1 /*
2 * Copyright (c) 2021, 2022, 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.
8 *
9 * This code is distributed in the hope that it will be useful, but WITHOUT
10 * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
11 * FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License
12 * version 2 for more details (a copy is included in the LICENSE file that
13 * accompanied this code).
14 *
15 * You should have received a copy of the GNU General Public License version
16 * 2 along with this work; if not, write to the Free Software Foundation,
17 * Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA.
18 *
19 * Please contact Oracle, 500 Oracle Parkway, Redwood Shores, CA 94065 USA
20 * or visit www.oracle.com if you need additional information or have any
21 * questions.
22 *
23 */
24 package org.openjdk.bench.jdk.incubator.vector;
25
26 import java.lang.foreign.Arena;
27 import java.lang.foreign.MemorySegment;
28 import java.nio.ByteOrder;
29 import java.util.concurrent.TimeUnit;
30 import jdk.incubator.vector.ByteVector;
31 import jdk.incubator.vector.VectorOperators;
32 import jdk.incubator.vector.VectorSpecies;
33 import org.openjdk.jmh.annotations.Benchmark;
34 import org.openjdk.jmh.annotations.BenchmarkMode;
35 import org.openjdk.jmh.annotations.CompilerControl;
36 import org.openjdk.jmh.annotations.Fork;
37 import org.openjdk.jmh.annotations.Measurement;
38 import org.openjdk.jmh.annotations.Mode;
39 import org.openjdk.jmh.annotations.OutputTimeUnit;
40 import org.openjdk.jmh.annotations.Param;
41 import org.openjdk.jmh.annotations.Setup;
42 import org.openjdk.jmh.annotations.State;
43 import org.openjdk.jmh.annotations.Warmup;
44
45 @BenchmarkMode(Mode.AverageTime)
46 @Warmup(iterations = 5, time = 500, timeUnit = TimeUnit.MILLISECONDS)
47 @Measurement(iterations = 10, time = 500, timeUnit = TimeUnit.MILLISECONDS)
48 @State(org.openjdk.jmh.annotations.Scope.Thread)
49 @OutputTimeUnit(TimeUnit.NANOSECONDS)
50 @Fork(value = 1, jvmArgsAppend = {
51 "--add-modules=jdk.incubator.vector",
52 "--enable-preview",
53 "--enable-native-access", "ALL-UNNAMED",
54 "-Djdk.incubator.vector.VECTOR_ACCESS_OOB_CHECK=1"})
55 public class TestLoadStoreBytes {
56 private static final VectorSpecies<Byte> SPECIES = VectorSpecies.ofLargestShape(byte.class);
57
58 @Param("1024")
59 private int size;
60
61 private byte[] srcArray;
62
63 private byte[] dstArray;
64
65
66 private MemorySegment srcSegmentHeap;
67
68 private MemorySegment dstSegmentHeap;
69
70 private MemorySegment srcSegment;
71
72 private MemorySegment dstSegment;
73
74 private byte[] a, b, c;
75
76 @Setup
77 public void setup() {
78 srcArray = new byte[size];
79 dstArray = srcArray.clone();
80 for (int i = 0; i < srcArray.length; i++) {
81 srcArray[i] = (byte) i;
82 }
83
84 srcSegmentHeap = MemorySegment.ofArray(new byte[size]);
85 dstSegmentHeap = MemorySegment.ofArray(new byte[size]);
86
87 srcSegment = Arena.ofAuto().allocate(size, SPECIES.vectorByteSize());
88 dstSegment = Arena.ofAuto().allocate(size, SPECIES.vectorByteSize());
89
90 a = new byte[size];
91 b = new byte[size];
92 c = new byte[size];
93 }
94
95
96 @Benchmark
97 public void array() {
98 for (int i = 0; i < SPECIES.loopBound(srcArray.length); i += SPECIES.length()) {
99 var v = ByteVector.fromArray(SPECIES, srcArray, i);
100 v.intoArray(dstArray, i);
101 }
102 }
103
104 @Benchmark
105 public void arrayScalar() {
106 for (int i = 0; i < SPECIES.loopBound(srcArray.length); i ++) {
107 var v = srcArray[i];
108 dstArray[i] = v;
109 }
110 }
111
112 @Benchmark
113 public void vectAdd1() {
114 var a = this.a;
115 var b = this.b;
116 var c = this.c;
117
118 for (int i = 0; i < a.length; i += SPECIES.length()) {
119 ByteVector av = ByteVector.fromArray(SPECIES, a, i);
120 ByteVector bv = ByteVector.fromArray(SPECIES, b, i);
121 av.lanewise(VectorOperators.ADD, bv).intoArray(c, i);
122 }
123 }
124
125 @Benchmark
126 public void vectAdd2() {
127 var a = this.a;
128 var b = this.b;
129 var c = this.c;
130
131 for (int i = 0; i < a.length/SPECIES.length(); i++) {
132 ByteVector av = ByteVector.fromArray(SPECIES, a, (i*SPECIES.length()));
133 ByteVector bv = ByteVector.fromArray(SPECIES, b, (i*SPECIES.length()));
134 av.lanewise(VectorOperators.ADD, bv).intoArray(c, (i*SPECIES.length()));
135 }
136 }
137
138 @Benchmark
139 public void arrayAdd() {
140 for (int i = 0; i < SPECIES.loopBound(srcArray.length); i += SPECIES.length()) {
141 var v = ByteVector.fromArray(SPECIES, srcArray, i);
142 v = v.add(v);
143 v.intoArray(dstArray, i);
144 }
145 }
146
147 @Benchmark
148 public void segmentHeap() {
149 for (long i = 0; i < SPECIES.loopBound(srcArray.length); i += SPECIES.length()) {
150 var v = ByteVector.fromMemorySegment(SPECIES, srcSegmentHeap, i, ByteOrder.nativeOrder());
151 v.intoMemorySegment(dstSegmentHeap, i, ByteOrder.nativeOrder());
152 }
153 }
154
155 @Benchmark
156 public void segmentNativeImplicit() {
157 for (long i = 0; i < SPECIES.loopBound(srcArray.length); i += SPECIES.length()) {
158 var v = ByteVector.fromMemorySegment(SPECIES, srcSegment, i, ByteOrder.nativeOrder());
159 v.intoMemorySegment(dstSegment, i, ByteOrder.nativeOrder());
160 }
161 }
162
163 @Benchmark
164 public void segmentNativeConfined() {
165 try (final var arena = Arena.ofConfined()) {
166 final var srcSegmentConfined = srcSegment.reinterpret(arena, null);
167 final var dstSegmentConfined = dstSegment.reinterpret(arena, null);
168
169 for (long i = 0; i < SPECIES.loopBound(srcArray.length); i += SPECIES.length()) {
170 var v = ByteVector.fromMemorySegment(SPECIES, srcSegmentConfined, i, ByteOrder.nativeOrder());
171 v.intoMemorySegment(dstSegmentConfined, i, ByteOrder.nativeOrder());
172 }
173 }
174 }
175 }