Could someone clear up to me how these things correlate:
Task
Thread
ThreadPool's thread
Paraller.For/ForEach/Invoke
I.e. when I create a Task and run it, where does it get a thread to execute on? And when I call Parallel.* what is really going on under the covers?
Any links to articles, blogposts, etc are also very welcomed!
The ideal state of a system is to have 1 actively running thread per CPU core. By defining work in more general terms of "tasks", the TPL can dynamically decide how many threads to use and which tasks to do on each one in order to come closest to achieving that ideal state. These are decisions that are almost always best made dynamically at runtime because when writing the code you can't know for sure how many CPU cores will be available to your application, how busy they are with other work, etc.
Thread: is a real OS thread, has handle and ID.
ThreadPool: is a collection of already-created OS Threads. These threads are owned/maintained by the runtime, and your code is only allowed to "borrow" them for a while, you can only do work short-termed work in these threads, and you can't modify any thread state, nor delete these threads.
Best guesses on these two:
Task: might get run on a pre-created thread in the thread pool, or might get run as part of user-mode scheduling, this is all depending on what the runtime thinks is best. Another guess: with TPL, the user-mode scheduling is NOT based on OS Fibers, but is its own complete (and working) implementation).
Parallel.For: actually, no clue how this is implemented. The runtime might create new threads to do the parallel bits, or much more likely use the thread pool's threads for the parallelism.
Related
I'm trying to make a thread pool for a game engine and I've been considering how my system should react to third party libraries spawning their own threads.
From what I've read, it is ideal to only have one thread for each CPU you have access to. So if my third party physics update spawns four threads, it would be ideal to turn off four threads from my thread pool while it is running, then turn them back on afterwards, that way multiple threads are never contending over one CPU.
My question is about the underlying mechanics behind functionality like conditional variables. Since spawning threads is expensive, having four threads wait on a conditional variable and then notifying them when the physics is done seems like a much better option than joining four threads and re-spawning them afterwards. But if they are waiting on a variable, are the threads truly "asleep" or are they still contending for CPU resources in the background?
Although you did not write what platform you are programming on, in most implementations threads that are waiting consume little to no CPU resources.
They do however use some memory (to save the stack, etc.), so you should avoid spawning an excessive number of threads and trying to reuse them as much as possible, since as you noted, spawning a new thread is an expensive operation on most platforms.
Even though you did not provide a lot of information, I'm guessing that in your scenario letting the threads wait is a much better option, as a small number of threads will not use a lot of resources and possibly having to spawn new threads frequently will affect performance badly on almost all platforms.
Any good third party library should give you the option of running it's work through your thread pool, to avoid that problem in the first place.
For example here's the documentation on how you can do that with PhysX - https://developer.nvidia.com/sites/default/files/akamai/physx/Docs/TaskManager.html
I've learned that a process has running, ready, blocked, and suspended states. Threads also have these states except for suspended because it lives in the process's address space.
A process blocks most of the time when it is doing a blocking i/o or waiting for an event.
I can easily picture out a process getting blocked if its single-threaded or if it follows a one-to-many model, but how does it work if the process is multi-threaded?
For example:
I have a process with two threads in a system that follows a one-to-one model. One handles the gui and the other handles the blocking i/o. I know the process remains responsive because the other thread handles the i/o.
So is there by any chance the process gets blocked or should I just rule it out in this case?
I'm just getting into these stuff so forgive me If I haven't understand some of the important details yet.
Let's say you have a work queue where the UI thread schedules work to be done and the I\O thread looks there for work to do. The work queue itself is data that is read and modified from both threads, therefor you must synchronize access somehow or race conditions result.
The naive approach is to synchronize access to the queue using a lock (aka critical section). If the I\O thread acquires the lock and then blocks, the UI thread will only remain responsive until it decides it needs to schedule work and tries to acquire the lock. A better approach is to use a lock-free queue about which much has been written and you can easily search for more info.
But to answer your question, yes, it is still much easier than you might think to cause UI to stutter / hang even when using multiple threads. There are various libraries that make it easier or harder to solve this problem, so depending on your OS and language of choice, there may be something better than just OS primitives. Win32 (from what I remember) doesn't it make it very easy at all despite having all sorts of synchronization primitives. Pthreads and Boost never seemed very straightforward to me either. Apple's GCD makes it semantically much easier to express what you want (in my opinion), though there are still pitfalls one must be aware of (such as scheduling too many blocking operations on a single work queue to be done in parallel and causing the processor to thrash when they all wake up at the same time).
My advice is to just dive in and write lots of multithreaded code. It can be tough to debug but you will learn a lot and eventually it becomes second nature.
Suppose we have a dual-core machine with a mainstream, modern OS capable to utilize both the cores.
If I have two threads, P1 and Q1 within the same process, and they happen to commence creating child threads, say, P2 and Q2, at approximately the same machine cycle, will OS perform the thread creation concurrently?
I heard thread creation is expensive, so the question came forth...
Thanks in advance.
