SynchronousQueue、LinkedBlockingQueue、ArrayBlockingQueue性能测试

　　JDK6对SynchronousQueue做了性能优化，避免对竞争资源加锁，所以想试试到底平时是选择SynchronousQueue还是其他BlockingQueue。

　　对于容器类在并发环境下的比较，一是是否线程安全，二是并发性能如何。BlockingQueue的实现都是线程安全的，所以只能比比它们的并发性能了。在不同的应用场景中，对容器的使用情况不同，有的读取操作多修改写入操作少，有的修改写入操作多，这对容器的性能会造成不同的影响。但对于Queue的使用，个人认为是比较一致的，简单点就是put和get，不会修改某个元素的内容再被读取，也很少只读取的操作，那是不是有最佳实践了?

　　LinkedBlockingQueue性能表现远超ArrayBlcokingQueue，不管线程多少，不管Queue长短，LinkedBlockingQueue都胜过ArrayBlockingQueue。SynchronousQueue表现很稳定，而且在20个线程之内不管Queue长短，SynchronousQueue性能表现是最好的，（其实SynchronousQueue跟 Queue长短没有关系），如果Queue的capability只能是1，那么毫无疑问选择SynchronousQueue，这也是设计 SynchronousQueue的目的吧。但大家也可以看到当超过1000个线程时，SynchronousQueue性能就直线下降了，只有最高峰的一半左右，而且当Queue大于30时，LinkedBlockingQueue性能就超过SynchronousQueue。

结论：

线程多（>20），Queue长度长（>30），使用LinkedBlockingQueue
线程少 (<20) ，Queue长度短 (<30) , 使用SynchronousQueue

当然，使用SynchronousQueue的时候不要忘记应用的扩展，如果将来需要进行扩展还是选择LinkedBlockingQueue好，尽量把SynchronousQueue限制在特殊场景中使用。

少用ArrayBlcokingQueue，似乎没找到它的好处，高手给给建议吧！

最后看看测试代码和结果：（Win7 64bit + JDK7 + CPU4 + 4GB)

import java.util.concurrent.ArrayBlockingQueue;

import java.util.concurrent.BlockingQueue;

import java.util.concurrent.Callable;

import java.util.concurrent.CompletionService;

import java.util.concurrent.ExecutorCompletionService;

import java.util.concurrent.ExecutorService;

import java.util.concurrent.Executors;

import java.util.concurrent.LinkedBlockingQueue;

import java.util.concurrent.SynchronousQueue;

public class TestSynchronousQueue {

private static int THREAD_NUM;

private static int N = 1000000;

private static ExecutorService executor;

public static void main(String[] args) throws Exception {

System.out.println("Producer\tConsumer\tcapacity \t LinkedBlockingQueue \t ArrayBlockingQueue \t SynchronousQueue");

for(int j = 0; j<10; j++){

THREAD_NUM = (int) Math.pow(2, j);

executor = Executors.newFixedThreadPool(THREAD_NUM * 2);

for (int i = 0; i < 10; i++) {

int length = (i == 0) ? 1 : i * 10;

System.out.print(THREAD_NUM + "\t\t");

System.out.print(length + "\t\t");

System.out.print(doTest2(new LinkedBlockingQueue<Integer>(length), N) + "/s\t\t\t");

System.out.print(doTest2(new ArrayBlockingQueue<Integer>(length), N) + "/s\t\t\t");

System.out.print(doTest2(new SynchronousQueue<Integer>(), N) + "/s");

System.out.println();

}

executor.shutdown();

}

private static class Producer implements Runnable{

int n;

BlockingQueue<Integer> q;

public Producer(int initN, BlockingQueue<Integer> initQ){

n = initN;

q = initQ;

}

public void run() {

for (int i = 0; i < n; i++)

try {

q.put(i);

} catch (InterruptedException ex) {

}

private static class Consumer implements Callable<Long>{

int n;

BlockingQueue<Integer> q;

public Consumer(int initN, BlockingQueue<Integer> initQ){

n = initN;

q = initQ;

}

public Long call() {

long sum = 0;

for (int i = 0; i < n; i++)

try {

sum += q.take();

} catch (InterruptedException ex) {

}

return sum;

}

private static long doTest2(final BlockingQueue<Integer> q, final int n)

throws Exception {

CompletionService<Long> completionServ = new ExecutorCompletionService<Long>(executor);

long t = System.nanoTime();

for(int i=0; i<THREAD_NUM; i++){

executor.submit(new Producer(n/THREAD_NUM, q));

}

for(int i=0; i<THREAD_NUM; i++){

completionServ.submit(new Consumer(n/THREAD_NUM, q));

}

for(int i=0; i<THREAD_NUM; i++){

completionServ.take().get();

}

t = System.nanoTime() - t;

return (long) (1000000000.0 * N / t); // Throughput, items/sec

}

程序运行结果：

100

101

Producer    Consumer    capacity     LinkedBlockingQueue     ArrayBlockingQueue      SynchronousQueue

