mirror of
https://github.com/openjdk/jdk.git
synced 2026-01-28 03:58:21 +00:00
JDK-8373026
This commit is contained in:
Emanuel Peter
parent
293fec7e28
commit
e734152cf4
@ -0,0 +1,50 @@
|
||||
/*
|
||||
* Copyright (c) 2025, Oracle and/or its affiliates. All rights reserved.
|
||||
* DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
|
||||
*
|
||||
* This code is free software; you can redistribute it and/or modify it
|
||||
* under the terms of the GNU General Public License version 2 only, as
|
||||
* published by the Free Software Foundation.
|
||||
*
|
||||
* This code is distributed in the hope that it will be useful, but WITHOUT
|
||||
* ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
|
||||
* FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License
|
||||
* version 2 for more details (a copy is included in the LICENSE file that
|
||||
* accompanied this code).
|
||||
*
|
||||
* You should have received a copy of the GNU General Public License version
|
||||
* 2 along with this work; if not, write to the Free Software Foundation,
|
||||
* Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA.
|
||||
*
|
||||
* Please contact Oracle, 500 Oracle Parkway, Redwood Shores, CA 94065 USA
|
||||
* or visit www.oracle.com if you need additional information or have any
|
||||
* questions.
|
||||
*/
|
||||
|
||||
/*
|
||||
* @test
|
||||
* @bug 8373026
|
||||
* @summary Test auto vectorization and Vector API with some vector
|
||||
* algorithms. Related benchmark: VectorAlgorithms.java
|
||||
* @library /test/lib /
|
||||
* @run driver ${test.main.class}
|
||||
*/
|
||||
|
||||
package compiler.loopopts.superword;
|
||||
|
||||
import compiler.lib.ir_framework.*;
|
||||
import compiler.lib.verify.*;
|
||||
|
||||
/**
|
||||
* The goal of this benchmark is to show the power of auto vectorization
|
||||
* and the Vector API.
|
||||
*
|
||||
* Please only modify this benchark in synchronization with the JMH benchmark:
|
||||
* micro/org/openjdk/bench/vm/compiler/VectorAlgorithms.java
|
||||
*/
|
||||
public class TestVectorAlgorithms {
|
||||
|
||||
public static void main(String[] args) {
|
||||
}
|
||||
}
|
||||
|
||||
166
test/micro/org/openjdk/bench/vm/compiler/VectorAlgorithms.java
Normal file
166
test/micro/org/openjdk/bench/vm/compiler/VectorAlgorithms.java
Normal file
@ -0,0 +1,166 @@
|
||||
/*
|
||||
* Copyright (c) 2025, Oracle and/or its affiliates. All rights reserved.
|
||||
* DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
|
||||
*
|
||||
* This code is free software; you can redistribute it and/or modify it
|
||||
* under the terms of the GNU General Public License version 2 only, as
|
||||
* published by the Free Software Foundation.
|
||||
*
|
||||
* This code is distributed in the hope that it will be useful, but WITHOUT
|
||||
* ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
|
||||
* FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License
|
||||
* version 2 for more details (a copy is included in the LICENSE file that
|
||||
* accompanied this code).
|
||||
*
|
||||
* You should have received a copy of the GNU General Public License version
|
||||
* 2 along with this work; if not, write to the Free Software Foundation,
|
||||
* Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA.
|
||||
*
|
||||
* Please contact Oracle, 500 Oracle Parkway, Redwood Shores, CA 94065 USA
|
||||
* or visit www.oracle.com if you need additional information or have any
|
||||
* questions.
|
||||
*
|
||||
*/
|
||||
|
||||
package org.openjdk.bench.vm.compiler;
|
||||
|
||||
import jdk.incubator.vector.*;
|
||||
import java.util.concurrent.TimeUnit;
|
||||
import org.openjdk.jmh.annotations.*;
|
||||
|
||||
/**
|
||||
* The goal of this benchmark is to show the power of auto vectorization
|
||||
* and the Vector API.
