I have a program that loads slowly, which I guess is due to the amount of image resources I have to load at the beginning. I thought multi-threading would help, but now I'm not so sure. Here is my automatic multi-threading method.
private static Thread[] t;
private static int currentThreads;
public static void loadWithThreads(Object[] array, IntegerRunnable r) {
final int threads = Runtime.getRuntime().availableProcessors();
t = new Thread[threads];
for (int i = 0; i < threads; i ++) {
t[i] = new Thread("HMediaConverter") {
final int id = currentThreads;
int items = (array.length / threads) * currentThreads;
#Override
public void run() {
super.run();
for (int i = items; i < (items + (array.length / threads)); i ++) {
r.run(i);
}
//Recycle this thread so it can be used for another time.
try {
t[id].join();
lock.notifyAll();
currentThreads --;
} catch (InterruptedException e) {
e.printStackTrace();
}
}
};
t[i].setPriority(Thread.MAX_PRIORITY);
t[i].start();
currentThreads ++;
}
}
And here is my image loading code:
public static ImageIcon loadImageIcon(String path) {
return new ImageIcon(ImageIO.read(Tools.class.getClassLoader().getResource(path));
}
Surely there is a way to speed things up? I'm running this on a perfectly good Intel i5, it shouldn't be this slow, so it must be my code.
Loading 113 images of a total of 159.14mb with...
public static void loadWithoutThreads(File[] array) {
for (File file : array) {
try {
ImageIO.read(file);
} catch (IOException ex) {
ex.printStackTrace();
}
}
}
Took ~15s
With...
public static void loadWithThreads(File[] array) {
final int threads = Runtime.getRuntime().availableProcessors();
t = new Thread[threads];
CountDownLatch latch = new CountDownLatch(threads);
for (int i = 0; i < threads; i++) {
t[i] = new Thread("HMediaConverter") {
final int id = currentThreads;
int items = (array.length / threads) * currentThreads;
#Override
public void run() {
try {
System.out.println("Starting " + id);
for (int i = items; i < (items + (array.length / threads)); i++) {
try {
System.out.println(i + ": " + array[i]);
ImageIO.read(array[i]);
} catch (IOException ex) {
ex.printStackTrace();
}
}
} finally {
latch.countDown();
}
}
};
t[i].setPriority(Thread.MAX_PRIORITY);
System.out.println("Start " + i);
t[i].start();
currentThreads++;
}
try {
latch.await();
} catch (InterruptedException ex) {
ex.printStackTrace();
}
}
took ~11s
With...
public static void loadWithExecutor(File[] images) {
ExecutorService service = Executors.newFixedThreadPool(2);
List<ImageLoadingTask> tasks = new ArrayList<>(images.length);
for (File file : images) {
tasks.add(new ImageLoadingTask(file));
}
try {
List<Future<BufferedImage>> results = service.invokeAll(tasks);
} catch (InterruptedException ex) {
ex.printStackTrace();
}
service.shutdown();
}
public static class ImageLoadingTask implements Callable<BufferedImage> {
private File file;
public ImageLoadingTask(File file) {
this.file = file;
}
#Override
public BufferedImage call() throws Exception {
return ImageIO.read(file);
}
}
Took ~7s
The ExecutorService is more efficient because when one thread is processing a larger file, the other can be processing a number of small files. This is achieved by pooling the threads that aren't doing any work until they are needed, allowing a thread to perform a lot of short work, while the other thread(s) are also busy. You don't need to wait as long
Have a look at Executors for more details
The following is a re-write that should work that is close to what the op wrote. A re-write into A fixed-size thread pool would probably be better.
//import java.util.concurrent.atomic.AtomicInteger;
private static Thread[] t;
private static AtomicInteger completedLoads = new AtomicInteger(0);
public static void loadWithThreads(Object[] array, IntegerRunnable r) {
final int threads = Runtime.getRuntime().availableProcessors();
t = new Thread[threads];
completedLoads = new AtomicInteger(0);
int targetLoads = array.length;
int itemsPerThread = (array.length / threads);
for (int i = 0; i < threads; i ++) {
t[i] = new Thread("HMediaConverter" + i) {
int startItem = itemsPerThread * i;
#Override
public void run() {
super.run();
for (int i = startItem; i < startItem + itemsPerThread; i ++) {
try {
r.run(i);
}
finally {
completedLoads.incrementAndGet();
}
}
}
};
t[i].setPriority(Thread.MAX_PRIORITY);
t[i].start();
}
// Wait for the images to load
while (completedLoads.get() < targetLoads)
{
try {
Thread.sleep(100);
}
catch (InterruptedException ie) {
// ignore
}
}
}
Isolate which part does the slowing down - e.g by running System.currentTimeMillis() btween major segmnst then show us where is the biggest time - or show us all the program.
