Develop Reference

Thread executes "slowly"

I am starting with threads and wrote for the sake of learning the following simple program, which later would be used to calculate about 100,000 times a formula (it is a relatively simple one but which takes an iterated range of values).
The problem with it is that I expected every thread to execute in almost no time and thus the complete program to finish nearly immediately, but the fact is that everything runs too slow (about 10s)...
static readonly double TotalIterations = 1000;
public static Iterations ActualIterations = new Iterations();
public static void Main()
{
var par1 = "foo";
var par2 = "boo";
var par3 = 3;
for (int i = 0; i < TotalIterations; i++)
{
new Thread(() => new Calculations().Calculate(par1, par2, par3)).Start();
}
AwaitingThreads();
}
static void AwaitThreads()
{
Console.WriteLine("Awaiting threads to finished...");
while (true)
{
lock (ActualIterations)
{
if (ActualIterations.Progress() == TotalIterations) break;
}
Thread.Sleep(1 * 1000);
}
Console.WriteLine("All threads finished!");
}
public class Calculations {
public bool Calculate(string par1, string par2, int par3)
{
// ...
bool result = false;
lock (ActualIterations)
{
ActualIterations.Incr();
}
return result;
}
}
public class Iterations
{
int progress = 0;
public void Incr()
{
progress++;
}
public int Progress()
{
return progress;
}
}
I also tried using a ThreadPool like this, but there was no improvement...
static readonly double TotalIterations = 1000;
static string par1 = "foo";
static string par2 = "boo";
static int par3 = 3;
public static Iterations ActualIterations = new Iterations();
public static void Main()
{
ThreadPool.QueueUserWorkItem(MyThreadPool);
AwaitThreads();
}
static void AwaitThreads()
{
Console.WriteLine("Awaiting threads to finished...");
while (true)
{
lock (ActualIterations)
{
if (ActualIterations.Progress() == TotalIterations) break;
}
Thread.Sleep(1 * 1000);
}
Console.WriteLine("All threads finished!");
}
static void MyThreadPool(Object stateInfo)
{
for (int i = 0; i < TotalIterations; i++)
{
new Thread(() => new Calculations().Calculate(par1, par2, par3)).Start();
}
}
public class Calculations {
public bool Calculate(string par1, string par2, int par3)
{
// ...
bool result = false;
lock (ActualIterations)
{
ActualIterations.Incr();
}
return result;
}
}
public class Iterations
{
int progress = 0;
public void Incr()
{
progress++;
}
public int Progress()
{
return progress;
}
}
When I quit using threads in this example and use a static method, executing it sequentially in my for loop, the program finishes in 1s...
Can anybody enlighten me what I am doing wrong here with those threads?

The problem with it is that I expected every thread to execute in almost no time
Right. You're ignoring the fact that creating a new thread is a relatively expensive operation. Far, far more expensive than "acquiring a lock and incrementing an integer" which is the work you're doing in the thread.
To give a real world comparison, it's a little like ordering a new car, waiting it to be delivered, and then driving it 1km. That's going to be slower than just walking 1km.
Using the thread pool would be faster, but you're not using it correctly - you're launching one thread pool task which then creates all the other threads again.
I would encourage you to look at using Task<T> instead, which normally uses the thread pool under the hood, and is a generally more modern abstraction for this sort of work.

This is the way to proceed doing what you wanted to do:
class Program
{
static void Main(string[] args)
{
List<Task> tasks = new List<Task>();
for (int i = 0; i < 1000; i++)
{
tasks.Add(Task.Run(() =>
{
Console.WriteLine("Calculations " + DateTime.Now);
}));
}
Task.WaitAll(tasks.ToArray());
}
}
Tasks are actually optimized and programmer-friendly to use if you need to work with threads.
Another advice i want to give you is to create an Object just for locking purposes, example:
class Program
{
private static Object _locker = new Object();
static void Main(string[] args)
{
List<Task> tasks = new List<Task>();
for (int i = 0; i < 1000; i++)
{
tasks.Add(Task.Run(() =>
{
lock (_locker)
{
Console.WriteLine("Calculations " + DateTime.Now);
}
}));
}
Task.WaitAll(tasks.ToArray());
}
}

