1 /*
   2  * Copyright (c) 2017, Oracle and/or its affiliates. All rights reserved.
   3  * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
   4  *
   5  * This code is free software; you can redistribute it and/or modify it
   6  * under the terms of the GNU General Public License version 2 only, as
   7  * published by the Free Software Foundation.  Oracle designates this
   8  * particular file as subject to the "Classpath" exception as provided
   9  * by Oracle in the LICENSE file that accompanied this code.
  10  *
  11  * This code is distributed in the hope that it will be useful, but WITHOUT
  12  * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
  13  * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
  14  * version 2 for more details (a copy is included in the LICENSE file that
  15  * accompanied this code).
  16  *
  17  * You should have received a copy of the GNU General Public License version
  18  * 2 along with this work; if not, write to the Free Software Foundation,
  19  * Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA.
  20  *
  21  * Please contact Oracle, 500 Oracle Parkway, Redwood Shores, CA 94065 USA
  22  * or visit www.oracle.com if you need additional information or have
  23  * questions.
  24  */
  25 package jdk.incubator.vector;
  26 
  27 import java.nio.ByteBuffer;
  28 import java.nio.ByteOrder;
  29 import java.nio.ReadOnlyBufferException;
  30 import java.util.Arrays;
  31 import java.util.Objects;
  32 import java.util.function.IntUnaryOperator;
  33 
  34 import jdk.internal.misc.Unsafe;
  35 import jdk.internal.vm.annotation.ForceInline;
  36 import static jdk.incubator.vector.VectorIntrinsics.*;
  37 
  38 @SuppressWarnings("cast")
  39 final class Byte512Vector extends ByteVector {
  40     private static final VectorSpecies<Byte> SPECIES = ByteVector.SPECIES_512;
  41 
  42     static final Byte512Vector ZERO = new Byte512Vector();
  43 
  44     static final int LENGTH = SPECIES.length();
  45 
  46     private final byte[] vec; // Don't access directly, use getElements() instead.
  47 
  48     private byte[] getElements() {
  49         return VectorIntrinsics.maybeRebox(this).vec;
  50     }
  51 
  52     Byte512Vector() {
  53         vec = new byte[SPECIES.length()];
  54     }
  55 
  56     Byte512Vector(byte[] v) {
  57         vec = v;
  58     }
  59 
  60     @Override
  61     public int length() { return LENGTH; }
  62 
  63     // Unary operator
  64 
  65     @Override
  66     Byte512Vector uOp(FUnOp f) {
  67         byte[] vec = getElements();
  68         byte[] res = new byte[length()];
  69         for (int i = 0; i < length(); i++) {
  70             res[i] = f.apply(i, vec[i]);
  71         }
  72         return new Byte512Vector(res);
  73     }
  74 
  75     @Override
  76     Byte512Vector uOp(VectorMask<Byte> o, FUnOp f) {
  77         byte[] vec = getElements();
  78         byte[] res = new byte[length()];
  79         boolean[] mbits = ((Byte512Mask)o).getBits();
  80         for (int i = 0; i < length(); i++) {
  81             res[i] = mbits[i] ? f.apply(i, vec[i]) : vec[i];
  82         }
  83         return new Byte512Vector(res);
  84     }
  85 
  86     // Binary operator
  87 
  88     @Override
  89     Byte512Vector bOp(Vector<Byte> o, FBinOp f) {
  90         byte[] res = new byte[length()];
  91         byte[] vec1 = this.getElements();
  92         byte[] vec2 = ((Byte512Vector)o).getElements();
  93         for (int i = 0; i < length(); i++) {
  94             res[i] = f.apply(i, vec1[i], vec2[i]);
  95         }
  96         return new Byte512Vector(res);
  97     }
  98 
  99     @Override
 100     Byte512Vector bOp(Vector<Byte> o1, VectorMask<Byte> o2, FBinOp f) {
 101         byte[] res = new byte[length()];
 102         byte[] vec1 = this.getElements();
 103         byte[] vec2 = ((Byte512Vector)o1).getElements();
 104         boolean[] mbits = ((Byte512Mask)o2).getBits();
 105         for (int i = 0; i < length(); i++) {
 106             res[i] = mbits[i] ? f.apply(i, vec1[i], vec2[i]) : vec1[i];
 107         }
 108         return new Byte512Vector(res);
 109     }
 110 
 111     // Trinary operator
 112 
 113     @Override
 114     Byte512Vector tOp(Vector<Byte> o1, Vector<Byte> o2, FTriOp f) {
 115         byte[] res = new byte[length()];
 116         byte[] vec1 = this.getElements();
 117         byte[] vec2 = ((Byte512Vector)o1).getElements();
 118         byte[] vec3 = ((Byte512Vector)o2).getElements();
 119         for (int i = 0; i < length(); i++) {
 120             res[i] = f.apply(i, vec1[i], vec2[i], vec3[i]);
 121         }
 122         return new Byte512Vector(res);
 123     }
 124 
 125     @Override
 126     Byte512Vector tOp(Vector<Byte> o1, Vector<Byte> o2, VectorMask<Byte> o3, FTriOp f) {
 127         byte[] res = new byte[length()];
 128         byte[] vec1 = getElements();
 129         byte[] vec2 = ((Byte512Vector)o1).getElements();
 130         byte[] vec3 = ((Byte512Vector)o2).getElements();
 131         boolean[] mbits = ((Byte512Mask)o3).getBits();
 132         for (int i = 0; i < length(); i++) {
 133             res[i] = mbits[i] ? f.apply(i, vec1[i], vec2[i], vec3[i]) : vec1[i];
 134         }
 135         return new Byte512Vector(res);
 136     }
 137 
 138     @Override
 139     byte rOp(byte v, FBinOp f) {
 140         byte[] vec = getElements();
 141         for (int i = 0; i < length(); i++) {
 142             v = f.apply(i, v, vec[i]);
 143         }
 144         return v;
 145     }
 146 
 147     @Override
 148     @ForceInline
 149     public <F> Vector<F> cast(VectorSpecies<F> s) {
 150         Objects.requireNonNull(s);
 151         if (s.length() != LENGTH)
 152             throw new IllegalArgumentException("Vector length this species length differ");
 153 
 154         return VectorIntrinsics.cast(
 155             Byte512Vector.class,
 156             byte.class, LENGTH,
 157             s.boxType(),
 158             s.elementType(), LENGTH,
 159             this, s,