Threads handling as noted is questionable and you shouldn't use methods such as join etc out of the box unless you have seen it sometwhere provably working.
So post times and we'll take it from there - it could be the images it could be the threads
Related
I am trying out the Producer-Consumer problem using Semaphore. The program looks fine to me except for one place.
public class ProducerConsumerWithSemaphores
{
private final ArrayList<Integer> list = new ArrayList<>(5);
private final Semaphore semaphoreProducer = new Semaphore(1);
private final Semaphore semaphoreConsumer = new Semaphore(0);
private void produce() throws InterruptedException
{
for(int i = 0;i< 5;i++)
{
semaphoreProducer.acquire();
list.add(i);
System.out.println("Produced: " + i);
semaphoreConsumer.release();
}
}
private void consumer() throws InterruptedException
{
while (!list.isEmpty()) /// This line is where I have the doubt
{
semaphoreConsumer.acquire();
System.out.println("Consumer: " + list.remove(list.size()-1));
semaphoreProducer.release();
Thread.sleep(100);
}
}
public static void main(String[] args)
{
final ProducerConsumerWithSemaphores obj = new ProducerConsumerWithSemaphores();
new Thread(new Runnable()
{
#Override
public void run()
{
try
{
obj.produce();
} catch (InterruptedException e)
{
e.printStackTrace();
}
}
}).start();
new Thread(new Runnable()
{
#Override
public void run()
{
try
{
obj.consumer();
} catch (InterruptedException e)
{
e.printStackTrace();
}
}
}).start();
}
}
Is it okay to check the list if it is not empty before acquiring the semaphore? Will this cause any problem in multithreaded environment?
private void consumer() throws InterruptedException
{
while (!list.isEmpty()) /// This line is where I have the doubt
The problem is, if consumer runs faster than producer, your consumer quit immediately, then you have no consumer!!
The correct example looks like,
Producer–consumer problem#Using semaphores. I believe your intention is not to use true as endless loop because you want Producer/Consumer to quit when job is done. If that's your intention, you can 1. set a totalCount to end the loop. 2. Or a boolean flag which will be set by producer after putItemIntoBuffer when producer put the last one. The flag must be protected as well as the buffer.(update: this method doesn't work if there's multiple producers/consumers) 3. Simulate EOF ( idea taken from producer - consume; how does the consumer stop?)
Will this cause any problem in multithreaded environment?
Your critical section (your list) is not protected . Usually we use 3 semaphores. The 3rd one is used as a mutex to protect the buffer.