|
||||
*
|
||||
* Please only modify this benchark in synchronization with the IR test:
|
||||
* test/hotspot/jtreg/compiler/vectorization/TestVectorAlgorithms.java
|
||||
*/
|
||||
@BenchmarkMode(Mode.AverageTime)
|
||||
@OutputTimeUnit(TimeUnit.NANOSECONDS)
|
||||
@State(Scope.Thread)
|
||||
@Warmup(iterations = 2, time = 1)
|
||||
@Measurement(iterations = 3, time = 1)
|
||||
@Fork(value = 1, jvmArgs = {"--add-modules=jdk.incubator.vector"})
|
||||
public class VectorAlgorithms {
|
||||
private static final VectorSpecies<Integer> SPECIES_I = IntVector.SPECIES_512;
|
||||
|
||||
@Param({"640000"})
|
||||
public int SIZE;
|
||||
|
||||
public static int[] AI;
|
||||
public static int[] RI;
|
||||
|
||||
@Setup
|
||||
public void init() {
|
||||
AI = new int[SIZE];
|
||||
RI = new int[SIZE];
|
||||
}
|
||||
|
||||
//@Benchmark
|
||||
//public int reduceAddI_loop() {
|
||||
// int sum = 0;
|
||||
// for (int i = 0; i < AI.length; i++) {
|
||||
// sum += AI[i];
|
||||
// }
|
||||
// return sum;
|
||||
//}
|
||||
|
||||
//@Benchmark
|
||||
//public int reduceAddI_loop_reassociate() {
|
||||
// int sum = 0;
|
||||
// for (int i = 0; i < AI.length; i+=4) {
|
||||
// sum += (AI[i + 0] + AI[i + 1]) + (AI[i + 2] + AI[i + 3]);
|
||||
// }
|
||||
// return sum;
|
||||
//}
|
||||
|
||||
//@Benchmark
|
||||
//public int reduceAddI_VectorAPI_naive() {
|
||||
// var sum = 0;
|
||||
// for (int i = 0; i < SPECIES_I.loopBound(AI.length); i += SPECIES_I.length()) {
|
||||
// IntVector v = IntVector.fromArray(SPECIES_I, AI, i);
|
||||
// sum += v.reduceLanes(VectorOperators.ADD);
|
||||
// }
|
||||
// return sum;
|
||||
//}
|
||||
|
||||
//@Benchmark
|
||||
//public int reduceAddI_VectorAPI_reduction_after_loop() {
|
||||
// var acc = IntVector.broadcast(SPECIES_I, 0);
|
||||
// for (int i = 0; i < SPECIES_I.loopBound(AI.length); i += SPECIES_I.length()) {
|
||||
// IntVector v = IntVector.fromArray(SPECIES_I, AI, i);
|
||||
// acc = acc.add(v);
|
||||
// }
|
||||
// return acc.reduceLanes(VectorOperators.ADD);
|
||||
//}
|
||||
|
||||
//@Benchmark
|
||||
//public void scanAddI_loop() {
|
||||
// int sum = 0;
|
||||
// for (int i = 0; i < AI.length; i++) {
|
||||
// sum += AI[i];
|
||||
// RI[i] = sum;
|
||||
// }
|
||||
//}
|
||||
|
||||
//@Benchmark
|
||||
//public void scanAddI_loop_reassociate() {
|
||||
// int sum = 0;
|
||||
// for (int i = 0; i < AI.length; i+=4) {
|
||||
// // We cut the latency by a factor of 4, but increase the number of additions.