Any reasonably well designed OS can have multiple processors executing kernel code at the same time. Therefore some of the tasks involved in a thread creation can be happening concurrently. But there will be some necessary serialization to manipulate some shared data structures (e.g. allocating memory, inserting a newly created threat structure into a global list). The processors could contend for the same lock thereby reducing concurrency.
Systems/applications which make new threads so often that the overhead of thread creation actually matters are probably designed wrong (doing too little useful work in a thread relative to the startup time, and not taking advantage of the obvious optimization of reusing short-lived threads from a pool).
It will be sorta-concurrently. There are aspects of thread-creation that cannot proceed in parallel - it would be unfortunate if the kernel memory-manager allocated both threads the same stack!
Thread creation is sufficiently expensive that it's worth while avoiding doing it at all during an app. run, hence the popularity of thread pools. Long-running tasks that block can be threaded off and left for the life of the app - often this means that explicit thread termination, (awkward at best, almost impossible at worst, from user code), is not necessary.
I think developers continually start and stop threads because they like to think of them as 'functions', where you 'pass parameters' in at the start and 'return' results when the thread ends. Ths is not the best way of conceptualizing threads.
If anybody has had a lot of experience timing code running on the main VCL thread vs a background thread, I'd like to get an opinion. I have some code that does some heavy string processing running in my Delphi 6 application on the main thread. Each time I run an operation, the time for each operation hovers around 50 ms on a single thread on my i5 Quad core. What makes me really suspicious is that the same code running on an old Pentium 4 that I have, shows the same time for the operation when usually I see code running about 4 times slower on the Pentium 4 than the Quad Core. I am beginning to wonder if the code might be consuming significantly less time than 50 ms but that there's something about the main VCL thread, perhaps Windows message handling or executing Windows API calls, that is creating an artificial "floor" for the operation. Note, an operation is triggered by an incoming request on a socket if that matters, but the time measurement does not take place until the data is fully received.
Before I undertake the work of moving all the code on to a background thread for testing, I am wondering if anyone has any general knowledge in this area? What have your experiences been with code running on and off the main VCL thread? Note, the timing measurements are being done when there is absolutely no user triggered activity going on during the tests.
I'm also wondering if raising the priority of the thread to just below real-time would do any good. I've never seen much improvement in my run times when experimenting with those flags.
-- roschler
Given all threads have the same priority, as they normally do, there can't be a difference, for the following reasons. If you're seeing a difference, re-evaluate the code (make sure you run the same thing in both VCL and background threads) and make sure you time it properly:
The compiler generates the exact same code, it doesn't care if the code is going to run in the main thread or in a background thread. In fact you can put the whole code in a procedure and call that from both your worker thread's Execute() and from the main VCL thread.
For the CPU all cores, and all threads, are equal. Unless it's actually a Hyper Threading CPU, where not all cores are real, but then see the next bullet.
Even if not all CPU cores are equal, your thread will very unlikely run on the same core, the operating system is free to move it around at will (and does actually schedule your thread to run on different cores at different times).
Messaging overhead doesn't matter for the main VCL thread, because unless you're calling Application.ProcessMessages() manually, the message pump is simply stopped while your procedure does it's work. The message pump is passive, your thread needs to request messages from the queue, but since the thread is busy doing your work, it's not requesting any messages so no overhead there.
There's just one place where threads are not equal, and this can change the perceived speed of execution: It's the operating system that schedules threads to execution units (cores), and for the operating system threads have different priorities. You can tell the OS a certain thread needs to be treated differently using the SetThreadPriority() API (which is used by the TThread.Priority property).
Without simple source code to reproduce the issue, and how you are timing your threads, it will be difficult to understand what occurs in your software.
Sounds definitively like either:
An Architecture issue - how are your threads defined?
A measurement issue - how are you timing your threads?
A typical scaling issue of both the memory manager and the RTL string-related implementation.
About the latest point, consider this:
The current memory manager (FastMM4) is not scaling well on multi-core CPU; try with a per-thread memory manager, like our experimental SynScaleMM - note e.g. that the Free Pascal Compiler team has written a new scaling MM from scratch recently, to avoid such issue;
Try changing the string process implementation to avoid memory allocation (use static buffers), and string reference-counting (every string reference counting access produces a LOCK DEC/INC which do not scale so well on multi-code CPU - use per-thread char-level process, using e.g. PChar on static buffers instead of string).
I'm sure that without string operations, you'll find that all threads are equivalent.
In short: neither the current Delphi MM, neither the current string implementation scales well on multi-core CPU. You just found out a known issue of the current RTL. Read this SO question.
When your code has control of the VCL thread, for instance if it is in one method and doesn't call out to any VCL controls or call Application.ProcessMessages, then the run time will not be affected just because it's in the main VCL thread.
There is no overhead, since you "own" the whole processing power of the thread when you are in your own code.
I would suggest that you use a profiling tool to find where the actual bottleneck is.
Performance can't be assessed statically. For that you need to get AQTime, or some other performance profiler for Delphi. I use AQtime, and I love it, but I'm aware it's considered expensive.
Your code will not magically get faster just because you moved it to a background thread. If anything, your all-inclusive-time until you see results in your UI might get a little slower, if you have to send a lot of data from the background thread to the foreground thread via some synchronization mechanisms.