I see the problem in the AwaitThreads method.
It uses the same lock (ActualIterations) as working thread and it makes working threads to wait for shared resource additionally.
Also (as it was mentioned by #Euphoric) the thread working code you have shown is just about single increment and it uses the shared resource between all threads.
You have to change it in some another way and try to avoid shared resource usage in multi threaded environment.
For example, if you need to make some calculation on huge data array you have to feed each thread own data part to be processed and then wait for all tasks to be finished. There is Task concept and Task.WaitAll

How do I use Interlocked.Add() correctly?

From the documentation here: "The methods of this class help protect against errors that can occur when the scheduler switches contexts while a thread is updating a variable that can be accessed by other threads..."
Also, an answer to this question states "INTERLOCKED METHODS ARE CONCURRENTLY SAFE ON ANY NUMBER OF COREs OR CPUs" which seems pretty clear.
Based on the above I thought Interlocked.Add() would be sufficient for multiple threads to do addition on a variable. Apparently I'm wrong or I'm using the method incorrectly. In the runnable code below I expect Downloader.ActiveRequestCount to be zero when Run() completes. If I do not lock around the call to Interlocked.Add I get a random non-zero result. What is the correct usage of Interlocked.Add()?
class Program
{
private Downloader downloader { get; set; }
static void Main(string[] args)
{
new Program().Run().Wait();
}
public async Task Run()
{
downloader = new Downloader();
List<Task> tasks = new List<Task>(100);
for (int i = 0; i < 100; i++)
tasks.Add(Task.Run(Download));
await Task.WhenAll(tasks);
Console.Clear();
//expected:0, actual when lock is not used:random number i.e. 51,115
Console.WriteLine($"ActiveRequestCount is : {downloader.ActiveRequestCount}");
Console.ReadLine();
}
private async Task Download()
{
for (int i = 0; i < 100; i++)
await downloader.Download();
}
}
public class Downloader :INotifyPropertyChanged
{
private object locker = new object();
private int _ActiveRequestCount;
public int ActiveRequestCount { get => _ActiveRequestCount; private set => _ActiveRequestCount = value; }
public async Task<string> Download()
{
string result = string.Empty;
try
{
IncrementActiveRequestCount(1);
result = await Task.FromResult("boo");
}
catch (Exception ex)
{
Console.WriteLine("oops");
}
finally
{
IncrementActiveRequestCount(-1);
}
return result;
}
public void IncrementActiveRequestCount(int value)
{
//lock (locker) // is this redundant
//{
_ActiveRequestCount = Interlocked.Add(ref _ActiveRequestCount, value);
//}
RaisePropertyChanged(nameof(ActiveRequestCount));
}
#region INotifyPropertyChanged implementation
public event PropertyChangedEventHandler PropertyChanged;
public void RaisePropertyChanged([CallerMemberNameAttribute] string propertyName = "") => PropertyChanged?.Invoke(this, new PropertyChangedEventArgs(propertyName));
#endregion
}

Replace
_ActiveRequestCount = Interlocked.Add(ref _ActiveRequestCount, value);
with
Interlocked.Add(ref _ActiveRequestCount, value);
Interlocked.Add is thread-safe and takes a ref parameter, so that it can do the assignment safely. You additionally perform an (unnecessary) unsafe assignment (=). Just remove it.