 160             (species, vector) -> vector.castDefault(species)
 161         );
 162     }
 163 
 164     @SuppressWarnings("unchecked")
 165     @ForceInline
 166     private <F> Vector<F> castDefault(VectorSpecies<F> s) {
 167         int limit = s.length();
 168 
 169         Class<?> stype = s.elementType();
 170         if (stype == byte.class) {
 171             byte[] a = new byte[limit];
 172             for (int i = 0; i < limit; i++) {
 173                 a[i] = (byte) this.get(i);
 174             }
 175             return (Vector) ByteVector.fromArray((VectorSpecies<Byte>) s, a, 0);
 176         } else if (stype == short.class) {
 177             short[] a = new short[limit];
 178             for (int i = 0; i < limit; i++) {
 179                 a[i] = (short) this.get(i);
 180             }
 181             return (Vector) ShortVector.fromArray((VectorSpecies<Short>) s, a, 0);
 182         } else if (stype == int.class) {
 183             int[] a = new int[limit];
 184             for (int i = 0; i < limit; i++) {
 185                 a[i] = (int) this.get(i);
 186             }
 187             return (Vector) IntVector.fromArray((VectorSpecies<Integer>) s, a, 0);
 188         } else if (stype == long.class) {
 189             long[] a = new long[limit];
 190             for (int i = 0; i < limit; i++) {
 191                 a[i] = (long) this.get(i);
 192             }
 193             return (Vector) LongVector.fromArray((VectorSpecies<Long>) s, a, 0);
 194         } else if (stype == float.class) {
 195             float[] a = new float[limit];
 196             for (int i = 0; i < limit; i++) {
 197                 a[i] = (float) this.get(i);
 198             }
 199             return (Vector) FloatVector.fromArray((VectorSpecies<Float>) s, a, 0);
 200         } else if (stype == double.class) {
 201             double[] a = new double[limit];
 202             for (int i = 0; i < limit; i++) {
 203                 a[i] = (double) this.get(i);
 204             }
 205             return (Vector) DoubleVector.fromArray((VectorSpecies<Double>) s, a, 0);
 206         } else {
 207             throw new UnsupportedOperationException("Bad lane type for casting.");
 208         }
 209     }
 210 
 211     @Override
 212     @ForceInline
 213     @SuppressWarnings("unchecked")
 214     public <F> Vector<F> reinterpret(VectorSpecies<F> s) {
 215         Objects.requireNonNull(s);
 216 
 217         if(s.elementType().equals(byte.class)) {
 218             return (Vector<F>) reshape((VectorSpecies<Byte>)s);
 219         }
 220         if(s.bitSize() == bitSize()) {
 221             return reinterpretType(s);
 222         }
 223 
 224         return defaultReinterpret(s);
 225     }
 226 
 227     @ForceInline
 228     private <F> Vector<F> reinterpretType(VectorSpecies<F> s) {
 229         Objects.requireNonNull(s);
 230 
 231         Class<?> stype = s.elementType();
 232         if (stype == byte.class) {
 233             return VectorIntrinsics.reinterpret(
 234                 Byte512Vector.class,
 235                 byte.class, LENGTH,
 236                 Byte512Vector.class,
 237                 byte.class, Byte512Vector.LENGTH,
 238                 this, s,
 239                 (species, vector) -> vector.defaultReinterpret(species)
 240             );
 241         } else if (stype == short.class) {
 242             return VectorIntrinsics.reinterpret(
 243                 Byte512Vector.class,
 244                 byte.class, LENGTH,
 245                 Short512Vector.class,
 246                 short.class, Short512Vector.LENGTH,
 247                 this, s,
 248                 (species, vector) -> vector.defaultReinterpret(species)
 249             );
 250         } else if (stype == int.class) {
 251             return VectorIntrinsics.reinterpret(
 252                 Byte512Vector.class,
 253                 byte.class, LENGTH,
 254                 Int512Vector.class,
 255                 int.class, Int512Vector.LENGTH,
 256                 this, s,
 257                 (species, vector) -> vector.defaultReinterpret(species)
 258             );
 259         } else if (stype == long.class) {
 260             return VectorIntrinsics.reinterpret(
 261                 Byte512Vector.class,
 262                 byte.class, LENGTH,
 263                 Long512Vector.class,
 264                 long.class, Long512Vector.LENGTH,
 265                 this, s,
 266                 (species, vector) -> vector.defaultReinterpret(species)
 267             );
 268         } else if (stype == float.class) {
 269             return VectorIntrinsics.reinterpret(
 270                 Byte512Vector.class,
 271                 byte.class, LENGTH,
 272                 Float512Vector.class,
 273                 float.class, Float512Vector.LENGTH,
 274                 this, s,
 275                 (species, vector) -> vector.defaultReinterpret(species)
 276             );
 277         } else if (stype == double.class) {
 278             return VectorIntrinsics.reinterpret(
 279                 Byte512Vector.class,
 280                 byte.class, LENGTH,
 281                 Double512Vector.class,
 282                 double.class, Double512Vector.LENGTH,
 283                 this, s,
 284                 (species, vector) -> vector.defaultReinterpret(species)
 285             );
 286         } else {
 287             throw new UnsupportedOperationException("Bad lane type for casting.");
 288         }
 289     }
 290 
 291     @Override
 292     @ForceInline
 293     public ByteVector reshape(VectorSpecies<Byte> s) {
 294         Objects.requireNonNull(s);
 295         if (s.bitSize() == 64 && (s.boxType() == Byte64Vector.class)) {
 296             return VectorIntrinsics.reinterpret(
 297                 Byte512Vector.class,
 298                 byte.class, LENGTH,
 299                 Byte64Vector.class,
 300                 byte.class, Byte64Vector.LENGTH,
 301                 this, s,
