author | thartmann |
Wed, 10 Feb 2016 15:24:22 +0100 | |
changeset 36057 | 867e857daf84 |
parent 30588 | 24fc4b3a964e |
child 40059 | c2304140ed64 |
permissions | -rw-r--r-- |
30588 | 1 |
/* |
2 |
* Copyright (c) 2015, 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 |
||
25 |
/** |
|
26 |
* @test |
|
27 |
* @bug 8074981 |
|
28 |
* @summary Add C2 x86 Superword support for scalar product reduction optimizations : int test |
|
36057
867e857daf84
8149123: [TESTBUG] compiler/loopopts/superword/SumRed* tests running on non-x86 platforms
thartmann
parents:
30588
diff
changeset
|
29 |
* @requires os.arch=="x86" | os.arch=="i386" | os.arch=="amd64" | os.arch=="x86_64" | os.arch=="aarch64" |
30588 | 30 |
* |
31 |
* @run main/othervm -XX:+IgnoreUnrecognizedVMOptions -XX:+SuperWordReductions -XX:LoopUnrollLimit=250 -XX:CompileThresholdScaling=0.1 -XX:CompileCommand=exclude,ReductionPerf::main ReductionPerf |
|
32 |
* @run main/othervm -XX:+IgnoreUnrecognizedVMOptions -XX:-SuperWordReductions -XX:LoopUnrollLimit=250 -XX:CompileThresholdScaling=0.1 -XX:CompileCommand=exclude,ReductionPerf::main ReductionPerf |
|
33 |
*/ |
|
34 |
||
35 |
public class ReductionPerf { |
|
36 |
public static void main(String[] args) throws Exception { |
|
37 |
int[] a1 = new int[8*1024]; |
|
38 |
int[] a2 = new int[8*1024]; |
|
39 |
int[] a3 = new int[8*1024]; |
|
40 |
long[] b1 = new long[8*1024]; |
|
41 |
long[] b2 = new long[8*1024]; |
|
42 |
long[] b3 = new long[8*1024]; |
|
43 |
float[] c1 = new float[8*1024]; |
|
44 |
float[] c2 = new float[8*1024]; |
|
45 |
float[] c3 = new float[8*1024]; |
|
46 |
double[] d1 = new double[8*1024]; |
|
47 |
double[] d2 = new double[8*1024]; |
|
48 |
double[] d3 = new double[8*1024]; |
|
49 |
||
50 |
ReductionInit(a1,a2,a3,b1,b2,b3,c1,c2,c3,d1,d2,d3); |
|
51 |
||
52 |
int sumIv = sumInt(a1,a2,a3); |
|
53 |
long sumLv = sumLong(b1,b2,b3); |
|
54 |
float sumFv = sumFloat(c1,c2,c3); |
|
55 |
double sumDv = sumDouble(d1,d2,d3); |
|
56 |
int mulIv = prodInt(a1,a2,a3); |
|
57 |
long mulLv = prodLong(b1,b2,b3); |
|
58 |
float mulFv = prodFloat(c1,c2,c3); |
|
59 |
double mulDv = prodDouble(d1,d2,d3); |
|
60 |
||
61 |
int sumI = 0; |
|
62 |
long sumL = 0; |
|
63 |
float sumF = 0.f; |
|
64 |
double sumD = 0.; |
|
65 |
int mulI = 0; |
|
66 |
long mulL = 0; |
|
67 |
float mulF = 0.f; |
|
68 |
double mulD = 0.; |
|
69 |
||
70 |
System.out.println("Warmup ..."); |
|
71 |
long start = System.currentTimeMillis(); |
|
72 |
||
73 |
for(int j = 0; j < 2000; j++) { |
|
74 |
sumI = sumInt(a1,a2,a3); |
|
75 |
sumL = sumLong(b1,b2,b3); |
|
76 |
sumF = sumFloat(c1,c2,c3); |
|
77 |
sumD = sumDouble(d1,d2,d3); |
|
78 |
mulI = prodInt(a1,a2,a3); |
|
79 |
mulL = prodLong(b1,b2,b3); |
|
80 |
mulF = prodFloat(c1,c2,c3); |
|
81 |
mulD = prodDouble(d1,d2,d3); |
|
82 |
} |
|
83 |
||
84 |
long stop = System.currentTimeMillis(); |
|
85 |
System.out.println(" Warmup is done in " + (stop - start) + " msec"); |
|
