qq2); 937 t1 = z1 * (poly3[j1] + z1 * poly4[j1]); 938 t2 = z2 * (poly3[j2] + z2 * poly4[j2]); 939 w0 = x0 * (one + z0 * (pp1 + z0 * pp2)); 940 t1 = z1 * (poly1[j1] + z1 * (poly2[j1] + t1)); 941 t2 = z2 * (poly1[j2] + z2 * (poly2[j2] + t2)); 942 j0 = (((j0 & ~0x80000000) - 0x3fc40000) >> 13) & ~0x3; 943 xsb0 = (xsb0 >> 30) & 2; 944 n0 ^= (xsb0 & ~(n0 << 1)); 945 xsb0 |= 1; 946 a0 = __vlibm_TBL_sincos_hi[j0+n0]; 947 t0 = (__vlibm_TBL_sincos_hi[j0+((n0+xsb0)&3)] * w0 + a0 * t0) + __vlibm_TBL_sincos_lo[j0+n0]; 948 t1 = x1_or_one[n1] + (y1_or_zero[n1] + x1_or_one[n1] * t1); 949 t2 = x2_or_one[n2] + (y2_or_zero[n2] + x2_or_one[n2] * t2); 950 *py0 = ( a0 + t0 ); 951 *py1 = t1; 952 *py2 = t2; 953 break; 954 955 case 7: 956 j0 = n0 & 1; 957 j1 = n1 & 1; 958 j2 = n2 & 1; 959 x0_or_one[0] = x0; 960 x0_or_one[2] = -x0; 961 x1_or_one[0] = x1; 962 x1_or_one[2] = -x1; 963 x2_or_one[0] = x2; 964 x2_or_one[2] = -x2; 965 y0_or_zero[0] = y0; 966 y0_or_zero[2] = -y0; 967 y1_or_zero[0] = y1; 968 y1_or_zero[2] = -y1; 969 y2_or_zero[0] = y2; 970 y2_or_zero[2] = -y2; 971 z0 = x0 * x0; 972 z1 = x1 * x1; 973 z2 = x2 * x2; 974 t0 = z0 * (poly3[j0] + z0 * poly4[j0]); 975 t1 = z1 * (poly3[j1] + z1 * poly4[j1]); 976 t2 = z2 * (poly3[j2] + z2 * poly4[j2]); 977 t0 = z0 * (poly1[j0] + z0 * (poly2[j0] + t0)); 978 t1 = z1 * (poly1[j1] + z1 * (poly2[j1] + t1)); 979 t2 = z2 * (poly1[j2] + z2 * (poly2[j2] + t2)); 980 t0 = x0_or_one[n0] + (y0_or_zero[n0] + x0_or_one[n0] * t0); 981 t1 = x1_or_one[n1] + (y1_or_zero[n1] + x1_or_one[n1] * t1); 982 t2 = x2_or_one[n2] + (y2_or_zero[n2] + x2_or_one[n2] * t2); 983 *py0 = t0; 984 *py1 = t1; 985 *py2 = t2; 986 break; 987 } 988 989 x += stridex; 990 y += stridey; 991 i = 0; 992 } while (--n > 0); 993 994 if (i > 0) 995 { 996 double fn0, fn1, a0, a1, w0, w1, y0, y1; 997 double t0, t1, z0, z1; 998 unsigned j0, j1; 999 int n0, n1; 1000 1001 if (i > 1) 1002 { 1003 n1 = (int) (x1 * invpio2 + half[xsb1]); 1004 fn1 = (double) n1; 1005 n1 = (n1 + 1) & 3; /* Add 1 (before the mod) to make sin into cos */ 1006 a1 = x1 - fn1 * pio2_1; 1007 w1 = fn1 * pio2_2; 1008 x1 = a1 - w1; 1009 y1 = (a1 - x1) - w1; 1010 a1 = x1; 1011 w1 = fn1 * pio2_3 - y1; 1012 x1 = a1 - w1; 1013 y1 = (a1 - x1) - w1; 1014 a1 = x1; 1015 w1 = fn1 * pio2_3t - y1; 1016 x1 = a1 - w1; 1017 y1 = (a1 - x1) - w1; 1018 xsb1 = HI(&x1); 1019 if ((xsb1 & ~0x80000000) < thresh[n1&1]) 1020 { 1021 j1 = n1 & 1; 1022 x1_or_one[0] = x1; 1023 x1_or_one[2] = -x1; 1024 y1_or_zero[0] = y1; 1025 y1_or_zero[2] = -y1; 1026 z1 = x1 * x1; 1027 t1 = z1 * (poly3[j1] + z1 * poly4[j1]); 1028 t1 = z1 * (poly1[j1] + z1 * (poly2[j1] + t1)); 1029 t1 = x1_or_one[n1] + (y1_or_zero[n1] + x1_or_one[n1] * t1); 1030 *py1 = t1; 1031 } 1032 else 1033 { 1034 j1 = (xsb1 + 0x4000) & 0xffff8000; 1035 HI(&t1) = j1; 1036 LO(&t1) = 0; 1037 x1 = (x1 - t1) + y1; 1038 z1 = x1 * x1; 1039 t1 = z1 * (qq1 + z1 * qq2); 1040 w1 = x1 * (one + z1 * (pp1 + z1 * pp2)); 1041 j1 = (((j1 & ~0x80000000) - 0x3fc40000) >> 13) & ~0x3; 1042 xsb1 = (xsb1 >> 30) & 2; 1043 n1 ^= (xsb1 & ~(n1 << 1)); 1044 xsb1 |= 1; 1045 a1 = __vlibm_TBL_sincos_hi[j1+n1]; 1046 t1 = (__vlibm_TBL_sincos_hi[j1+((n1+xsb1)&3)] * w1 + a1 * t1) + __vlibm_TBL_sincos_lo[j1+n1]; 1047 *py1 = ( a1 + t1 ); 1048 } 1049 } 1050 n0 = (int) (x0 * invpio2 + half[xsb0]); 1051 fn0 = (double) n0; 1052 n0 = (n0 + 1) & 3; /* Add 1 (before the mod) to make sin into cos */ 1053 a0 = x0 - fn0 * pio2_1; 1054 w0 = fn0 * pio2_2; 1055 x0 = a0 - w0; 1056 y0 = (a0 - x0) - w0; 1057 a0 = x0; 1058 w0 = fn0 * pio2_3 - y0; 1059 x0 = a0 - w0; 1060 y0 = (a0 - x0) - w0; 1061 a0 = x0; 1062 w0 = fn0 * pio2_3t - y0; 1063 x0 = a0 - w0; 1064 y0 = (a0 - x0) - w0; 1065 xsb0 = HI(&x0); 1066 if ((xsb0 & ~0x80000000) < thresh[n0&1]) 1067 { 1068 j0 = n0 & 1; 1069 x0_or_one[0] = x0; 1070 x0_or_one[2] = -x0; 1071 y0_or_zero[0] = y0; 1072 y0_or_zero[2] = -y0; 1073 z0 = x0 * x0; 1074 t0 = z0 * (poly3[j0] + z0 * poly4[j0]); 1075 t0 = z0 * (poly1[j0] + z0 * (poly2[j0] + t0)); 1076 t0 = x0_or_one[n0] + (y0_or_zero[n0] + x0_or_one[n0] * t0); 1077 *py0 = t0; 1078 } 1079 else 1080 { 1081 j0 = (xsb0 + 0x4000) & 0xffff8000; 1082 HI(&t0) = j0; 1083 LO(&t0) = 0; 1084 x0 = (x0 - t0) + y0; 1085 z0 = x0 * x0; 1086 t0 = z0 * (qq1 + z0 * qq2); 1087 w0 = x0 * (one + z0 * (pp1 + z0 * pp2)); 1088 j0 = (((j0 & ~0x80000000) - 0x3fc40000) >> 13) & ~0x3; 1089 xsb0 = (xsb0 >> 30) & 2; 1090 n0 ^= (xsb0 & ~(n0 << 1)); 1091 xsb0 |= 1; 1092 a0 = __vlibm_TBL_sincos_hi[j0+n0]; 1093 t0 = (__vlibm_TBL_sincos_hi[j0+((n0+xsb0)&3)] * w0 + a0 * t0) + __vlibm_TBL_sincos_lo[j0+n0]; 1094 *py0 = ( a0 + t0 ); 1095 } 1096 } 1097 1098 if (biguns) 1099 __vlibm_vcos_big(nsave, xsave, sxsave, ysave, sysave, 0x413921fb); 1100 } 1101