Read-write synchronization class implementation

I am writing a read-write synchronization class, and would like some advice on what I to do next. For some reason, it sometimes allows a Read to happen in the middle of a Write, and I cannot find the reason.
This is what I want from this class:
Reads not allowed at the same time as writes.
Multiples reads can happen at the same time.
Only one write can happen at a time.
When a write is needed, all already executing reads continue,
no new reads are allowed, when all reads finish the write executes.
I know that .Net framework has a class to do this... but what I want is to understand and to reproduce something like that. I'm not reinventing the wheel, I am trying to understand it by making my own wheel... happens that my wheel is kinda squared a bit.
What I have currently is this:
public class ReadWriteSync
{
private ManualResetEvent read = new ManualResetEvent(true);
private volatile int readingBlocks = 0;
private AutoResetEvent write = new AutoResetEvent(true);
private object locker = new object();
public IDisposable ReadLock()
{
lock (this.locker)
{
this.write.Reset();
Interlocked.Increment(ref this.readingBlocks);
this.read.WaitOne();
}
return new Disposer(() =>
{
if (Interlocked.Decrement(ref this.readingBlocks) == 0)
this.write.Set();
});
}
public IDisposable WriteLock()
{
lock (this.locker)
{
this.read.Reset();
this.write.WaitOne();
}
return new Disposer(() =>
{
this.read.Set();
if (this.readingBlocks == 0)
this.write.Set();
});
}
class Disposer : IDisposable
{
Action disposer;
public Disposer(Action disposer) { this.disposer = disposer; }
public void Dispose() { this.disposer(); }
}
}
This is my test program... when something goes wrong it prints the lines in red.
class Program
{
static ReadWriteSync sync = new ReadWriteSync();
static void Main(string[] args)
{
Console.BackgroundColor = ConsoleColor.DarkGray;
Console.ForegroundColor = ConsoleColor.Gray;
Console.Clear();
Task readTask1 = new Task(() => DoReads("A", 20));
Task readTask2 = new Task(() => DoReads("B", 30));
Task readTask3 = new Task(() => DoReads("C", 40));
Task readTask4 = new Task(() => DoReads("D", 50));
Task writeTask1 = new Task(() => DoWrites("E", 500));
Task writeTask2 = new Task(() => DoWrites("F", 200));
readTask1.Start();
readTask2.Start();
readTask3.Start();
readTask4.Start();
writeTask1.Start();
writeTask2.Start();
Task.WaitAll(
readTask1, readTask2, readTask3, readTask4,
writeTask1, writeTask2);
}
static volatile bool reading;
static volatile bool writing;
static void DoWrites(string name, int interval)
{
for (int i = 1; i < int.MaxValue; i += 2)
{
using (sync.WriteLock())
{
Console.ForegroundColor = (writing || reading) ? ConsoleColor.Red : ConsoleColor.Gray;
writing = true;
Console.WriteLine("WRITE {1}-{0} BEGIN", i, name);
Thread.Sleep(interval);
Console.WriteLine("WRITE {1}-{0} END", i, name);
writing = false;
}
Thread.Sleep(interval);
}
}
static void DoReads(string name, int interval)
{
for (int i = 0; i < int.MaxValue; i += 2)
{
using (sync.ReadLock())
{
Console.ForegroundColor = (writing) ? ConsoleColor.Red : ConsoleColor.Gray;
reading = true;
Console.WriteLine("READ {1}-{0} BEGIN", i, name);
Thread.Sleep(interval * 3);
Console.WriteLine("READ {1}-{0} END", i, name);
reading = false;
}
Thread.Sleep(interval);
}
}
}
What is wrong with all this... any advice on how to do it correctly?

