We Keep Coding

Simple .NET method of a parallel work queue with controllable concurrency

This sounds like an overly trivial question, and I think I am overcomplicating it because I haven't been able to find the answer for months. There are easy ways of doing this in Golang, Scala/Akka, etc but I can't seem to find anything in .NET.
What I need is an ability to have a list of Tasks that are all independent of each other, and the ability to execute them concurrently on a specified (and easily changeable) number of threads.
Basically something like:
int numberOfParallelThreads = 3; // changeable
Queue<Task> pendingTasks = GetPendingTasks(); // returns 80 items
await SomeBuiltInDotNetParallelExecutableManager.RunAllTasksWithSpecifiedConcurrency(pendingTasks, numberOfParallelThreads);
And that SomeBuiltInDotNetParallelExecutableManager would execute 80 tasks three at a time; i.e. when one finishes it draws the next one from the queue, until the queue is exhausted.
There is Task.WhenAll and Task.WaitAll, but you can't specify the max number of parallel threads in them.
Is there a built in, simple way to do this?

Parallel.ForEachAsync (or depending on actual workload it's sync counterpart - Parallel.ForEach, but it will not handle functions returning Task correctly):
IEnumerable<int> x = ...;
await Parallel.ForEachAsync(x, new ParallelOptions
{
MaxDegreeOfParallelism = 3
}, async (i, token) => await Task.Delay(i * 1000, token));
Also it is highly recommended that methods in C# return so called "hot", i.e. started tasks, so "idiomatically" Queue<Task> should be a collection of already started tasks, so you will have no control over number of them executing in parallel cause it will be controlled by ThreadPool/TaskScheduler.
And there is port of Akka to .NET - Akka.NET if you want to go down that route.

Microsoft's Reactive Framework makes this easy too:
IEnumerable<int> values = ...;
IDisposable subscription =
values
.ToObservable()
.Select(v => Observable.Defer(() => Observable.Start(() => { /* do work on each value */ })))
.Merge(3)
.Subscribe();

How is concurrency handled when not using Task.Run() to schedule the work?

If we fill a list of Tasks that need to do both CPU-bound and I/O bound work, by simply passing their method declaration to that list (Not by creating a new task and manually scheduling it by using Task.Start), how exactly are these tasks handled?
I know that they are not done in parallel, but concurrently.
Does that mean that a single thread will move along them, and that single thread might not be the same thread in the thread pool, or the same thread that initially started waiting for them all to complete/added them to the list?
EDIT: My question is about how exactly these items are handled in the list concurrently - is the calling thread moving through them, or something else is going on?
Code for those that need code:
public async Task SomeFancyMethod(int i)
{
doCPUBoundWork(i);
await doIOBoundWork(i);
}
//Main thread
List<Task> someFancyTaskList = new List<Task>();
for (int i = 0; i< 10; i++)
someFancyTaskList.Add(SomeFancyMethod(i));
// Do various other things here --
// how are the items handled in the meantime?
await Task.WhenAll(someFancyTaskList);
Thank you.

Asynchronous methods always start running synchronously. The magic happens at the first await. When the await keyword sees an incomplete Task, it returns its own incomplete Task. If it sees a complete Task, execution continues synchronously.
So at this line:
someFancyTaskList.Add(SomeFancyMethod(i));
You're calling SomeFancyMethod(i), which will:
Run doCPUBoundWork(i) synchronously.
Run doIOBoundWork(i).
If doIOBoundWork(i) returns an incomplete Task, then the await in SomeFancyMethod will return its own incomplete Task.
Only then will the returned Task be added to your list and your loop will continue. So the CPU-bound work is happening sequentially (one after the other).
There is some more reading about this here: Control flow in async programs (C#)
As each I/O operation completes, the continuations of those tasks are scheduled. How those are done depends on the type of application - particularly, if there is a context that it needs to return to (desktop and ASP.NET do unless you specify ConfigureAwait(false), ASP.NET Core doesn't). So they might run sequentially on the same thread, or in parallel on ThreadPool threads.
If you want to immediately move the CPU-bound work to another thread to run that in parallel, you can use Task.Run:
someFancyTaskList.Add(Task.Run(() => SomeFancyMethod(i)));
If this is in a desktop application, then this would be wise, since you want to keep CPU-heavy work off of the UI thread. However, then you've lost your context in SomeFancyMethod, which may or may not matter to you. In a desktop app, you can always marshall calls back to the UI thread fairly easily.