86 |
||
87 |
if (sumIv != sumI) { |
|
88 |
System.out.println("sum int: " + sumIv + " != " + sumI); |
|
89 |
} |
|
90 |
if (sumLv != sumL) { |
|
91 |
System.out.println("sum long: " + sumLv + " != " + sumL); |
|
92 |
} |
|
93 |
if (sumFv != sumF) { |
|
94 |
System.out.println("sum float: " + sumFv + " != " + sumF); |
|
95 |
} |
|
96 |
if (sumDv != sumD) { |
|
97 |
System.out.println("sum double: " + sumDv + " != " + sumD); |
|
98 |
} |
|
99 |
if (mulIv != mulI) { |
|
100 |
System.out.println("prod int: " + mulIv + " != " + mulI); |
|
101 |
} |
|
102 |
if (mulLv != mulL) { |
|
103 |
System.out.println("prod long: " + mulLv + " != " + mulL); |
|
104 |
} |
|
105 |
if (mulFv != mulF) { |
|
106 |
System.out.println("prod float: " + mulFv + " != " + mulF); |
|
107 |
} |
|
108 |
if (mulDv != mulD) { |
|
109 |
System.out.println("prod double: " + mulDv + " != " + mulD); |
|
110 |
} |
|
111 |
||
112 |
start = System.currentTimeMillis(); |
|
113 |
for (int j = 0; j < 5000; j++) { |
|
114 |
sumI = sumInt(a1, a2 ,a3); |
|
115 |
} |
|
116 |
stop = System.currentTimeMillis(); |
|
117 |
System.out.println("sum int: " + (stop - start)); |
|
118 |
||
119 |
start = System.currentTimeMillis(); |
|
120 |
for (int j = 0; j < 5000; j++) { |
|
121 |
sumL = sumLong(b1, b2, b3); |
|
122 |
} |
|
123 |
stop = System.currentTimeMillis(); |
|
124 |
System.out.println("sum long: " + (stop - start)); |
|
125 |
||
126 |
start = System.currentTimeMillis(); |
|
127 |
for (int j = 0; j < 5000; j++) { |
|
128 |
sumF = sumFloat(c1, c2, c3); |
|
129 |
} |
|
130 |
stop = System.currentTimeMillis(); |
|
131 |
System.out.println("sum float: " + (stop - start)); |
|
132 |
||
133 |
start = System.currentTimeMillis(); |
|
134 |
for (int j = 0; j < 5000; j++) { |
|
135 |
sumD = sumDouble(d1, d2, d3); |
|
136 |
} |
|
137 |
stop = System.currentTimeMillis(); |
|
138 |
System.out.println("sum double: " + (stop - start)); |
|
139 |
||
140 |
start = System.currentTimeMillis(); |
|
141 |
for (int j = 0; j < 5000; j++) { |
|
142 |
mulI = prodInt(a1, a2, a3); |
|
143 |
} |
|
144 |
stop = System.currentTimeMillis(); |
|
145 |
System.out.println("prod int: " + (stop - start)); |
|
146 |
||
147 |
start = System.currentTimeMillis(); |
|
148 |
for (int j = 0; j < 5000; j++) { |
|
149 |
mulL = prodLong(b1, b2 ,b3); |
|
150 |
} |
|
151 |
stop = System.currentTimeMillis(); |
|
152 |
System.out.println("prod long: " + (stop - start)); |
|
153 |
||
154 |
start = System.currentTimeMillis(); |
|
155 |
for (int j = 0; j < 5000; j++) { |
|
156 |
mulF = prodFloat(c1, c2, c3); |
|
157 |
} |
|
158 |
stop = System.currentTimeMillis(); |
|
159 |
System.out.println("prod float: " + (stop - start)); |
|
160 |
||
161 |
start = System.currentTimeMillis(); |
|
162 |
for (int j = 0; j < 5000; j++) { |
|
163 |
mulD = prodDouble(d1, d2, d3); |
|
164 |
} |
|
165 |
stop = System.currentTimeMillis(); |
|
166 |
System.out.println("prod double: " + (stop - start)); |
|