To stop producers/consumers,
Example code with method 1:
public class Test3 {
private Semaphore mutex = new Semaphore(1);
private Semaphore fillCount = new Semaphore(0);
private Semaphore emptyCount = new Semaphore(3);
private final List<Integer> list = new ArrayList<>();
class Producer implements Runnable {
private final int totalTasks;
Producer(int totalTasks) {
this.totalTasks = totalTasks;
}
#Override
public void run() {
try {
for (int i = 0; i < totalTasks; i++) {
emptyCount.acquire();
mutex.acquire();
list.add(i);
System.out.println("Produced: " + i);
mutex.release();
fillCount.release();
}
} catch (InterruptedException e) {
e.printStackTrace();
}
}
}
class Consumer implements Runnable {
private final int totalTasks;
Consumer(int totalTasks) {
this.totalTasks = totalTasks;
}
#Override
public void run() {
try {
for (int i = 0; i < totalTasks; i++) {
fillCount.acquire();
mutex.acquire();
int item = list.remove(list.size() - 1);
System.out.println("Consumed: " + item);
mutex.release();
emptyCount.release();
}
} catch (InterruptedException e) {
e.printStackTrace();
}
}
}
public void runTest() {
int numProducer = 3;
int tasksPerProducer = 10;
int numConsumer = 6;
int tasksPerConsumer = 5;
for (int i = 0; i < numProducer; i++) {
new Thread(new Producer(tasksPerProducer)).start();
}
for (int i = 0; i < numConsumer; i++) {
new Thread(new Consumer(tasksPerConsumer)).start();
}
}
public static void main(String[] args) throws IOException {
Test3 t = new Test3();
t.runTest();
}
}
Example code with method 3:
public class Test4 {
private Semaphore mutex = new Semaphore(1);
private Semaphore fillCount = new Semaphore(0);
private Semaphore emptyCount = new Semaphore(3);
private Integer EOF = Integer.MAX_VALUE;
private final Queue<Integer> list = new LinkedList<>(); // need to put/get data in FIFO
class Producer implements Runnable {
private final int totalTasks;
Producer(int totalTasks) {
this.totalTasks = totalTasks;
}
#Override
public void run() {
try {
for (int i = 0; i < totalTasks + 1; i++) {
emptyCount.acquire();
mutex.acquire();
if (i == totalTasks) {
list.offer(EOF);
} else {
// add a valid value
list.offer(i);
System.out.println("Produced: " + i);
}
mutex.release();
fillCount.release();
}
} catch (InterruptedException e) {
e.printStackTrace();
}
}
}
class Consumer implements Runnable {
#Override
public void run() {
try {
boolean finished = false;
while (!finished) {
fillCount.acquire();
mutex.acquire();
int item = list.poll();
if (EOF.equals(item)) {
// do not consume this item because it means EOF
finished = true;
} else {
// it's a valid value, consume it.
System.out.println("Consumed: " + item);
}
mutex.release();
emptyCount.release();
}
} catch (InterruptedException e) {
e.printStackTrace();
}
}
}
public void runTest() {
int numProducer = 3;
int tasksPerProducer = 10;
for (int i = 0; i < numProducer; i++) {
new Thread(new Producer(tasksPerProducer)).start();
}
int numConsumer = numProducer; // producers will put N EOFs to kill N consumers.
for (int i = 0; i < numConsumer; i++) {
new Thread(new Consumer()).start();
}
}
public static void main(String[] args) throws IOException {
Test4 t = new Test4();
t.runTest();
}
}
Instead of using two semaphores why dont you use a single semaphore to such that the synchronization is made between threads link
Additional you can use ArrayBlockingQueue which are thread safe to properly demonstrate the Producer Consumer Problem.
I want to print a series of 1 to 100 number using n number of threads (lets take 10 threads for this). Condition is 1st thread will have a sequence number from 1, 11,21....91, 2nd thread will have a sequence 2,12,22.....92 and so on. All other threads will have a sequence number like that. Now I want to print number in sequence 1 to 100. I know we can use synchronization, wait and notify method and using a variable or flag counter but I don't think this is a good idea to use it. I want to use without concurrency (like executors etc) how will I do that. Please suggest.