I assume you don't mean passing their method declaration, but just invoking the method, like so:
var tasks = new Task[] { MethodAsync("foo"),
MethodAsync("bar") };
And we'll compare that to using Task.Run:
var tasks = new Task[] { Task.Run(() => MethodAsync("foo")),
Task.Run(() => MethodAsync("bar")) };
First, let's get the quick answer out of the way. The first variant will have lower or equal parallelism to the second variant. Parts of MethodAsync will run the caller thread in the first case, but not in the second case. How much this actually affects the parallelism depends entirely on the implementation of MethodAsync.
To get a bit deeper, we need to understand how async methods work. We have a method like:
async Task MethodAsync(string argument)
{
DoSomePreparationWork();
await WaitForIO();
await DoSomeOtherWork();
}
What happens when you call such a method? There is no magic. The method is a method like any other, just rewritten as a state machine (similar to how yield return works). It will run as any other method until it encounters the first await. At that point, it may or may not return a Task object. You may or may not await that Task object in the caller code. Ideally, your code should not depend on the difference. Just like yield return, await on a (non-completed!) task returns control to the caller of the method. Essentially, the contract is:
If you have CPU work to do, use my thread.
If whatever you do would mean the thread isn't going to use the CPU, return a promise of the result (a Task object) to the caller.
It allows you to maximize the ratio of what CPU work each thread is doing. If the asynchronous operation doesn't need the CPU, it will let the caller do something else. It doesn't inherently allow for parallelism, but it gives you the tools to do any kind of asynchronous operation, including parallel operations. One of the operations you can do is Task.Run, which is just another asynchronous method that returns a task, but which returns to the caller immediately.
So, the difference between:
MethodAsync("foo");
MethodAsync("bar");
and
Task.Run(() => MethodAsync("foo"));
Task.Run(() => MethodAsync("bar"));
is that the former will return (and continue to execute the next MethodAsync) after it reaches the first await on a non-completed task, while the latter will always return immediately.
You should usually decide based on your actual requirements:
Do you need to use the CPU efficiently and minimize context switching etc., or do you expect the async method to have negligible CPU work to do? Invoke the method directly.
Do you want to encourage parallelism or do you expect the async method to do interesting amounts of CPU work? Use Task.Run.

Here is your code rewritten without async/await, with old-school continuations instead. Hopefully it will make it easier to understand what's going on.
public Task CompoundMethodAsync(int i)
{
doCPUBoundWork(i);
return doIOBoundWorkAsync(i).ContinueWith(_ =>
{
doMoreCPUBoundWork(i);
});
}
// Main thread
var tasks = new List<Task>();
for (int i = 0; i < 10; i++)
{
Task task = CompoundMethodAsync(i);
tasks.Add(task);
}
// The doCPUBoundWork has already ran synchronously 10 times at this point
// Do various things while the compound tasks are progressing concurrently
Task.WhenAll(tasks).ContinueWith(_ =>
{
// The doIOBoundWorkAsync/doMoreCPUBoundWork have completed 10 times at this point
// Do various things after all compound tasks have been completed
});
// No code should exist here. Move everything inside the continuation above.

What are the use cases for AttachedToParent?

One of the differences between Task.Run and TaskFactory.StartNew is the addition of the new DenyChildAttach option by default. But AttachedToParent was created for a reason. In what situation would you want to use attached child tasks?