1 1 1 154567/s 154100/s 3655071/s

1 1 10 1833165/s 1967491/s 3622405/s

1 1 20 3011779/s 2558451/s 3744037/s

1 1 30 3145926/s 2632099/s 3354525/s

1 1 40 3289673/s 2879696/s 3581858/s

1 1 50 3201828/s 3008838/s 3600100/s

1 1 60 3171374/s 2541672/s 3922617/s

1 1 70 3159786/s 2844493/s 3423066/s

1 1 80 3042835/s 2536290/s 3443517/s

1 1 90 3025808/s 3026241/s 3307096/s

2 2 1 141555/s 135653/s 2897927/s

2 2 10 1627066/s 785082/s 2908671/s

2 2 20 2199668/s 1604847/s 2937085/s

2 2 30 2309495/s 2115986/s 2922671/s

2 2 40 2335737/s 2424491/s 2942621/s

2 2 50 2394045/s 2405210/s 2918222/s

2 2 60 2499445/s 2471052/s 2881591/s

2 2 70 2368143/s 2454153/s 2914038/s

2 2 80 2381024/s 2457910/s 2937337/s

2 2 90 2509167/s 2461035/s 2789278/s

4 4 1 138177/s 138101/s 2736238/s

4 4 10 1654165/s 478171/s 2693045/s

4 4 20 2443373/s 779452/s 2728493/s

4 4 30 2646300/s 1169313/s 2787315/s

4 4 40 2755774/s 1487883/s 2874789/s

4 4 50 2774736/s 1579152/s 2804046/s

4 4 60 2804725/s 1998602/s 2803680/s

4 4 70 2797524/s 2388276/s 2936613/s

4 4 80 2887786/s 2557358/s 2899823/s

4 4 90 2878895/s 2539458/s 2839990/s

8 8 1 140745/s 135621/s 2711703/s

8 8 10 1650143/s 526018/s 2730710/s

8 8 20 2477902/s 798799/s 2696374/s

8 8 30 2658511/s 983456/s 2783054/s

8 8 40 2694167/s 1185732/s 2677500/s

8 8 50 2758267/s 1110716/s 2766695/s

8 8 60 2831922/s 1003692/s 2762232/s

8 8 70 2763751/s 1409142/s 2791901/s

8 8 80 2771897/s 1654843/s 2838479/s

8 8 90 2740467/s 1718642/s 2806164/s

16 16 1 131843/s 137943/s 2694036/s

16 16 10 1637213/s 491171/s 2725893/s

16 16 20 2523193/s 660559/s 2709892/s

16 16 30 2601176/s 899163/s 2689270/s

16 16 40 2794088/s 1054763/s 2759321/s

16 16 50 2777807/s 1111479/s 2663346/s

16 16 60 2893566/s 931713/s 2778294/s

16 16 70 2822779/s 1286067/s 2704785/s

16 16 80 2828238/s 1430581/s 2724927/s

16 16 90 2860943/s 1249650/s 2791520/s

32 32 1 132098/s 130805/s 2676121/s

32 32 10 1586372/s 402270/s 2674953/s

32 32 20 2467754/s 886059/s 2580989/s

32 32 30 2569709/s 772173/s 2599466/s

32 32 40 2659883/s 963633/s 2677042/s

32 32 50 2721213/s 910607/s 2677578/s

32 32 60 2779272/s 861786/s 2676874/s

32 32 70 2757921/s 1111937/s 2696416/s

32 32 80 2915294/s 1323776/s 2655641/s

32 32 90 2798313/s 1193225/s 2630231/s

64 64 1 126035/s 123764/s 2526632/s

64 64 10 1539034/s 394597/s 2582590/s

64 64 20 2449850/s 703790/s 2598631/s

64 64 30 2672792/s 758256/s 2529693/s

64 64 40 2797081/s 661028/s 2573380/s

64 64 50 2789848/s 1162143/s 2659469/s

64 64 60 2726806/s 1145495/s 2567020/s

64 64 70 2731554/s 1359939/s 2607615/s

64 64 80 2871116/s 1305428/s 2494839/s

64 64 90 2774416/s 1339611/s 2560153/s

128 128 1 223305/s 112828/s 2390234/s

128 128 10 1419592/s 404611/s 2401086/s

128 128 20 2365301/s 793815/s 2516045/s

128 128 30 2647136/s 915702/s 2463175/s

128 128 40 2721664/s 1081728/s 2400299/s

128 128 50 2688304/s 1149251/s 2489667/s

128 128 60 2774212/s 1145298/s 2453444/s

128 128 70 2782905/s 1165408/s 2403510/s

128 128 80 2818388/s 1392486/s 2389275/s

128 128 90 2738468/s 1546247/s 2425994/s

256 256 1 160146/s 80530/s 2369297/s

256 256 10 1214041/s 364460/s 2142039/s

256 256 20 1915432/s 901668/s 2156774/s

256 256 30 2371862/s 1124997/s 2237464/s

256 256 40 2630812/s 1123016/s 2216475/s

256 256 50 2666827/s 1239640/s 2267322/s

256 256 60 2635269/s 1276851/s 2318122/s

256 256 70 2663477/s 1333002/s 2188256/s

256 256 80 2672080/s 1659850/s 2315438/s

256 256 90 2804828/s 1497635/s 2194905/s

512 512 1 123294/s 68426/s 1892168/s

512 512 10 1028250/s 296454/s 1728199/s

512 512 20 1545215/s 604512/s 1963526/s

512 512 30 1968728/s 762240/s 2000386/s

512 512 40 2273678/s 854483/s 1948188/s

512 512 50 2295335/s 939350/s 1858429/s

512 512 60 2419257/s 1056918/s 1884224/s

512 512 70 2346088/s 980795/s 1852387/s

512 512 80 2341964/s 928496/s 1867498/s

512 512 90 2375789/s 1290064/s 1923461/s

秒客网

SynchronousQueue、LinkedBlockingQueue、ArrayBlockingQueue性能测试

相关文章