The primary issue that I see is that you are trying to make reset events encompass both the meanings of a read/write and the handling of their current state, without synchronizing in a consistent manner.
Here's an example of how the inconsistent synchronization may bite you in your specific code.
A write is disposing and a read is coming in.
The read acquires the lock
The write sets the read ManualResetEvent (MRE)
The write checks the current read count, finding 0
The read resets the write AutoResetEvent (ARE)
The read increments the read count
The read finds its MRE has been set and begins to read
All is fine so far, but the write hasn't finished yet...
A second write comes in and acquires the lock
The second write resets the read MRE
The first write finishes by setting the write ARE
The second write finds its ARE has been set and begins to write
When thinking about multiple threads, unless you are within a lock of some sort, you must take the view that all other data is wildly fluctuating and cannot be trusted.
A naive implementation of this may split out the queueing logic from the state logic and synchronize appropriately.
public class ReadWrite
{
private static int readerCount = 0;
private static int writerCount = 0;
private int pendingReaderCount = 0;
private int pendingWriterCount = 0;
private readonly object decision = new object();
private class WakeLock:IDisposable
{
private readonly object wakeLock;
public WakeLock(object wakeLock) { this.wakeLock = wakeLock; }
public virtual void Dispose() { lock(this.wakeLock) Monitor.PulseAll(this.wakeLock); }
}
private class ReadLock:WakeLock
{
public ReadLock(object wakeLock) : base(wakeLock) { Interlocked.Increment(ref readerCount); }
public override void Dispose()
{
Interlocked.Decrement(ref readerCount);
base.Dispose();
}
}
private class WriteLock:WakeLock
{
public WriteLock(object wakeLock) : base(wakeLock) { Interlocked.Increment(ref writerCount); }
public override void Dispose()
{
Interlocked.Decrement(ref writerCount);
base.Dispose();
}
}
public IDisposable TakeReadLock()
{
lock(decision)
{
pendingReaderCount++;
while (pendingWriterCount > 0 || Thread.VolatileRead(ref writerCount) > 0)
Monitor.Wait(decision);
pendingReaderCount--;
return new ReadLock(this.decision);
}
}
public IDisposable TakeWriteLock()
{
lock(decision)
{
pendingWriterCount++;
while (Thread.VolatileRead(ref readerCount) > 0 || Thread.VolatileRead(ref writerCount) > 0)
Monitor.Wait(decision);
pendingWriterCount--;
return new WriteLock(this.decision);
}
}
}

Mutithreading with sequence

I have a main task that is spawning threads to do some work. When the work is completed it will write to the console.
My problem is that some of the threads that are created later will finish faster than those created earlier. However I need the writing to the console to be done in the same exact sequence as the thread was created.
So if a thread had completed its task, while some earlier threads had not, it has to wait till those earlier threads complete too.
public class DoRead
{
public DoRead()
{
}
private void StartReading()
{
int i = 1;
while (i < 10000)
{
Runner r = new Runner(i, "Work" + i.ToString());
r.StartThread();
i += 1;
}
}
}
internal class Runner : System.IDisposable
{
int _count;
string _work = "";
public Runner(int Count, string Work)
{
_count = Count;
_work = Work;
}
public void StartThread()
{
ThreadPool.QueueUserWorkItem(new WaitCallback(runThreadInPool), this);
}
public static void runThreadInPool(object obj)
{
((Runner)obj).run();
}
public void run()
{
try
{
Random r = new Random();
int num = r.Next(1000, 2000);
DateTime end = DateTime.Now.AddMilliseconds(num);
while (end > DateTime.Now)
{
}
Console.WriteLine(_count.ToString() + " : Done!");
}
catch
{
}
finally
{
_work = null;
}
}
public void Dispose()
{
this._work = null;
}
}