 302                 (species, vector) -> (ByteVector) vector.defaultReinterpret(species)
 303             );
 304         } else if (s.bitSize() == 128 && (s.boxType() == Byte128Vector.class)) {
 305             return VectorIntrinsics.reinterpret(
 306                 Byte512Vector.class,
 307                 byte.class, LENGTH,
 308                 Byte128Vector.class,
 309                 byte.class, Byte128Vector.LENGTH,
 310                 this, s,
 311                 (species, vector) -> (ByteVector) vector.defaultReinterpret(species)
 312             );
 313         } else if (s.bitSize() == 256 && (s.boxType() == Byte256Vector.class)) {
 314             return VectorIntrinsics.reinterpret(
 315                 Byte512Vector.class,
 316                 byte.class, LENGTH,
 317                 Byte256Vector.class,
 318                 byte.class, Byte256Vector.LENGTH,
 319                 this, s,
 320                 (species, vector) -> (ByteVector) vector.defaultReinterpret(species)
 321             );
 322         } else if (s.bitSize() == 512 && (s.boxType() == Byte512Vector.class)) {
 323             return VectorIntrinsics.reinterpret(
 324                 Byte512Vector.class,
 325                 byte.class, LENGTH,
 326                 Byte512Vector.class,
 327                 byte.class, Byte512Vector.LENGTH,
 328                 this, s,
 329                 (species, vector) -> (ByteVector) vector.defaultReinterpret(species)
 330             );
 331         } else if ((s.bitSize() > 0) && (s.bitSize() <= 2048)
 332                 && (s.bitSize() % 128 == 0) && (s.boxType() == ByteMaxVector.class)) {
 333             return VectorIntrinsics.reinterpret(
 334                 Byte512Vector.class,
 335                 byte.class, LENGTH,
 336                 ByteMaxVector.class,
 337                 byte.class, ByteMaxVector.LENGTH,
 338                 this, s,
 339                 (species, vector) -> (ByteVector) vector.defaultReinterpret(species)
 340             );
 341         } else {
 342             throw new InternalError("Unimplemented size");
 343         }
 344     }
 345 
 346     // Binary operations with scalars
 347 
 348     @Override
 349     @ForceInline
 350     public ByteVector add(byte o) {
 351         return add((Byte512Vector)ByteVector.broadcast(SPECIES, o));
 352     }
 353 
 354     @Override
 355     @ForceInline
 356     public ByteVector add(byte o, VectorMask<Byte> m) {
 357         return add((Byte512Vector)ByteVector.broadcast(SPECIES, o), m);
 358     }
 359 
 360     @Override
 361     @ForceInline
 362     public ByteVector sub(byte o) {
 363         return sub((Byte512Vector)ByteVector.broadcast(SPECIES, o));
 364     }
 365 
 366     @Override
 367     @ForceInline
 368     public ByteVector sub(byte o, VectorMask<Byte> m) {
 369         return sub((Byte512Vector)ByteVector.broadcast(SPECIES, o), m);
 370     }
 371 
 372     @Override
 373     @ForceInline
 374     public ByteVector mul(byte o) {
 375         return mul((Byte512Vector)ByteVector.broadcast(SPECIES, o));
 376     }
 377 
 378     @Override
 379     @ForceInline
 380     public ByteVector mul(byte o, VectorMask<Byte> m) {
 381         return mul((Byte512Vector)ByteVector.broadcast(SPECIES, o), m);
 382     }
 383 
 384     @Override
 385     @ForceInline
 386     public ByteVector min(byte o) {
 387         return min((Byte512Vector)ByteVector.broadcast(SPECIES, o));
 388     }
 389 
 390     @Override
 391     @ForceInline
 392     public ByteVector max(byte o) {
 393         return max((Byte512Vector)ByteVector.broadcast(SPECIES, o));
 394     }
 395 
 396     @Override
 397     @ForceInline
 398     public VectorMask<Byte> equal(byte o) {
 399         return equal((Byte512Vector)ByteVector.broadcast(SPECIES, o));
 400     }
 401 
 402     @Override
 403     @ForceInline
 404     public VectorMask<Byte> notEqual(byte o) {
 405         return notEqual((Byte512Vector)ByteVector.broadcast(SPECIES, o));
 406     }
 407 
 408     @Override
 409     @ForceInline
 410     public VectorMask<Byte> lessThan(byte o) {
 411         return lessThan((Byte512Vector)ByteVector.broadcast(SPECIES, o));
 412     }
 413 
 414     @Override
 415     @ForceInline
 416     public VectorMask<Byte> lessThanEq(byte o) {
 417         return lessThanEq((Byte512Vector)ByteVector.broadcast(SPECIES, o));
 418     }
 419 
 420     @Override
 421     @ForceInline
 422     public VectorMask<Byte> greaterThan(byte o) {
 423         return greaterThan((Byte512Vector)ByteVector.broadcast(SPECIES, o));
 424     }
 425 
 426     @Override
 427     @ForceInline
 428     public VectorMask<Byte> greaterThanEq(byte o) {
 429         return greaterThanEq((Byte512Vector)ByteVector.broadcast(SPECIES, o));
 430     }
 431 
 432     @Override
 433     @ForceInline
 434     public ByteVector blend(byte o, VectorMask<Byte> m) {
 435         return blend((Byte512Vector)ByteVector.broadcast(SPECIES, o), m);
 436     }
 437 
 438 
 439     @Override
 440     @ForceInline
 441     public ByteVector and(byte o) {
 442         return and((Byte512Vector)ByteVector.broadcast(SPECIES, o));
 443     }
 444 
 445     @Override
 446     @ForceInline
 447     public ByteVector and(byte o, VectorMask<Byte> m) {
 448         return and((Byte512Vector)ByteVector.broadcast(SPECIES, o), m);
 449     }
 450 
 451     @Override
 452     @ForceInline
 453     public ByteVector or(byte o) {
 454         return or((Byte512Vector)ByteVector.broadcast(SPECIES, o));
 455     }
 456 
 457     @Override
 458     @ForceInline
 459     public ByteVector or(byte o, VectorMask<Byte> m) {
 460         return or((Byte512Vector)ByteVector.broadcast(SPECIES, o), m);
 461     }
 462 
 463     @Override
 464     @ForceInline
 465     public ByteVector xor(byte o) {
 466         return xor((Byte512Vector)ByteVector.broadcast(SPECIES, o));
 467     }
 468 
 469     @Override
 470     @ForceInline
 471     public ByteVector xor(byte o, VectorMask<Byte> m) {