The use case for AttachedToParent is when you have a nested dynamic task parallelism scenario. That is:
You have an algorithm to execute in parallel (that is, CPU code, not I/O operations), and
The tasks your algorithm must perform vary while the algorithm is running (that is, at the time your algorithm starts, it doesn't know how many tasks it needs), and
There is a hierarchical or parent/child relationship in the completion of these tasks (that is, the "parent" should not be considered complete until all its children are complete, and if any children are failed/canceled, then the parent should be failed/canceled as well, even if its code did not error).
Since the vast majority of concurrency problems are I/O-based (not CPU-based), and since the vast majority of parallelism scenarios are data-based parallelism (not dynamic task parallelism), and since dynamic task parallel problems may or may not have a hierarchical nature, this scenario almost never comes up.
Unfortunately, there is a logical parent/child relationship between tasks (including asynchronous tasks), which have caused a lot of developers to incorrectly attempt to use the AttachedToParent flag with async tasks. Thus, the introduction of the DenyChildAttach flag (which prevents AttachedToParent from taking effect).

The only situation I can think of is made obsolete by async-await. E.g. if you want a task to wait on its child tasks you will await them.
var parentTask = Task.Run(async () =>
{
await Task.Run(() => Thread.Sleep(1000));
Console.WriteLine("parent task completed");
});
But in .Net 4.0 you would have to Wait() on them. E.g.
var parentTask = Task.Factory.StartNew(() =>
{
Task.Factory.StartNew(() => Thread.Sleep(1000)).Wait();
Console.WriteLine("parent task completed");
});
Unlike the first example this will block the thread until the child task is complete. With attached tasks we can get the same behaviour like this.
var parentTask = Task.Factory.StartNew(() =>
{
Task.Factory.StartNew(() => Thread.Sleep(1000), TaskCreationOptions.AttachedToParent)
.ContinueWith(antecedent => Console.WriteLine("parent task completed", TaskContinuationOptions.AttachedToParent);
});

How to execute nested async/await code in parallel while maintaining the same thread on await continuations?

This might be the worst StackOverflow title I've ever written. What I'm actually trying to do is execute an asynchronous method that uses the async/await convention (and itself contains additional await calls) from within a synchronous method multiple times in parallel while maintaining the same thread throughout the execution of each branch of the parallel execution, including for all await continuations. To put it another way, I want to execute some async code synchronously, but I want to do it multiple times in parallel. Now you can see why the title was so bad. Perhaps this is best illustrated with some code...
Assume I have the following:
public class MyAsyncCode
{
async Task MethodA()
{
// Do some stuff...
await MethodB();
// Some other stuff
}
async Task MethodB()
{
// Do some stuff...
await MethodC();
// Some other stuff
}
async Task MethodC()
{
// Do some stuff...
}
}
The caller is synchronous (from a console application). Let me try illustrating what I'm trying to do with an attempt to use Task.WaitAll(...) and wrapper tasks:
public void MyCallingMethod()
{
List<Task> tasks = new List<Task>();
for(int c = 0 ; c < 4 ; c++)
{
MyAsyncCode asyncCode = new MyAsyncCode();
tasks.Add(Task.Run(() => asyncCode.MethodA()));
}
Task.WaitAll(tasks.ToArray());
}
The desired behavior is for MethodA, MethodB, and MethodC to all be run on the same thread, both before and after the continuation, and for this to happen 4 times in parallel on 4 different threads. To put it yet another way, I want to remove the asynchronous behavior of my await calls since I'm making the calls parallel from the caller.
Now, before I go any further, I do understand that there's a difference between asynchronous code and parallel/multi-threaded code and that the former doesn't imply or suggest the latter. I'm also aware the easiest way to achieve this behavior is to remove the async/await declarations. Unfortunately, I don't have the option to do this (it's in a library) and there are reasons why I need the continuations to all be on the same thread (having to do with poor design of said library). But even more than that, this has piqued my interest and now I want to know from an academic perspective.
I've attempted to run this using PLINQ and immediate task execution with .AsParallel().Select(x => x.MethodA().Result). I've also attempted to use the AsyncHelper class found here and there, which really just uses .Unwrap().GetAwaiter().GetResult(). I've also tried some other stuff and I can't seem to get the desired behavior. I either end up with all the calls on the same thread (which obviously isn't parallel) or end up with the continuations executing on different threads.
Is what I'm trying to do even possible, or are async/await and the TPL just too different (despite both being based on Tasks)?