There may be a simpler way to do this than I used, (I'm used to .Net 4.0).
using System;
using System.Collections.Generic;
using System.Text;
using System.Threading;
namespace ConsoleApplication5
{
class Program
{
public static readonly int numOfTasks = 100;
public static int numTasksLeft = numOfTasks;
public static readonly object TaskDecrementLock = new object();
static void Main(string[] args)
{
DoRead dr = new DoRead();
dr.StartReading();
int tmpNumTasks = numTasksLeft;
while ( tmpNumTasks > 0 )
{
Thread.Sleep(1000);
tmpNumTasks = numTasksLeft;
}
List<string> strings = new List<string>();
lock( DoRead.locker )
{
for (int i = 1; i <= Program.numOfTasks; i++)
{
strings.Add( DoRead.dicto[i] );
}
}
foreach (string s in strings)
{
Console.WriteLine(s);
}
Console.ReadLine();
}
public class DoRead
{
public static readonly object locker = new object();
public static Dictionary<int, string> dicto = new Dictionary<int, string>();
public DoRead()
{
}
public void StartReading()
{
int i = 1;
while (i <= Program.numOfTasks )
{
Runner r = new Runner(i, "Work" + i.ToString());
r.StartThread();
i += 1;
}
}
}
internal class Runner : System.IDisposable
{
int _count;
string _work = "";
public Runner(int Count, string Work)
{
_count = Count;
_work = Work;
}
public void StartThread()
{
ThreadPool.QueueUserWorkItem(new WaitCallback(runThreadInPool), this);
}
public static void runThreadInPool(object obj)
{
Runner theRunner = ((Runner)obj);
string theString = theRunner.run();
lock (DoRead.locker)
{
DoRead.dicto.Add( theRunner._count, theString);
}
lock (Program.TaskDecrementLock)
{
Program.numTasksLeft--;
}
}
public string run()
{
try
{
Random r = new Random();
int num = r.Next(1000, 2000);
Thread.Sleep(num);
string theString = _count.ToString() + " : Done!";
return theString;
}
catch
{
}
finally
{
_work = null;
}
return "";
}
public void Dispose()
{
this._work = null;
}
}
}
}
Basically, I store the string you want printed from each task into a dictionary where the index is the task#. (I use a lock to make accessing the dictionary safe).
Next, so that the main program waits until all the background threads are done, I used another locked access to a NumTasksLeft variable.
I added stuff into the callback for the Runner.
It is bad practice to use busy loops, so I changed it to a Thread.Sleep( num ) statement.
Just change numOfTasks to 10000 to match your example.
I pull the return strings out of the dictionary in order, and then print it to the screen.
I'm sure you could refactor this to move or otherwise deal with the global variables, but this works.
Also, you might have noticed I didn't use the lock in the command
tmpNumTasks = numTasksLeft;
That's threadsafe, since numTasksLeft is an int which is read atomically on 32-bit computers and higher.

I don't know much on C#, but the whole idea of multi-threading is that you have multiple thread executing independently and you can never know which one will finish earlier (and you shouldn't expect earlier thread to end earlier).
One workaround is, instead writing out the finish message in the processing thread, have the processing thread setup a flag somewhere (probably a list with no of elements = no of thread spawned), and have a separate thread print out the finish message base on the flags in that list, and report up to the position that previous flag is consecutively "finished".
Honestly I don't feel that reasonable for you to print finish message like this anyway. I think changing the design is way better to have such meaningless "feature".

Typically, such requirements are met with an incrementing sequence number, much as you have already done.
Usually, the output from the processing threads is fed through a filter object that contains a list, (or dictionary), of all out-of-order result objects, 'holding them back' until all results with a lower seqeuence-number have come in. Again, similar to what you have already done.
What is not necessary is any kind of sleep() loop. The work threads themselves can operate the filter object, (which would beed a lock), or the work threads can producer-consumer-queue the results to an 'output thread' that operates the out-of-order filter.
This scheme works fine with pooled work threads, ie. those without continual create/terminate/destroy overhead.

A robust method of tracking failed workers with ThreadPool

I'm looking for a good method of tracking (counting) which workers have failed when queued with a Threadpool and using WaitHandle.WaitAll() for all threads to finish.
Is Interlocking a counter a good technique or is there a more robust strategy?