 472         return xor((Byte512Vector)ByteVector.broadcast(SPECIES, o), m);
 473     }
 474 
 475     @Override
 476     @ForceInline
 477     public Byte512Vector neg() {
 478         return (Byte512Vector)zero(SPECIES).sub(this);
 479     }
 480 
 481     // Unary operations
 482 
 483     @ForceInline
 484     @Override
 485     public Byte512Vector neg(VectorMask<Byte> m) {
 486         return blend(neg(), m);
 487     }
 488 
 489     @Override
 490     @ForceInline
 491     public Byte512Vector abs() {
 492         return VectorIntrinsics.unaryOp(
 493             VECTOR_OP_ABS, Byte512Vector.class, byte.class, LENGTH,
 494             this,
 495             v1 -> v1.uOp((i, a) -> (byte) Math.abs(a)));
 496     }
 497 
 498     @ForceInline
 499     @Override
 500     public Byte512Vector abs(VectorMask<Byte> m) {
 501         return blend(abs(), m);
 502     }
 503 
 504 
 505     @Override
 506     @ForceInline
 507     public Byte512Vector not() {
 508         return VectorIntrinsics.unaryOp(
 509             VECTOR_OP_NOT, Byte512Vector.class, byte.class, LENGTH,
 510             this,
 511             v1 -> v1.uOp((i, a) -> (byte) ~a));
 512     }
 513 
 514     @ForceInline
 515     @Override
 516     public Byte512Vector not(VectorMask<Byte> m) {
 517         return blend(not(), m);
 518     }
 519     // Binary operations
 520 
 521     @Override
 522     @ForceInline
 523     public Byte512Vector add(Vector<Byte> o) {
 524         Objects.requireNonNull(o);
 525         Byte512Vector v = (Byte512Vector)o;
 526         return VectorIntrinsics.binaryOp(
 527             VECTOR_OP_ADD, Byte512Vector.class, byte.class, LENGTH,
 528             this, v,
 529             (v1, v2) -> v1.bOp(v2, (i, a, b) -> (byte)(a + b)));
 530     }
 531 
 532     @Override
 533     @ForceInline
 534     public Byte512Vector add(Vector<Byte> v, VectorMask<Byte> m) {
 535         return blend(add(v), m);
 536     }
 537 
 538     @Override
 539     @ForceInline
 540     public Byte512Vector sub(Vector<Byte> o) {
 541         Objects.requireNonNull(o);
 542         Byte512Vector v = (Byte512Vector)o;
 543         return VectorIntrinsics.binaryOp(
 544             VECTOR_OP_SUB, Byte512Vector.class, byte.class, LENGTH,
 545             this, v,
 546             (v1, v2) -> v1.bOp(v2, (i, a, b) -> (byte)(a - b)));
 547     }
 548 
 549     @Override
 550     @ForceInline
 551     public Byte512Vector sub(Vector<Byte> v, VectorMask<Byte> m) {
 552         return blend(sub(v), m);
 553     }
 554 
 555     @Override
 556     @ForceInline
 557     public Byte512Vector mul(Vector<Byte> o) {
 558         Objects.requireNonNull(o);
 559         Byte512Vector v = (Byte512Vector)o;
 560         return VectorIntrinsics.binaryOp(
 561             VECTOR_OP_MUL, Byte512Vector.class, byte.class, LENGTH,
 562             this, v,
 563             (v1, v2) -> v1.bOp(v2, (i, a, b) -> (byte)(a * b)));
 564     }
 565 
 566     @Override
 567     @ForceInline
 568     public Byte512Vector mul(Vector<Byte> v, VectorMask<Byte> m) {
 569         return blend(mul(v), m);
 570     }
 571 
 572     @Override
 573     @ForceInline
 574     public Byte512Vector min(Vector<Byte> o) {
 575         Objects.requireNonNull(o);
 576         Byte512Vector v = (Byte512Vector)o;
 577         return (Byte512Vector) VectorIntrinsics.binaryOp(
 578             VECTOR_OP_MIN, Byte512Vector.class, byte.class, LENGTH,
 579             this, v,
 580             (v1, v2) -> v1.bOp(v2, (i, a, b) -> (byte) Math.min(a, b)));
 581     }
 582 
 583     @Override
 584     @ForceInline
 585     public Byte512Vector min(Vector<Byte> v, VectorMask<Byte> m) {
 586         return blend(min(v), m);
 587     }
 588 
 589     @Override
 590     @ForceInline
 591     public Byte512Vector max(Vector<Byte> o) {
 592         Objects.requireNonNull(o);
 593         Byte512Vector v = (Byte512Vector)o;
 594         return VectorIntrinsics.binaryOp(
 595             VECTOR_OP_MAX, Byte512Vector.class, byte.class, LENGTH,
 596             this, v,
 597             (v1, v2) -> v1.bOp(v2, (i, a, b) -> (byte) Math.max(a, b)));
 598         }
 599 
 600     @Override
 601     @ForceInline
 602     public Byte512Vector max(Vector<Byte> v, VectorMask<Byte> m) {
 603         return blend(max(v), m);
 604     }
 605 
 606     @Override
 607     @ForceInline
 608     public Byte512Vector and(Vector<Byte> o) {
 609         Objects.requireNonNull(o);
 610         Byte512Vector v = (Byte512Vector)o;
 611         return VectorIntrinsics.binaryOp(
 612             VECTOR_OP_AND, Byte512Vector.class, byte.class, LENGTH,
 613             this, v,
 614             (v1, v2) -> v1.bOp(v2, (i, a, b) -> (byte)(a & b)));
 615     }
 616 
 617     @Override
 618     @ForceInline
 619     public Byte512Vector or(Vector<Byte> o) {
 620         Objects.requireNonNull(o);
 621         Byte512Vector v = (Byte512Vector)o;
 622         return VectorIntrinsics.binaryOp(
 623             VECTOR_OP_OR, Byte512Vector.class, byte.class, LENGTH,
 624             this, v,
 625             (v1, v2) -> v1.bOp(v2, (i, a, b) -> (byte)(a | b)));
 626     }
 627 
 628     @Override
 629     @ForceInline
 630     public Byte512Vector xor(Vector<Byte> o) {
 631         Objects.requireNonNull(o);
 632         Byte512Vector v = (Byte512Vector)o;
 633         return VectorIntrinsics.binaryOp(
 634             VECTOR_OP_XOR, Byte512Vector.class, byte.class, LENGTH,
 635             this, v,
 636             (v1, v2) -> v1.bOp(v2, (i, a, b) -> (byte)(a ^ b)));
 637     }
 638 
 639     @Override
 640     @ForceInline
 641     public Byte512Vector and(Vector<Byte> v, VectorMask<Byte> m) {
 642         return blend(and(v), m);
 643     }
 644 
 645     @Override
 646     @ForceInline
 647     public Byte512Vector or(Vector<Byte> v, VectorMask<Byte> m) {
 648         return blend(or(v), m);
 649     }
 650 
 651     @Override
 652     @ForceInline
 653     public Byte512Vector xor(Vector<Byte> v, VectorMask<Byte> m) {
 654         return blend(xor(v), m);
 655     }
 656 