public class PrintNumberSequenceUsingRunnable {
int notifyValue = 1;
public static void main(String[] args) {
PrintNumberSequenceUsingRunnable sequence = new PrintNumberSequenceUsingRunnable();
Thread f = new Thread(new First(sequence), "Fisrt");
Thread s = new Thread(new Second(sequence), "Second");
Thread t = new Thread(new Third(sequence), "Third");
f.start();
s.start();
t.start();
}
}
class First implements Runnable {
PrintNumberSequenceUsingRunnable sequence;
public First(PrintNumberSequenceUsingRunnable sequence) {
this.sequence = sequence;
}
#Override
public void run() {
printFist();
}
private void printFist() {
synchronized (sequence) {
for (int i = 1; i <= 20; i += 3) {
while (sequence.notifyValue != 1) {
try {
sequence.wait();
} catch (InterruptedException e) {
// TODO Auto-generated catch block
e.printStackTrace();
}
}
System.out.println(Thread.currentThread().getName() + " " + i);
sequence.notifyValue = 2;
sequence.notifyAll();
}
}
}
}
class Second implements Runnable {
PrintNumberSequenceUsingRunnable sequence;
public Second(PrintNumberSequenceUsingRunnable sequence) {
this.sequence = sequence;
}
#Override
public void run() {
printSecond();
}
private void printSecond() {
synchronized (sequence) {
for (int i = 2; i <= 20; i += 3) {
while (sequence.notifyValue != 2) {
try {
sequence.wait();
} catch (InterruptedException e) {
// TODO Auto-generated catch block
e.printStackTrace();
}
}
System.out.println(Thread.currentThread().getName() + " " + i);
sequence.notifyValue = 3;
sequence.notifyAll();
}
}
}
}
class Third implements Runnable {
PrintNumberSequenceUsingRunnable sequence;
public Third(PrintNumberSequenceUsingRunnable sequence) {
this.sequence = sequence;
}
#Override
public void run() {
printThrid();
}
private void printThrid() {
synchronized (sequence) {
for (int i = 3; i <= 20; i += 3) {
while (sequence.notifyValue != 3) {
try {
sequence.wait();
} catch (InterruptedException e) {
// TODO Auto-generated catch block
e.printStackTrace();
}
}
System.out.println(Thread.currentThread().getName() + " " + i);
sequence.notifyValue = 1;
sequence.notifyAll();
}
}
}
}
You need to have values sorted on each threads. Each time a thread writes a number, it triggers an event in an event bus. All threads are subscribed to the event.
You start the system by triggering the event [minimum value - 1].
Each thread will receive a notification that the value [minimum value - 1] has been published. Only the thread that has the value [minimum value] will act and will trigger a new event for value [minimum value + 1].
Edit: I haven't tested it, but something like this.
static void main(String[] args) {
List<Deque<Integer>> publishQueues = new ArrayList<>();
for (int i = 1; i <= 10; i++) {
new Thread(new Worker(i, publishQueues)).start();
}
}
class Worker implements Runnable {
Deque subscriberQueue;
List<Deque<Integer>> publishQueues;
int i;
Worker(int i, List<Deque<Integer>> publishQueues) {
this.i = i;
this.publishQueues = publishQueues;
this.subscriberQueue = new ConcurrentLinkedDeque<>();
this.publishQueues.add(this.subscriberQueue);
}
void Run() {
LinkedList<Integer> ints = new LinkedList<>();
for (int j = i; j <= 100; j+=10) {
ints.add(j);
}
while (true) {
Integer publishedInteger = subscriberQueue.poll();
if (publishedInteger == ints.getFirst() - 1) {
Integer integer = ints.poll();
System.out.println(integer);
for (Dequeu<Integer> publishQueue : publishQueues) {
publishQueue.addLast(integer);
}
}
}
}
}
I'm trying to solve the producer consumer problem with threads in java, but the code won't run in parallell/concurrently. The producer always fills up the buffer completely before the consumer starts to consume, and I don't get why. The point is trying to do it using only synchronized blocks, wait() and notify().
Main :
String [] data = {"Fisk", "Katt", "Hund", "Sau", "Fugl", "Elg", "Tiger",
"Kameleon", "Isbjørn", "Puma"};
ProducerConsumer pc = new ProducerConsumer(5);
Thread[] thrds = new Thread[2];
thrds[0] = new Thread(new MyThread1(pc, data)); // producer
thrds[1] = new Thread(new MyThread2(pc)); // consumer
thrds[0].start();
thrds[1].start();
for(int i = 0; i < 2; i++) { // wait for all threads to die
try {
thrds[i].join();
}
catch (InterruptedException ie) {}
}
System.exit(0);
ProducerConsumer.java:
import java.util.LinkedList;
