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.nio.ByteOrder;
28 import java.util.concurrent.TimeUnit;
29
30 import java.lang.foreign.MemorySegment;
31
32 import jdk.incubator.vector.ShortVector;
33 import jdk.incubator.vector.VectorOperators;
34 import jdk.incubator.vector.VectorSpecies;
35 import org.openjdk.jmh.annotations.Benchmark;
36 import org.openjdk.jmh.annotations.BenchmarkMode;
37 import org.openjdk.jmh.annotations.CompilerControl;
38 import org.openjdk.jmh.annotations.Fork;
39 import org.openjdk.jmh.annotations.Measurement;
40 import org.openjdk.jmh.annotations.Mode;
41 import org.openjdk.jmh.annotations.OutputTimeUnit;
42 import org.openjdk.jmh.annotations.Param;
43 import org.openjdk.jmh.annotations.Setup;
44 import org.openjdk.jmh.annotations.State;
45 import org.openjdk.jmh.annotations.TearDown;
46 import org.openjdk.jmh.annotations.Warmup;
47
48 @BenchmarkMode(Mode.AverageTime)
49 @Warmup(iterations = 5, time = 500, timeUnit = TimeUnit.MILLISECONDS)
50 @Measurement(iterations = 10, time = 500, timeUnit = TimeUnit.MILLISECONDS)
51 @State(org.openjdk.jmh.annotations.Scope.Thread)
52 @OutputTimeUnit(TimeUnit.NANOSECONDS)
53 @Fork(value = 1, jvmArgsAppend = {
54 "--add-modules=jdk.incubator.vector",
55 "--enable-native-access", "ALL-UNNAMED"})
56 public class TestLoadStoreShorts {
57 private static final VectorSpecies<Short> SPECIES = VectorSpecies.ofLargestShape(short.class);
58
59 @Param("256")
60 private int size;
61
62 private int longSize;
63
64 private short[] srcArray;
65
66 private short[] dstArray;
67
68
69 private MemorySegment srcSegmentHeap;
70
71 private MemorySegment dstSegmentHeap;
72
73 private MemorySegment srcSegment;
74
75 private MemorySegment dstSegment;
76
77 private short[] a, b, c;
78
79 @Setup
80 public void setup() {
81 var longSize = size / Short.BYTES;
82 srcArray = new short[longSize];
83 dstArray = srcArray.clone();
84 for (int i = 0; i < srcArray.length; i++) {
85 srcArray[i] = (short) i;
86 }
87
88 srcSegmentHeap = MemorySegment.ofArray(new byte[size]);
89 dstSegmentHeap = MemorySegment.ofArray(new byte[size]);
90
91 srcSegment = Arena.ofAuto().allocate(size, SPECIES.vectorByteSize());
92 dstSegment = Arena.ofAuto().allocate(size, SPECIES.vectorByteSize());
93
94 this.longSize = longSize;
95
96 a = new short[size];
97 b = new short[size];
98 c = new short[size];
99 }
100
101 @Benchmark
102 public void array() {
103 for (int i = 0; i < SPECIES.loopBound(srcArray.length); i += SPECIES.length()) {
104 var v = ShortVector.fromArray(SPECIES, srcArray, i);
105 v.intoArray(dstArray, i);
106 }
107 }
108
109 @Benchmark
110 public void vectAdd1() {
111 var a = this.a;
112 var b = this.b;
113 var c = this.c;
114
115 for (int i = 0; i < a.length; i += SPECIES.length()) {
116 ShortVector av = ShortVector.fromArray(SPECIES, a, i);
117 ShortVector bv = ShortVector.fromArray(SPECIES, b, i);
118 av.lanewise(VectorOperators.ADD, bv).intoArray(c, i);
119 }
120 }
121
122 @Benchmark
123 public void vectAdd2() {
124 var a = this.a;
125 var b = this.b;
126 var c = this.c;
127
128 for (int i = 0; i < a.length/SPECIES.length(); i++) {
129 ShortVector av = ShortVector.fromArray(SPECIES, a, (i*SPECIES.length()));
130 ShortVector bv = ShortVector.fromArray(SPECIES, b, (i*SPECIES.length()));
131 av.lanewise(VectorOperators.ADD, bv).intoArray(c, (i*SPECIES.length()));
132 }
133 }
134
135 @Benchmark
136 public void arrayAdd() {
137 for (int i = 0; i < SPECIES.loopBound(srcArray.length); i += SPECIES.length()) {
138 var v = ShortVector.fromArray(SPECIES, srcArray, i);
139 v = v.add(v);
140 v.intoArray(dstArray, i);
141 }
142 }
143
144 @Benchmark
145 public void heapSegment() {
146 for (long i = 0; i < SPECIES.loopBound(longSize); i += SPECIES.length()) {
147 var v = ShortVector.fromMemorySegment(SPECIES, srcSegmentHeap, i, ByteOrder.nativeOrder());
148 v.intoMemorySegment(dstSegmentHeap, i, ByteOrder.nativeOrder());
149 }
150 }
151
152 @Benchmark
153 public void segmentNativeImplicit() {
154 for (long i = 0; i < SPECIES.loopBound(srcArray.length); i += SPECIES.length()) {
155 var v = ShortVector.fromMemorySegment(SPECIES, srcSegment, i, ByteOrder.nativeOrder());
156 v.intoMemorySegment(dstSegment, i, ByteOrder.nativeOrder());
157 }
158 }
159
160 @Benchmark
161 public void segmentNativeConfined() {
162 try (final var arena = Arena.ofConfined()) {
163 final var srcSegmentConfined = srcSegment.reinterpret(arena, null);
164 final var dstSegmentConfined = dstSegment.reinterpret(arena, null);
165
166 for (long i = 0; i < SPECIES.loopBound(srcArray.length); i += SPECIES.length()) {
167 var v = ShortVector.fromMemorySegment(SPECIES, srcSegmentConfined, i, ByteOrder.nativeOrder());
168 v.intoMemorySegment(dstSegmentConfined, i, ByteOrder.nativeOrder());
169 }
170 }
171 }
172 }