 657     @Override
 658     @ForceInline
 659     public Byte512Vector shiftL(int s) {
 660         return VectorIntrinsics.broadcastInt(
 661             VECTOR_OP_LSHIFT, Byte512Vector.class, byte.class, LENGTH,
 662             this, s,
 663             (v, i) -> v.uOp((__, a) -> (byte) (a << (i & 7))));
 664     }
 665 
 666     @Override
 667     @ForceInline
 668     public Byte512Vector shiftL(int s, VectorMask<Byte> m) {
 669         return blend(shiftL(s), m);
 670     }
 671 
 672     @Override
 673     @ForceInline
 674     public Byte512Vector shiftR(int s) {
 675         return VectorIntrinsics.broadcastInt(
 676             VECTOR_OP_URSHIFT, Byte512Vector.class, byte.class, LENGTH,
 677             this, s,
 678             (v, i) -> v.uOp((__, a) -> (byte) ((a & 0xFF) >>> (i & 7))));
 679     }
 680 
 681     @Override
 682     @ForceInline
 683     public Byte512Vector shiftR(int s, VectorMask<Byte> m) {
 684         return blend(shiftR(s), m);
 685     }
 686 
 687     @Override
 688     @ForceInline
 689     public Byte512Vector aShiftR(int s) {
 690         return VectorIntrinsics.broadcastInt(
 691             VECTOR_OP_RSHIFT, Byte512Vector.class, byte.class, LENGTH,
 692             this, s,
 693             (v, i) -> v.uOp((__, a) -> (byte) (a >> (i & 7))));
 694     }
 695 
 696     @Override
 697     @ForceInline
 698     public Byte512Vector aShiftR(int s, VectorMask<Byte> m) {
 699         return blend(aShiftR(s), m);
 700     }
 701     // Ternary operations
 702 
 703 
 704     // Type specific horizontal reductions
 705 
 706     @Override
 707     @ForceInline
 708     public byte addAll() {
 709         return (byte) VectorIntrinsics.reductionCoerced(
 710             VECTOR_OP_ADD, Byte512Vector.class, byte.class, LENGTH,
 711             this,
 712             v -> (long) v.rOp((byte) 0, (i, a, b) -> (byte) (a + b)));
 713     }
 714 
 715     @Override
 716     @ForceInline
 717     public byte andAll() {
 718         return (byte) VectorIntrinsics.reductionCoerced(
 719             VECTOR_OP_AND, Byte512Vector.class, byte.class, LENGTH,
 720             this,
 721             v -> (long) v.rOp((byte) -1, (i, a, b) -> (byte) (a & b)));
 722     }
 723 
 724     @Override
 725     @ForceInline
 726     public byte andAll(VectorMask<Byte> m) {
 727         return blend((Byte512Vector)ByteVector.broadcast(SPECIES, (byte) -1), m).andAll();
 728     }
 729 
 730     @Override
 731     @ForceInline
 732     public byte minAll() {
 733         return (byte) VectorIntrinsics.reductionCoerced(
 734             VECTOR_OP_MIN, Byte512Vector.class, byte.class, LENGTH,
 735             this,
 736             v -> (long) v.rOp(Byte.MAX_VALUE , (i, a, b) -> (byte) Math.min(a, b)));
 737     }
 738 
 739     @Override
 740     @ForceInline
 741     public byte maxAll() {
 742         return (byte) VectorIntrinsics.reductionCoerced(
 743             VECTOR_OP_MAX, Byte512Vector.class, byte.class, LENGTH,
 744             this,
 745             v -> (long) v.rOp(Byte.MIN_VALUE , (i, a, b) -> (byte) Math.max(a, b)));
 746     }
 747 
 748     @Override
 749     @ForceInline
 750     public byte mulAll() {
 751         return (byte) VectorIntrinsics.reductionCoerced(
 752             VECTOR_OP_MUL, Byte512Vector.class, byte.class, LENGTH,
 753             this,
 754             v -> (long) v.rOp((byte) 1, (i, a, b) -> (byte) (a * b)));
 755     }
 756 
 757     @Override
 758     @ForceInline
 759     public byte orAll() {
 760         return (byte) VectorIntrinsics.reductionCoerced(
 761             VECTOR_OP_OR, Byte512Vector.class, byte.class, LENGTH,
 762             this,
 763             v -> (long) v.rOp((byte) 0, (i, a, b) -> (byte) (a | b)));
 764     }
 765 
 766     @Override
 767     @ForceInline
 768     public byte orAll(VectorMask<Byte> m) {
 769         return blend((Byte512Vector)ByteVector.broadcast(SPECIES, (byte) 0), m).orAll();
 770     }
 771 
 772     @Override
 773     @ForceInline
 774     public byte xorAll() {
 775         return (byte) VectorIntrinsics.reductionCoerced(
 776             VECTOR_OP_XOR, Byte512Vector.class, byte.class, LENGTH,
 777             this,
 778             v -> (long) v.rOp((byte) 0, (i, a, b) -> (byte) (a ^ b)));
 779     }
 780 
 781     @Override
 782     @ForceInline
 783     public byte xorAll(VectorMask<Byte> m) {
 784         return blend((Byte512Vector)ByteVector.broadcast(SPECIES, (byte) 0), m).xorAll();
 785     }
 786 
 787 
 788     @Override
 789     @ForceInline
 790     public byte addAll(VectorMask<Byte> m) {
 791         return blend((Byte512Vector)ByteVector.broadcast(SPECIES, (byte) 0), m).addAll();
 792     }
 793 
 794 
 795     @Override
 796     @ForceInline
 797     public byte mulAll(VectorMask<Byte> m) {
 798         return blend((Byte512Vector)ByteVector.broadcast(SPECIES, (byte) 1), m).mulAll();
 799     }
 800 
 801     @Override
 802     @ForceInline
 803     public byte minAll(VectorMask<Byte> m) {
 804         return blend((Byte512Vector)ByteVector.broadcast(SPECIES, Byte.MAX_VALUE), m).minAll();
 805     }
 806 
 807     @Override
 808     @ForceInline
 809     public byte maxAll(VectorMask<Byte> m) {
 810         return blend((Byte512Vector)ByteVector.broadcast(SPECIES, Byte.MIN_VALUE), m).maxAll();
 811     }
 812 
 813     @Override
 814     @ForceInline
 815     public VectorShuffle<Byte> toShuffle() {
 816         byte[] a = toArray();
 817         int[] sa = new int[a.length];
 818         for (int i = 0; i < a.length; i++) {
 819             sa[i] = (int) a[i];
 820         }
 821         return VectorShuffle.fromArray(SPECIES, sa, 0);
 822     }
 823 
 824     // Memory operations
 825 
 826     private static final int ARRAY_SHIFT         = 31 - Integer.numberOfLeadingZeros(Unsafe.ARRAY_BYTE_INDEX_SCALE);
 827     private static final int BOOLEAN_ARRAY_SHIFT = 31 - Integer.numberOfLeadingZeros(Unsafe.ARRAY_BOOLEAN_INDEX_SCALE);
 828 
 829     @Override
 830     @ForceInline
 831     public void intoArray(byte[] a, int ix) {