|
||||
// int old_sum = sum;
|
||||
// int v0 = AI[i + 0];
|
||||
// int v1 = AI[i + 1];
|
||||
// int v2 = AI[i + 2];
|
||||
// int v3 = AI[i + 3];
|
||||
// int v01 = v0 + v1;
|
||||
// int v23 = v2 + v3;
|
||||
// int v0123 = v01 + v23;
|
||||
// sum += v0123;
|
||||
// RI[i + 0] = old_sum + v0;
|
||||
// RI[i + 1] = old_sum + v01;
|
||||
// RI[i + 2] = old_sum + v01 + v2;
|
||||
// RI[i + 3] = old_sum + v0123;
|
||||
// }
|
||||
//}
|
||||
|
||||
//@Benchmark
|
||||
//public void scanAddI_VectorAPI_shift_blend_add() {
|
||||
// // Using Naive Parallel Algorithm: Hills and Steele
|
||||
// int sum = 0;
|
||||
// for (int i = 0; i < SPECIES_I.loopBound(AI.length); i += SPECIES_I.length()) {
|
||||
// IntVector v = IntVector.fromArray(SPECIES_I, AI, i);
|
||||
// v = v.add(v.lanewise(VectorOperators.LSHL, 1 ).blend(0, VectorMask.fromLong(SPECIES_I, 0b1111111111111110)));
|
||||
// v = v.add(v.lanewise(VectorOperators.LSHL, 2 ).blend(0, VectorMask.fromLong(SPECIES_I, 0b1111111111111100)));
|
||||
// v = v.add(v.lanewise(VectorOperators.LSHL, 4 ).blend(0, VectorMask.fromLong(SPECIES_I, 0b1111111111110000)));
|
||||
// v = v.add(v.lanewise(VectorOperators.LSHL, 8 ).blend(0, VectorMask.fromLong(SPECIES_I, 0b1111111100000000)));
|
||||
// v = v.add(sum);
|
||||
// v.intoArray(RI, i);
|
||||
// sum = v.lane(SPECIES_I.length() - 1);
|
||||
// }
|
||||
//}
|
||||
|
||||
//@Benchmark
|
||||
//public void scanAddI_VectorAPI_permute_add() {
|
||||
// // Using Naive Parallel Algorithm: Hills and Steele
|
||||
// int sum = 0;
|
||||
// var shf1 = VectorShuffle.fromArray(SPECIES_I, new int[]{-1, 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14}, 0);
|
||||
// var shf2 = VectorShuffle.fromArray(SPECIES_I, new int[]{-1, -1, 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13}, 0);
|
||||
// var shf3 = VectorShuffle.fromArray(SPECIES_I, new int[]{-1, -1, -1, -1, 0, 1, 2, 3, 4, 6, 7, 8, 9, 10, 11, 12}, 0);
|
||||
// var shf4 = VectorShuffle.fromArray(SPECIES_I, new int[]{-1, -1, -1, -1, -1, -1, -1, -1, 0, 1, 2, 3, 4, 6, 7, 8}, 0);
|
||||
// var mask1 = VectorMask.fromLong(SPECIES_I, 0b1111111111111110);
|
||||
// var mask2 = VectorMask.fromLong(SPECIES_I, 0b1111111111111100);
|
||||
// var mask3 = VectorMask.fromLong(SPECIES_I, 0b1111111111110000);
|
||||
// var mask4 = VectorMask.fromLong(SPECIES_I, 0b1111111100000000);
|
||||
// for (int i = 0; i < SPECIES_I.loopBound(AI.length); i += SPECIES_I.length()) {
|
||||
// IntVector v = IntVector.fromArray(SPECIES_I, AI, i);
|
||||
// v = v.add(v.rearrange(shf1), mask1);
|
||||
// v = v.add(v.rearrange(shf2), mask2);
|
||||
// v = v.add(v.rearrange(shf3), mask3);
|
||||
// v = v.add(v.rearrange(shf4), mask4);
|
||||
// v = v.add(sum);
|
||||
// v.intoArray(RI, i);
|
||||
// sum = v.lane(SPECIES_I.length() - 1);
|
||||
// }
|
||||
//}
|
||||
}
|
||||
Loading…
x
Reference in New Issue
Block a user