import java.util.Queue;
public class ProducerConsumer implements Runnable {
private int bufferSize;
private Queue<String> buffer;
public ProducerConsumer(int size) {
bufferSize = size;
buffer = new LinkedList<String>();
}
public void produce(String item) throws InterruptedException {
synchronized(buffer) {
while (buffer.size() >= bufferSize) {
try {
System.out.println("Full buffer. Waiting for consumer...");
buffer.wait();
}catch (Exception e) {}
}
buffer.add(item);
System.out.println("Producer is putting " + item + " in the buffer");
buffer.notify();
}
}
public void consume() throws InterruptedException {
synchronized (buffer) {
while (buffer.size() == 0) {
try {
System.out.println("Empty buffer. Waiting for production...");
buffer.wait();
}catch (Exception e) {}
}
System.out.println("Consumer is consuming " + buffer.remove() + ".");
buffer.notify();
}
}
#Override
public void run() {
}
}
MyThread1 :
/*
* PRODUCER - Thread
*/
public class MyThread1 implements Runnable {
private String [] data;
private ProducerConsumer pc;
public MyThread1(ProducerConsumer pc, String [] data) {
this.pc = pc;
this.data = data;
}
#Override
public void run() {
for (int i = 0; i < data.length; i++) {
try {
pc.produce(data[i]);
} catch (InterruptedException ex) {}
}
}
}
MyThread2:
//THE CONSUMER - Thread
public class MyThread2 implements Runnable{
private ProducerConsumer pc;
public MyThread2(ProducerConsumer pc) {
this.pc = pc;
}
//Run consume
#Override
public void run() {
while (true) {
try {
pc.consume();
Thread.sleep(2);
}
catch(InterruptedException e) {}
}
}
}
On recent machines, with short queues like this, you will never see actual multithreading effects like, in this case, producer and consumer taking turns unless you slow both of them down a bit. You only slowed down the consumer. Instead of using a short array, put a million Integers in a queue and see what happens.
import java.util.List;
import java.util.concurrent.CopyOnWriteArrayList;
public class ProduserConsumerDemo {
public static void main(String[] args) {
List<Integer> list = new CopyOnWriteArrayList<>();
int size = 5;
Producer producer = new Producer(list, size);
Consumer consumer = new Consumer(list);
Thread t1 = new Thread(producer, "Producer");
Thread t2 = new Thread(consumer, "Consumer");
t1.start();
t2.start();
}
}
class Producer implements Runnable {
private final List<Integer> list;
private final int size;
public Producer(List<Integer> list, final int size) {
this.list = list;
this.size = size;
}
public void run() {
try {
produce();
} catch (InterruptedException e) {
e.printStackTrace();
}
}
private void produce() throws InterruptedException {
int i = 0;
while (i >= 0) {
synchronized (list) {
while (list.size() == size) {
System.out.println(
"List is full." + Thread.currentThread().getName() + " is waiting. Size:" + list.size());
list.wait();
}
System.out.println("Produce :" + i);
list.add(i++);
Thread.sleep(50);
list.notify();
}
}
}
}
class Consumer implements Runnable {
private final List<Integer> list;
public Consumer(List<Integer> list) {
this.list = list;
}
public void run() {
try {
consume();
} catch (InterruptedException e) {
e.printStackTrace();
}
}
private void consume() throws InterruptedException {
while (true) {
synchronized (list) {
while (list.isEmpty()) {
System.out.println(
"List is empty. " + Thread.currentThread().getName() + " is waiting. Size:" + list.size());
list.wait();
}
System.out.println("Consumed item:" + list.remove(0));
Thread.sleep(50);
list.notify();
}
}
}
}
My homework was to create project using parallelization that all should be proper.
However, I made my project but my profesor mentioned something is wrong in my code "please look at array list, something is not ok, maybe synchronization?".
I would like ask you community to help me and point what could be wrong. I think it might be problem with not covering by synchronize brackets my array list, am I right?
import java.util.ArrayList;
import java.util.Collections;
import java.util.List;
import java.util.logging.Level;
import java.util.logging.Logger;
/**
* My project finds all dividors for specific number
*It must use threads, so I made them. First I start them (first loop)
*then join them (second loop). My project must have that loops.
*Problem might be with not synchronizing methods array list...