 832         Objects.requireNonNull(a);
 833         ix = VectorIntrinsics.checkIndex(ix, a.length, LENGTH);
 834         VectorIntrinsics.store(Byte512Vector.class, byte.class, LENGTH,
 835                                a, (((long) ix) << ARRAY_SHIFT) + Unsafe.ARRAY_BYTE_BASE_OFFSET,
 836                                this,
 837                                a, ix,
 838                                (arr, idx, v) -> v.forEach((i, e) -> arr[idx + i] = e));
 839     }
 840 
 841     @Override
 842     @ForceInline
 843     public final void intoArray(byte[] a, int ax, VectorMask<Byte> m) {
 844         ByteVector oldVal = ByteVector.fromArray(SPECIES, a, ax);
 845         ByteVector newVal = oldVal.blend(this, m);
 846         newVal.intoArray(a, ax);
 847     }
 848 
 849     @Override
 850     @ForceInline
 851     public void intoByteArray(byte[] a, int ix) {
 852         Objects.requireNonNull(a);
 853         ix = VectorIntrinsics.checkIndex(ix, a.length, bitSize() / Byte.SIZE);
 854         VectorIntrinsics.store(Byte512Vector.class, byte.class, LENGTH,
 855                                a, ((long) ix) + Unsafe.ARRAY_BYTE_BASE_OFFSET,
 856                                this,
 857                                a, ix,
 858                                (c, idx, v) -> {
 859                                    ByteBuffer bbc = ByteBuffer.wrap(c, idx, c.length - idx).order(ByteOrder.nativeOrder());
 860                                    ByteBuffer tb = bbc;
 861                                    v.forEach((i, e) -> tb.put(e));
 862                                });
 863     }
 864 
 865     @Override
 866     @ForceInline
 867     public final void intoByteArray(byte[] a, int ix, VectorMask<Byte> m) {
 868         Byte512Vector oldVal = (Byte512Vector) ByteVector.fromByteArray(SPECIES, a, ix);
 869         Byte512Vector newVal = oldVal.blend(this, m);
 870         newVal.intoByteArray(a, ix);
 871     }
 872 
 873     @Override
 874     @ForceInline
 875     public void intoByteBuffer(ByteBuffer bb, int ix) {
 876         if (bb.order() != ByteOrder.nativeOrder()) {
 877             throw new IllegalArgumentException();
 878         }
 879         if (bb.isReadOnly()) {
 880             throw new ReadOnlyBufferException();
 881         }
 882         ix = VectorIntrinsics.checkIndex(ix, bb.limit(), bitSize() / Byte.SIZE);
 883         VectorIntrinsics.store(Byte512Vector.class, byte.class, LENGTH,
 884                                U.getReference(bb, BYTE_BUFFER_HB), ix + U.getLong(bb, BUFFER_ADDRESS),
 885                                this,
 886                                bb, ix,
 887                                (c, idx, v) -> {
 888                                    ByteBuffer bbc = c.duplicate().position(idx).order(ByteOrder.nativeOrder());
 889                                    ByteBuffer tb = bbc;
 890                                    v.forEach((i, e) -> tb.put(e));
 891                                });
 892     }
 893 
 894     @Override
 895     @ForceInline
 896     public void intoByteBuffer(ByteBuffer bb, int ix, VectorMask<Byte> m) {
 897         Byte512Vector oldVal = (Byte512Vector) ByteVector.fromByteBuffer(SPECIES, bb, ix);
 898         Byte512Vector newVal = oldVal.blend(this, m);
 899         newVal.intoByteBuffer(bb, ix);
 900     }
 901 
 902     //
 903 
 904     @Override
 905     public String toString() {
 906         return Arrays.toString(getElements());
 907     }
 908 
 909     @Override
 910     public boolean equals(Object o) {
 911         if (this == o) return true;
 912         if (o == null || this.getClass() != o.getClass()) return false;
 913 
 914         Byte512Vector that = (Byte512Vector) o;
 915         return this.equal(that).allTrue();
 916     }
 917 
 918     @Override
 919     public int hashCode() {
 920         return Arrays.hashCode(vec);
 921     }
 922 
 923     // Binary test
 924 
 925     @Override
 926     Byte512Mask bTest(Vector<Byte> o, FBinTest f) {
 927         byte[] vec1 = getElements();
 928         byte[] vec2 = ((Byte512Vector)o).getElements();
 929         boolean[] bits = new boolean[length()];
 930         for (int i = 0; i < length(); i++){
 931             bits[i] = f.apply(i, vec1[i], vec2[i]);
 932         }
 933         return new Byte512Mask(bits);
 934     }
 935 
 936     // Comparisons
 937 
 938     @Override
 939     @ForceInline
 940     public Byte512Mask equal(Vector<Byte> o) {
 941         Objects.requireNonNull(o);
 942         Byte512Vector v = (Byte512Vector)o;
 943 
 944         return VectorIntrinsics.compare(
 945             BT_eq, Byte512Vector.class, Byte512Mask.class, byte.class, LENGTH,
 946             this, v,
 947             (v1, v2) -> v1.bTest(v2, (i, a, b) -> a == b));
 948     }
 949 
 950     @Override
 951     @ForceInline
 952     public Byte512Mask notEqual(Vector<Byte> o) {
 953         Objects.requireNonNull(o);
 954         Byte512Vector v = (Byte512Vector)o;
 955 
 956         return VectorIntrinsics.compare(
 957             BT_ne, Byte512Vector.class, Byte512Mask.class, byte.class, LENGTH,
 958             this, v,
 959             (v1, v2) -> v1.bTest(v2, (i, a, b) -> a != b));
 960     }
 961 
 962     @Override
 963     @ForceInline
 964     public Byte512Mask lessThan(Vector<Byte> o) {
 965         Objects.requireNonNull(o);
 966         Byte512Vector v = (Byte512Vector)o;
 967 
 968         return VectorIntrinsics.compare(
 969             BT_lt, Byte512Vector.class, Byte512Mask.class, byte.class, LENGTH,
 970             this, v,
 971             (v1, v2) -> v1.bTest(v2, (i, a, b) -> a < b));
 972     }
 973 
 974     @Override
 975     @ForceInline
 976     public Byte512Mask lessThanEq(Vector<Byte> o) {
 977         Objects.requireNonNull(o);
 978         Byte512Vector v = (Byte512Vector)o;
 979 
 980         return VectorIntrinsics.compare(
 981             BT_le, Byte512Vector.class, Byte512Mask.class, byte.class, LENGTH,
 982             this, v,