167 |
||
168 |
} |
|
169 |
||
170 |
public static void ReductionInit(int[] a1, int[] a2, int[] a3, |
|
171 |
long[] b1, long[] b2, long[] b3, |
|
172 |
float[] c1, float[] c2, float[] c3, |
|
173 |
double[] d1, double[] d2, double[] d3 ) { |
|
174 |
for(int i = 0; i < a1.length; i++) { |
|
175 |
a1[i] = (i + 0); |
|
176 |
a2[i] = (i + 1); |
|
177 |
a3[i] = (i + 2); |
|
178 |
b1[i] = (long) (i + 0); |
|
179 |
b2[i] = (long) (i + 1); |
|
180 |
b3[i] = (long) (i + 2); |
|
181 |
c1[i] = (float) (i + 0); |
|
182 |
c2[i] = (float) (i + 1); |
|
183 |
c3[i] = (float) (i + 2); |
|
184 |
d1[i] = (double) (i + 0); |
|
185 |
d2[i] = (double) (i + 1); |
|
186 |
d3[i] = (double) (i + 2); |
|
187 |
} |
|
188 |
} |
|
189 |
||
190 |
public static int sumInt(int[] a1, int[] a2, int[] a3) { |
|
191 |
int total = 0; |
|
192 |
for(int i = 0; i < a1.length; i++) { |
|
193 |
total += (a1[i] * a2[i]) + (a1[i] * a3[i]) + (a2[i] * a3[i]); |
|
194 |
} |
|
195 |
return total; |
|
196 |
} |
|
197 |
||
198 |
public static long sumLong(long[] b1, long[] b2, long[] b3) { |
|
199 |
long total = 0; |
|
200 |
for(int i = 0; i < b1.length; i++) { |
|
201 |
total += (b1[i] * b2[i]) + (b1[i] * b3[i]) + (b2[i] * b3[i]); |
|
202 |
} |
|
203 |
return total; |
|
204 |
} |
|
205 |
||
206 |
public static float sumFloat(float[] c1, float[] c2, float[] c3) { |
|
207 |
float total = 0; |
|
208 |
for(int i = 0; i < c1.length; i++) { |
|
209 |
total += (c1[i] * c2[i]) + (c1[i] * c3[i]) + (c2[i] * c3[i]); |
|
210 |
} |
|
211 |
return total; |
|
212 |
} |
|
213 |
||
214 |
public static double sumDouble(double[] d1, double[] d2, double[] d3) { |
|
215 |
double total = 0; |
|
216 |
for(int i = 0; i < d1.length; i++) { |
|
217 |
total += (d1[i] * d2[i]) + (d1[i] * d3[i]) + (d2[i] * d3[i]); |
|
218 |
} |
|
219 |
return total; |
|
220 |
} |
|
221 |
||
222 |
public static int prodInt(int[] a1, int[] a2, int[] a3) { |
|
223 |
int total = 1; |
|
224 |
for(int i = 0; i < a1.length; i++) { |
|
225 |
total *= (a1[i] * a2[i]) + (a1[i] * a3[i]) + (a2[i] * a3[i]); |
|
226 |
} |
|
227 |
return total; |
|
228 |
} |
|
229 |
||
230 |
public static long prodLong(long[] b1, long[] b2, long[] b3) { |
|
231 |
long total = 1; |
|
232 |
for(int i = 0; i < b1.length; i++) { |
|
233 |
total *= (b1[i] * b2[i]) + (b1[i] * b3[i]) + (b2[i] * b3[i]); |
|
234 |
} |
|
235 |
return total; |
|
236 |
} |
|
237 |
||
238 |
public static float prodFloat(float[] c1, float[] c2, float[] c3) { |
|
239 |
float total = 1; |
|
240 |
for(int i = 0; i < c1.length; i++) { |
|
241 |
total *= (c1[i] * c2[i]) + (c1[i] * c3[i]) + (c2[i] * c3[i]); |
|
242 |
} |
|
243 |
return total; |
|
244 |
} |
|
245 |
||
246 |
public static double prodDouble(double[] d1, double[] d2, double[] d3) { |
|
247 |
double total = 1; |
|
248 |
for(int i = 0; i < d1.length; i++) { |
|
249 |
total *= (d1[i] * d2[i]) + (d1[i] * d3[i]) + (d2[i] * d3[i]); |
|
250 |
} |
|
251 |
return total; |
|
252 |
} |
|
253 |
} |