*/
public class Main {
private final static int NUMBER = 100;
private final static List<Integer> dividors = new ArrayList<Integer>();
public static void main(String[] args) {
new Main().doStuff();
}
private int sqr;
private int sqrp1;
private void doStuff() {
sqr = (int) Math.sqrt(NUMBER);
sqrp1 = sqr + 1;
Thread[] t = new Thread[sqrp1];
//starting tasks
for (int i = 1; i < sqrp1; i++) {
final int it = i;
if (NUMBER % i == 0) {
final int e = i;
t[i] = new Thread(new Runnable() {
#Override
public void run() {
System.out.println("sta"+e);
if (!checkContains(e)) {
addElement(e);
}
final int dividednumber = NUMBER / e;
if (!checkContains(dividednumber)) {
addElement(dividednumber);
}
}
});
t[i].start();
}
}
//calling join for tasks
for (int i = 1; i < sqrp1; i++) {
final int it = i;
if (NUMBER % i == 0) {
try {
System.out.println("sto"+i);
t[i].join();
} catch (InterruptedException ex) {
Logger.getLogger(Main.class.getName()).log(Level.SEVERE, null, ex);
}
}
}
System.out.println("xxx");
Collections.sort(dividors);
Integer[] arrayDividors = dividors.toArray(new Integer[0]);
for (int i = 0; i < arrayDividors.length; i++) {
System.out.println(arrayDividors[i]);
}
}
private synchronized void addElement(int element) {
dividors.add(element);
}
private synchronized boolean checkContains(int element) {
return dividors.contains(element);
}
}
Am I right changing this part, is it ok now?
t[i] = new Thread(new Runnable() {
#Override
public void run() {
System.out.println("waiting " + e);
synchronized (this) {
System.out.println("entering " + e);
if (!checkContains(e)) {
addElement(e);
}
final int dividednumber = NUMBER / e;
if (!checkContains(dividednumber)) {
addElement(dividednumber);
}
System.out.println("leaving " + e);
}
}
});
You need to turn this into a single atomic operation.
if (!checkContains(dividednumber)) {
addElement(dividednumber);
}
Imagine you have two threads.
T1: if (!checkContains(dividednumber)) { // false
T2: if (!checkContains(dividednumber)) { // false
T1: addElement(dividednumber); // adds number
T2: addElement(dividednumber); // adds same number
If you have one addElementWithoutDuplicates, this won't happen.
Let's say I want to have n threads running and I want to output something when ALL threads complete. Here are the methods I've tried:
//This uses a ThreadGroup called tGroup
while(tGroup.activeCount() > 0) {
try {
Thread.sleep(10000);
} catch (InterruptedException e) {
e.printStackTrace();
}
}
This next is just with a while loop and an ArrayList storing Threads
boolean alive = true;
int count = 0;
while (alive) {
count = 0;
for (int i = 0; i < numThreads; i++) {
if (!threads.get(i).isAlive()) {
count++;
}
if (count == numThreads) {
alive = false;
break;
}
}
}
Loop through all your threads and join() each one. The join() will block on any unfinished thread until it finishes.
I think what you want to use is a CountDownLatch as this was built specifically for just this type of situation. Each worker thread will notify the latch when it is complete, and then any threads that have called await() on the latch will hold operation until the count down is complete. Please look at the sample code in the API link I've given above to see how easy and flexible this is to use.
Edit:
Oops, I guess I was too late in posting this. But regardless that you accepted the other answer, you still owe it to yourself to check this out as it is quite elegant and easy to use.
For example:
import java.util.Random;
import java.util.concurrent.CountDownLatch;
public class CountDownLatchEg {
public static void main(String[] args) {
int threadCount = 8;
CountDownLatch latch = new CountDownLatch(threadCount);
System.out.println("Start all threads");
for (int i = 0; i < threadCount; i++) {
new Thread(new MyRunnable(latch, i)).start();
}
System.out.println("All threads started");
try {
latch.await();
} catch (InterruptedException e) {}
System.out.println("All threads finished");
}
}
class MyRunnable implements Runnable {
private CountDownLatch latch;
private Random rand = new Random();
private long delay;
private int id;
public MyRunnable(CountDownLatch latch, int id) {
this.latch = latch;
delay = (rand.nextInt(4) + 1) * 1000;
this.id = id;
}
#Override
public void run() {
System.out.println("Start thread: " + id);
try {
Thread.sleep(delay);
} catch (InterruptedException e) {}
System.out.println("End thread: " + id);
latch.countDown();
}
}
Don't you mean this:
boolean alive = true;
int count = 0;
while (alive) {
count = 0;
for (int i = 0; i < numThreads; i++) {
if (!threads.get(i).isAlive()) {
count++;
}
}
if (count == numThreads) {
alive = false;
}
}
?