 983             (v1, v2) -> v1.bTest(v2, (i, a, b) -> a <= b));
 984     }
 985 
 986     @Override
 987     @ForceInline
 988     public Byte512Mask greaterThan(Vector<Byte> o) {
 989         Objects.requireNonNull(o);
 990         Byte512Vector v = (Byte512Vector)o;
 991 
 992         return (Byte512Mask) VectorIntrinsics.compare(
 993             BT_gt, Byte512Vector.class, Byte512Mask.class, byte.class, LENGTH,
 994             this, v,
 995             (v1, v2) -> v1.bTest(v2, (i, a, b) -> a > b));
 996     }
 997 
 998     @Override
 999     @ForceInline
1000     public Byte512Mask greaterThanEq(Vector<Byte> o) {
1001         Objects.requireNonNull(o);
1002         Byte512Vector v = (Byte512Vector)o;
1003 
1004         return VectorIntrinsics.compare(
1005             BT_ge, Byte512Vector.class, Byte512Mask.class, byte.class, LENGTH,
1006             this, v,
1007             (v1, v2) -> v1.bTest(v2, (i, a, b) -> a >= b));
1008     }
1009 
1010     // Foreach
1011 
1012     @Override
1013     void forEach(FUnCon f) {
1014         byte[] vec = getElements();
1015         for (int i = 0; i < length(); i++) {
1016             f.apply(i, vec[i]);
1017         }
1018     }
1019 
1020     @Override
1021     void forEach(VectorMask<Byte> o, FUnCon f) {
1022         boolean[] mbits = ((Byte512Mask)o).getBits();
1023         forEach((i, a) -> {
1024             if (mbits[i]) { f.apply(i, a); }
1025         });
1026     }
1027 
1028 
1029 
1030     @Override
1031     public Byte512Vector rotateEL(int j) {
1032         byte[] vec = getElements();
1033         byte[] res = new byte[length()];
1034         for (int i = 0; i < length(); i++){
1035             res[(j + i) % length()] = vec[i];
1036         }
1037         return new Byte512Vector(res);
1038     }
1039 
1040     @Override
1041     public Byte512Vector rotateER(int j) {
1042         byte[] vec = getElements();
1043         byte[] res = new byte[length()];
1044         for (int i = 0; i < length(); i++){
1045             int z = i - j;
1046             if(j < 0) {
1047                 res[length() + z] = vec[i];
1048             } else {
1049                 res[z] = vec[i];
1050             }
1051         }
1052         return new Byte512Vector(res);
1053     }
1054 
1055     @Override
1056     public Byte512Vector shiftEL(int j) {
1057         byte[] vec = getElements();
1058         byte[] res = new byte[length()];
1059         for (int i = 0; i < length() - j; i++) {
1060             res[i] = vec[i + j];
1061         }
1062         return new Byte512Vector(res);
1063     }
1064 
1065     @Override
1066     public Byte512Vector shiftER(int j) {
1067         byte[] vec = getElements();
1068         byte[] res = new byte[length()];
1069         for (int i = 0; i < length() - j; i++){
1070             res[i + j] = vec[i];
1071         }
1072         return new Byte512Vector(res);
1073     }
1074 
1075     @Override
1076     @ForceInline
1077     public Byte512Vector rearrange(Vector<Byte> v,
1078                                   VectorShuffle<Byte> s, VectorMask<Byte> m) {
1079         return this.rearrange(s).blend(v.rearrange(s), m);
1080     }
1081 
1082     @Override
1083     @ForceInline
1084     public Byte512Vector rearrange(VectorShuffle<Byte> o1) {
1085         Objects.requireNonNull(o1);
1086         Byte512Shuffle s =  (Byte512Shuffle)o1;
1087 
1088         return VectorIntrinsics.rearrangeOp(
1089             Byte512Vector.class, Byte512Shuffle.class, byte.class, LENGTH,
1090             this, s,
1091             (v1, s_) -> v1.uOp((i, a) -> {
1092                 int ei = s_.getElement(i);
1093                 return v1.get(ei);
1094             }));
1095     }
1096 
1097     @Override
1098     @ForceInline
1099     public Byte512Vector blend(Vector<Byte> o1, VectorMask<Byte> o2) {
1100         Objects.requireNonNull(o1);
1101         Objects.requireNonNull(o2);
1102         Byte512Vector v = (Byte512Vector)o1;
1103         Byte512Mask   m = (Byte512Mask)o2;
1104 
1105         return VectorIntrinsics.blend(
1106             Byte512Vector.class, Byte512Mask.class, byte.class, LENGTH,
1107             this, v, m,
1108             (v1, v2, m_) -> v1.bOp(v2, (i, a, b) -> m_.getElement(i) ? b : a));
1109     }
1110 
1111     // Accessors
1112 
1113     @Override
1114     public byte get(int i) {
1115         if (i < 0 || i >= LENGTH) {
1116             throw new IllegalArgumentException("Index " + i + " must be zero or positive, and less than " + LENGTH);
1117         }
1118         return (byte) VectorIntrinsics.extract(
1119                                 Byte512Vector.class, byte.class, LENGTH,
1120                                 this, i,
1121                                 (vec, ix) -> {
1122                                     byte[] vecarr = vec.getElements();
1123                                     return (long)vecarr[ix];
1124                                 });
1125     }
1126 
1127     @Override
1128     public Byte512Vector with(int i, byte e) {
1129         if (i < 0 || i >= LENGTH) {
1130             throw new IllegalArgumentException("Index " + i + " must be zero or positive, and less than " + LENGTH);
1131         }
1132         return VectorIntrinsics.insert(
1133                                 Byte512Vector.class, byte.class, LENGTH,
1134                                 this, i, (long)e,
1135                                 (v, ix, bits) -> {
1136                                     byte[] res = v.getElements().clone();
1137                                     res[ix] = (byte)bits;
1138                                     return new Byte512Vector(res);
1139                                 });
1140     }
1141 
1142     // Mask
1143 
1144     static final class Byte512Mask extends AbstractMask<Byte> {
1145         static final Byte512Mask TRUE_MASK = new Byte512Mask(true);
1146         static final Byte512Mask FALSE_MASK = new Byte512Mask(false);
1147 
1148         private final boolean[] bits; // Don't access directly, use getBits() instead.