The methods that you are calling do not use ConfigureAwait(false). This means that we can force the continuations to resume in a context we like. Options:
Install a single-threaded synchronization context. I believe Nito.Async has that.
Use a custom TaskScheduler. await looks at TaskScheduler.Current and resumes at that scheduler if it is non-default.
I'm not sure if there are any pros and cons for either option. Option 2 has easier scoping I think. Option 2 would look like:
Task.Factory.StartNew(
() => MethodA()
, new ConcurrentExclusiveSchedulerPair().ExclusiveScheduler).Unwrap();
Call this once for each parallel invocation and use Task.WaitAll to join all those tasks. Probably you should dispose of that scheduler as well.
I'm (ab)using ConcurrentExclusiveSchedulerPair here to get a single-threaded scheduler.
If those methods are not particularly CPU-intensive you can just use the same scheduler/thread for all of them.

You can create 4 independent threads, each one executes MethodA with a limited-concurrency (actually, no concurrency at all) TaskScheduler. That will ensure that every Task, and continuation Tasks, that the thread creates, will be executed by that thread.
public void MyCallingMethod()
{
CancellationToken csl = new CancellationToken();
var threads = Enumerable.Range(0, 4).Select(p =>
{
var t = new Thread(_ =>
{
Task.Factory.StartNew(() => MethodA(), csl, TaskCreationOptions.None,
new LimitedConcurrencyLevelTaskScheduler(1)).Wait();
});
t.Start();
return t;
}).ToArray();
//You can block the main thread and wait for the other threads here...
}
That won't ensure you a 4th degree parallelism, of course.
You can see an implementation of such TaskScheduler in MSDN - https://msdn.microsoft.com/en-us/library/ee789351(v=vs.110).aspx

Is this a good candidate for using async await in this C# .NET code?

I have a collection of objects. Before I save these objects, I need to grab some reference data from the Database. I need to do a few calls to the DB for each object. These methods can happen in any order / are not dependent upon each other.
Can I use async / await to make these db operations happen at the same time ?
Parallel.ForEach(animals,
animal => {
animal.Owner = GetAnimalOwner(ownerId);
animal.FavouriteFood = GetFavouriteFood(foodId);
Database.Store(animal);
});
That's some pseudo code that explains what I'm trying to do.
Is this a good candidate for async / await ?

If all you want to do is to execute the two operations in parallel, then you don't need async-await. You can use old TPL methods, like Parallel.Invoke():
Parallel.ForEach(animals,
animal => {
Parallel.Invoke(
() => animal.Owner = GetAnimalOwner(ownerId),
() => animal.FavouriteFood = GetFavouriteFood(foodId));
Database.Store(animal);
});
Keep in mind that there is no guarantee that this will actually improve your performance, depending on your DB and your connection to it.
In general, async-await makes sense in two cases:
You want to offload some long-running code from the UI thread.
You want to use less threads to improve performance or scalability.
It seems to me like #1 is not your case. If you're in case #2, then you should rewrite your database methods into truly asynchronous ones (depending on your DB library, this may not be possible).
If you do that, then you could do something similar to what Scott Chamberlain suggested (but not using Parallel.ForEach()):
var tasks = animals.Select(
async animal => {
var getOwnerTask = GetAnimalOwnerAsync(ownerId);
var getFavouriteFoodTask = GetFavouriteFoodAsync(foodId);
animal.Owner = await getOwnerTask;
animal.FavouriteFood = await getFavouriteFoodTask;
await Database.StoreAsync(animal);
});
await Task.WhenAll(tasks);
This will execute with even bigger degree of parallelism then Parallel.ForEach() (because that's limited by the thread pool), but again, that may not actually improve performance. If you wanted to limit the degree of parallelism, you could add SemaphoreSlim or you could use TPL Dataflow instead.