I have the following situation:
In order to run a algorithm, i must run several threads and each thread will set a instance variable x, right before it dies. The problem is that these threads dont return immediately:
public Foo myAlgorithm()
{
//create n Runnables (n is big)
//start these runnables (may take long time do die)
//i need the x value of each runnable here, but they havent finished yet!
//get average x from all the runnables
return new Foo(averageX);
}
Should i use wait notify ? Or should i just embed a while loop and check for termination ?
Thanks everyone!
Create some shared storage to hold the x
value from each thread, or just store the sum if that's sufficient. Use a CountDownLatch
to wait for the threads to terminate. Each thread, when finished, would call CountDownLatch.countDown()
and your myAlgorithm
method would use the CountDownLatch.await()
method to wait for them.
Edit: Here's a complete example of the approach I suggested. It created 39 worker threads, each of which adds a random number to a shared sum. When all of the workers are finished, the average is computed and printed.
import java.util.Random;
import java.util.concurrent.CountDownLatch;
import java.util.concurrent.atomic.AtomicInteger;
class Worker implements Runnable {
private final AtomicInteger sum;
private final CountDownLatch latch;
public Worker(AtomicInteger sum, CountDownLatch latch) {
this.sum = sum;
this.latch = latch;
}
@Override
public void run() {
Random random = new Random();
try {
// Sleep a random length of time from 5-10s
Thread.sleep(random.nextInt(5000) + 5000);
} catch (InterruptedException e) {
e.printStackTrace();
}
// Compute x
int x = random.nextInt(500);
// Add to the shared sum
System.out.println("Adding " + x + " to sum");
sum.addAndGet(x);
// This runnable is finished, so count down
latch.countDown();
}
}
class Program {
public static void main(String[] args) {
// There will be 39 workers
final int N = 39;
// Holds the sum of all results from all workers
AtomicInteger sum = new AtomicInteger();
// Tracks how many workers are still working
CountDownLatch latch = new CountDownLatch(N);
System.out.println("Starting " + N + " workers");
for (int i = 0; i < N; i++) {
// Each worker uses the shared atomic sum and countdown latch.
Worker worker = new Worker(sum, latch);
// Start the worker
new Thread(worker).start();
}
try {
// Important: waits for all workers to finish.
latch.await();
} catch (InterruptedException e) {
e.printStackTrace();
}
// Compute the average
double average = (double) sum.get() / (double) N;
System.out.println(" Sum: " + sum.get());
System.out.println("Workers: " + N);
System.out.println("Average: " + average);
}
}
The output should be something like this:
Starting 39 workers
Adding 94 to sum
Adding 86 to sum
Adding 454 to sum
...
...
...
Adding 358 to sum
Adding 134 to sum
Adding 482 to sum
Sum: 10133
Workers: 39
Average: 259.8205128205128
Edit: Just for fun, here is an example using ExecutorService
, Callable
, and Future
.
import java.util.ArrayList;
import java.util.Collection;
import java.util.List;
import java.util.Random;
import java.util.concurrent.Callable;
import java.util.concurrent.ExecutionException;
import java.util.concurrent.ExecutorService;
import java.util.concurrent.Future;
import java.util.concurrent.ScheduledThreadPoolExecutor;
class Worker implements Callable<Integer> {
@Override
public Integer call() throws Exception {
Random random = new Random();
// Sleep a random length of time, from 5-10s
Thread.sleep(random.nextInt(5000) + 5000);
// Compute x
int x = random.nextInt(500);
System.out.println("Computed " + x);
return x;
}
}
public class Program {
public static void main(String[] args) {
// Thread pool size
final int POOL_SIZE = 10;
// There will be 39 workers
final int N = 39;
System.out.println("Starting " + N + " workers");
// Create the workers
Collection<Callable<Integer>> workers = new ArrayList<Callable<Integer>>(N);
for (int i = 0; i < N; i++) {
workers.add(new Worker());
}
// Create the executor service
ExecutorService executor = new ScheduledThreadPoolExecutor(POOL_SIZE);
// Execute all the workers, wait for the results
List<Future<Integer>> results = null;
try {
// Executes all tasks and waits for them to finish
results = executor.invokeAll(workers);
} catch (InterruptedException e) {
e.printStackTrace();
return;
}
// Compute the sum from the results
int sum = 0;
for (Future<Integer> future : results) {
try {
sum += future.get();
} catch (InterruptedException e) {
e.printStackTrace(); return;
} catch (ExecutionException e) {
e.printStackTrace(); return;
}
}
// Compute the average
double average = (double) sum / (double) N;
System.out.println(" Sum: " + sum);
System.out.println(" Workers: " + N);
System.out.println(" Average: " + average);
}
}
The output should look like this:
Starting 39 workers
Computed 419
Computed 36
Computed 338
...
...
...
Computed 261
Computed 354
Computed 112
Sum: 9526
Workers: 39
Average: 244.25641025641025
您可以使自己对java.util.concurrent.Future
以及所有相关联的东西(例如ThreadPools,Executors等)广为人知。Teaser: Future
是具有返回值的线程。
use a ExecutorService
and submit each task (as a Callable
) to it
you'll get a Future for each task submitted
List<Future<ResultType>> results = exec.invokeAll(tasks);//tasks is a set of Callable<ResultType>
//invokeAll blocks untill all tasks are finished
for(Future<ResultType> f:results){
ResultType x=f.get();//loop over Futures to get the result
//do something with x
}
The technical post webpages of this site follow the CC BY-SA 4.0 protocol. If you need to reprint, please indicate the site URL or the original address.Any question please contact:yoyou2525@163.com.