1149 
1150         public Byte512Mask(boolean[] bits) {
1151             this(bits, 0);
1152         }
1153 
1154         public Byte512Mask(boolean[] bits, int offset) {
1155             boolean[] a = new boolean[species().length()];
1156             for (int i = 0; i < a.length; i++) {
1157                 a[i] = bits[offset + i];
1158             }
1159             this.bits = a;
1160         }
1161 
1162         public Byte512Mask(boolean val) {
1163             boolean[] bits = new boolean[species().length()];
1164             Arrays.fill(bits, val);
1165             this.bits = bits;
1166         }
1167 
1168         boolean[] getBits() {
1169             return VectorIntrinsics.maybeRebox(this).bits;
1170         }
1171 
1172         @Override
1173         Byte512Mask uOp(MUnOp f) {
1174             boolean[] res = new boolean[species().length()];
1175             boolean[] bits = getBits();
1176             for (int i = 0; i < species().length(); i++) {
1177                 res[i] = f.apply(i, bits[i]);
1178             }
1179             return new Byte512Mask(res);
1180         }
1181 
1182         @Override
1183         Byte512Mask bOp(VectorMask<Byte> o, MBinOp f) {
1184             boolean[] res = new boolean[species().length()];
1185             boolean[] bits = getBits();
1186             boolean[] mbits = ((Byte512Mask)o).getBits();
1187             for (int i = 0; i < species().length(); i++) {
1188                 res[i] = f.apply(i, bits[i], mbits[i]);
1189             }
1190             return new Byte512Mask(res);
1191         }
1192 
1193         @Override
1194         public VectorSpecies<Byte> species() {
1195             return SPECIES;
1196         }
1197 
1198         @Override
1199         public Byte512Vector toVector() {
1200             byte[] res = new byte[species().length()];
1201             boolean[] bits = getBits();
1202             for (int i = 0; i < species().length(); i++) {
1203                 // -1 will result in the most significant bit being set in
1204                 // addition to some or all other bits
1205                 res[i] = (byte) (bits[i] ? -1 : 0);
1206             }
1207             return new Byte512Vector(res);
1208         }
1209 
1210         @Override
1211         @ForceInline
1212         @SuppressWarnings("unchecked")
1213         public <E> VectorMask<E> cast(VectorSpecies<E> species) {
1214             if (length() != species.length())
1215                 throw new IllegalArgumentException("VectorMask length and species length differ");
1216             Class<?> stype = species.elementType();
1217             boolean [] maskArray = toArray();
1218             if (stype == byte.class) {
1219                 return (VectorMask <E>) new Byte512Vector.Byte512Mask(maskArray);
1220             } else if (stype == short.class) {
1221                 return (VectorMask <E>) new Short512Vector.Short512Mask(maskArray);
1222             } else if (stype == int.class) {
1223                 return (VectorMask <E>) new Int512Vector.Int512Mask(maskArray);
1224             } else if (stype == long.class) {
1225                 return (VectorMask <E>) new Long512Vector.Long512Mask(maskArray);
1226             } else if (stype == float.class) {
1227                 return (VectorMask <E>) new Float512Vector.Float512Mask(maskArray);
1228             } else if (stype == double.class) {
1229                 return (VectorMask <E>) new Double512Vector.Double512Mask(maskArray);
1230             } else {
1231                 throw new UnsupportedOperationException("Bad lane type for casting.");
1232             }
1233         }
1234 
1235         // Unary operations
1236 
1237         @Override
1238         @ForceInline
1239         public Byte512Mask not() {
1240             return (Byte512Mask) VectorIntrinsics.unaryOp(
1241                                              VECTOR_OP_NOT, Byte512Mask.class, byte.class, LENGTH,
1242                                              this,
1243                                              (m1) -> m1.uOp((i, a) -> !a));
1244         }
1245 
1246         // Binary operations
1247 
1248         @Override
1249         @ForceInline
1250         public Byte512Mask and(VectorMask<Byte> o) {
1251             Objects.requireNonNull(o);
1252             Byte512Mask m = (Byte512Mask)o;
1253             return VectorIntrinsics.binaryOp(VECTOR_OP_AND, Byte512Mask.class, byte.class, LENGTH,
1254                                              this, m,
1255                                              (m1, m2) -> m1.bOp(m2, (i, a, b) -> a & b));
1256         }
1257 
1258         @Override
1259         @ForceInline
1260         public Byte512Mask or(VectorMask<Byte> o) {
1261             Objects.requireNonNull(o);
1262             Byte512Mask m = (Byte512Mask)o;
1263             return VectorIntrinsics.binaryOp(VECTOR_OP_OR, Byte512Mask.class, byte.class, LENGTH,
1264                                              this, m,
1265                                              (m1, m2) -> m1.bOp(m2, (i, a, b) -> a | b));
1266         }
1267 
1268         // Reductions
1269 
1270         @Override
1271         @ForceInline
1272         public boolean anyTrue() {
1273             return VectorIntrinsics.test(BT_ne, Byte512Mask.class, byte.class, LENGTH,
1274                                          this, this,
1275                                          (m, __) -> anyTrueHelper(((Byte512Mask)m).getBits()));
1276         }
1277 
1278         @Override
1279         @ForceInline
1280         public boolean allTrue() {
1281             return VectorIntrinsics.test(BT_overflow, Byte512Mask.class, byte.class, LENGTH,
1282                                          this, VectorMask.maskAllTrue(species()),
1283                                          (m, __) -> allTrueHelper(((Byte512Mask)m).getBits()));
1284         }
1285     }
1286 
1287     // Shuffle
1288 
1289     static final class Byte512Shuffle extends AbstractShuffle<Byte> {
1290         Byte512Shuffle(byte[] reorder) {
1291             super(reorder);
1292         }
1293 
1294         public Byte512Shuffle(int[] reorder) {
1295             super(reorder);
1296         }
1297 
1298         public Byte512Shuffle(int[] reorder, int i) {
1299             super(reorder, i);
1300         }
1301 
1302         public Byte512Shuffle(IntUnaryOperator f) {
1303             super(f);
1304         }
1305 
1306         @Override
1307         public VectorSpecies<Byte> species() {
1308             return SPECIES;
1309         }
1310 
1311         @Override
1312         public ByteVector toVector() {
1313             byte[] va = new byte[SPECIES.length()];
1314             for (int i = 0; i < va.length; i++) {
1315               va[i] = (byte) getElement(i);
1316             }
1317             return ByteVector.fromArray(SPECIES, va, 0);
1318         }
1319 
1320         @Override
1321         @ForceInline
1322         @SuppressWarnings("unchecked")
1323         public <F> VectorShuffle<F> cast(VectorSpecies<F> species) {
1324             if (length() != species.length())
1325                 throw new IllegalArgumentException("Shuffle length and species length differ");
1326             Class<?> stype = species.elementType();
1327             int [] shuffleArray = toArray();
1328             if (stype == byte.class) {
1329                 return (VectorShuffle<F>) new Byte512Vector.Byte512Shuffle(shuffleArray);
1330             } else if (stype == short.class) {
1331                 return (VectorShuffle<F>) new Short512Vector.Short512Shuffle(shuffleArray);
1332             } else if (stype == int.class) {
1333                 return (VectorShuffle<F>) new Int512Vector.Int512Shuffle(shuffleArray);
1334             } else if (stype == long.class) {
1335                 return (VectorShuffle<F>) new Long512Vector.Long512Shuffle(shuffleArray);
1336             } else if (stype == float.class) {
1337                 return (VectorShuffle<F>) new Float512Vector.Float512Shuffle(shuffleArray);
1338             } else if (stype == double.class) {
1339                 return (VectorShuffle<F>) new Double512Vector.Double512Shuffle(shuffleArray);
1340             } else {
1341                 throw new UnsupportedOperationException("Bad lane type for casting.");
1342             }
1343         }
1344 
1345         @Override
1346         public Byte512Shuffle rearrange(VectorShuffle<Byte> o) {
1347             Byte512Shuffle s = (Byte512Shuffle) o;
1348             byte[] r = new byte[reorder.length];
1349             for (int i = 0; i < reorder.length; i++) {
1350                 r[i] = reorder[s.reorder[i]];
1351             }
1352             return new Byte512Shuffle(r);
1353         }
1354     }
1355 
1356     // VectorSpecies
1357 
1358     @Override
1359     public VectorSpecies<Byte> species() {
1360         return SPECIES;
1361     }
1362 }