Okay, here's an approach that you could take. I've encapsulated the data that we want to track into a class TrackedWorkers. There is a constructor on this class that enables you to set how many workers will be working. Then, the workers are launched using LaunchWorkers which requires a delegate that eats an object and returns a bool. The object represents the input to the worker and the bool represents success or failure depending on true or false being the return value, respectively.
So basically what we do we have an array to track worker state. We launch the workers and set the status corresponding to that worker depending on the return value from the worker. When the worker returns, we set an AutoResetEvent and WaitHandle.WaitAll for all the AutoResetEvents to be set.
Note that there is an nested class to track the work (the delegate) the worker is supposed to do, the input to that work, and an ID used to set the status AutoResetEvent corresponding to that thread.
Note very carefully that once the work is done we are not holding a reference to the work delegate func nor to the input. This is important so that we don't accidentally prevent stuff from being garbage collected.
There are methods for getting the status of a particular worker, as well as all the indexes of the workers that succeeded and all the indexes of the workers that failed.
One last note: I do not consider this code production ready. It is merely a sketch of the approach that I would take. You need to take care to add testing, exception handling and other such details.
class TrackedWorkers {
class WorkerState {
public object Input { get; private set; }
public int ID { get; private set; }
public Func<object, bool> Func { get; private set; }
public WorkerState(Func<object, bool> func, object input, int id) {
Func = func;
Input = input;
ID = id;
}
}
AutoResetEvent[] events;
bool[] statuses;
bool _workComplete;
int _number;
public TrackedWorkers(int number) {
if (number <= 0 || number > 64) {
throw new ArgumentOutOfRangeException(
"number",
"number must be positive and at most 64"
);
}
this._number = number;
events = new AutoResetEvent[number];
statuses = new bool[number];
_workComplete = false;
}
void Initialize() {
_workComplete = false;
for (int i = 0; i < _number; i++) {
events[i] = new AutoResetEvent(false);
statuses[i] = true;
}
}
void DoWork(object state) {
WorkerState ws = (WorkerState)state;
statuses[ws.ID] = ws.Func(ws.Input);
events[ws.ID].Set();
}
public void LaunchWorkers(Func<object, bool> func, object[] inputs) {
Initialize();
for (int i = 0; i < _number; i++) {
WorkerState ws = new WorkerState(func, inputs[i], i);
ThreadPool.QueueUserWorkItem(this.DoWork, ws);
}
WaitHandle.WaitAll(events);
_workComplete = true;
}
void ThrowIfWorkIsNotDone() {
if (!_workComplete) {
throw new InvalidOperationException("work not complete");
}
}
public bool GetWorkerStatus(int i) {
ThrowIfWorkIsNotDone();
return statuses[i];
}
public IEnumerable<int> SuccessfulWorkers {
get {
return WorkersWhere(b => b);
}
}
public IEnumerable<int> FailedWorkers {
get {
return WorkersWhere(b => !b);
}
}
IEnumerable<int> WorkersWhere(Predicate<bool> predicate) {
ThrowIfWorkIsNotDone();
for (int i = 0; i < _number; i++) {
if (predicate(statuses[i])) {
yield return i;
}
}
}
}
Sample usage:
class Program {
static Random rg = new Random();
static object lockObject = new object();
static void Main(string[] args) {
int count = 64;
Pair[] pairs = new Pair[count];
for(int i = 0; i < count; i++) {
pairs[i] = new Pair(i, 2 * i);
}
TrackedWorkers workers = new TrackedWorkers(count);
workers.LaunchWorkers(SleepAndAdd, pairs.Cast<object>().ToArray());
Console.WriteLine(
"Number successful: {0}",
workers.SuccessfulWorkers.Count()
);
Console.WriteLine(
"Number failed: {0}",
workers.FailedWorkers.Count()
);
}
static bool SleepAndAdd(object o) {
Pair pair = (Pair)o;
int timeout;
double d;
lock (lockObject) {
timeout = rg.Next(1000);
d = rg.NextDouble();
}
Thread.Sleep(timeout);
bool success = d < 0.5;
if (success) {
Console.WriteLine(pair.First + pair.Second);
}
return (success);
}
}
The above program is going to launch sixty-four threads. The ith thread has the task of adding the numbers i and 2 * i and printing the result to the console. However, I have added a random amount of sleep (less than one second) to simulate busyness and I flip a coin to determine success or failure of the thread. Those that succeed print the sum they were tasked with and return true. Those that fail print nothing and return false.
Here I have used
struct Pair {
public int First { get; private set; }
public int Second { get; private set; }
public Pair(int first, int second) : this() {
this.First = first;
this.Second = second;
}
}