If however you could execute parts of your algorithm in parallel, that is, split your work so that you have 2 or more worker threads processing your data, and you have a quad core processor, then your total time to do a fixed load of work, could decrease. That doesn't mean the code would run any faster, but depending on a lot of factors, you might achieve a slight benefit from multithreading, up to the number of cores in your computer. It's never ever going to be a 2x performance boost, to use two threads instead of one, but you might get 20%-40% better performance, in your more-than-one-threaded parallel solutions, depending on how scalable your heap is under multithreaded loads, and how IO/memory/cache bound your workload is.
As for raising thread priorities, generally all you will do there is upset the delicate balance of your Windows system's performance. By raising the priorities you will achieve (sometimes) a nominal, but unrepeatable and non-guaranteeable increase in performance. Depending on the other things you do in your code, and your data sources, playing with priorities of threads can introduce subtle problems. See Dining Philosophers problem for more.
Your best bet for optimizing the speed of string operations is to first test it and find out exactly where it is using most of its time. Is it heap operations? Memory Copy and move operations? Without a profiler, even with advice from other people, you will still be comitting a cardinal sin of programming; premature optimization. Be results oriented. Be science based. Measure. Understand. Then decide.
Having said that, I've seen a lot of horrible code in my time, and there is one killer thing that people do that totally kills their threaded app performance; Using TThread.Synchronize too much.
Here's a pathological (Extreme) case, that sadly, occurs in the wild fairly frequently:
procedure TMyThread.Execute;
begin
while not Terminated do
Synchronize(DoWork);
end;
The problem here is that 100% of the work is really done in the foreground, other than the "if terminated" check, which executes in the thread context. To make the above code even worse, add a non-interruptible sleep.
For fast background thread code, use Synchronize sparingly or not at all, and make sure the code it calls is simple and executes quickly, or better yet, use TThread.Queue or PostMessage if you could really live with queueing main thread activity.
Are you asking if a background thread would be faster? If your background thread would run the same code as the main thread and there's nothing else going on in the main thread, you don't stand to gain anything with a background thread. Threads should be used to split and distribute processing loads that would otherwise contend with one another and/or block one another when running in the main thread. Since you seem to be dealing with a case where your main thread is otherwise idle, simply spawning a thread to run slow code will not help.
Threads aren't magic, they can't speed up slow code or eliminate processing bottlenecks in a particular segment not related to contention on the main thread. Make sure your code isn't doing something you don't know about and that your timing methodology is correct.
My first hunch would be that your interaction with the socket is affecting your timing in a way you haven't detected... (I know you said you're sure that's not involved - but maybe check again...)
Can someone list some comparison points between Thread Spawning vs Thread Pooling, which one is better? Please consider the .NET framework as a reference implementation that supports both.
Thread pool threads are much cheaper than a regular Thread, they pool the system resources required for threads. But they have a number of limitations that may make them unfit:
You cannot abort a threadpool thread
There is no easy way to detect that a threadpool completed, no Thread.Join()
There is no easy way to marshal exceptions from a threadpool thread
You cannot display any kind of UI on a threadpool thread beyond a message box
A threadpool thread should not run longer than a few seconds
A threadpool thread should not block for a long time
The latter two constraints are a side-effect of the threadpool scheduler, it tries to limit the number of active threads to the number of cores your CPU has available. This can cause long delays if you schedule many long running threads that block often.
Many other threadpool implementations have similar constraints, give or take.
A "pool" contains a list of available "threads" ready to be used whereas "spawning" refers to actually creating a new thread.
The usefulness of "Thread Pooling" lies in "lower time-to-use": creation time overhead is avoided.
In terms of "which one is better": it depends. If the creation-time overhead is a problem use Thread-pooling. This is a common problem in environments where lots of "short-lived tasks" need to be performed.
As pointed out by other folks, there is a "management overhead" for Thread-Pooling: this is minimal if properly implemented. E.g. limiting the number of threads in the pool is trivial.
For some definition of "better", you generally want to go with a thread pool. Without knowing what your use case is, consider that with a thread pool, you have a fixed number of threads which can all be created at startup or can be created on demand (but the number of threads cannot exceed the size of the pool). If a task is submitted and no thread is available, it is put into a queue until there is a thread free to handle it.
If you are spawning threads in response to requests or some other kind of trigger, you run the risk of depleting all your resources as there is nothing to cap the amount of threads created.
Another benefit to thread pooling is reuse - the same threads are used over and over to handle different tasks, rather than having to create a new thread each time.
As pointed out by others, if you have a small number of tasks that will run for a long time, this would negate the benefits gained by avoiding frequent thread creation (since you would not need to create a ton of threads anyway).
My feeling is that you should start just by creating a thread as needed... If the performance of this is OK, then you're done. If at some point, you detect that you need lower latency around thread creation you can generally drop in a thread pool without breaking anything...
All depends on your scenario. Creating new threads is resource intensive and an expensive operation. Most very short asynchronous operations (less than a few seconds max) could make use of the thread pool.
For longer running operations that you want to run in the background, you'd typically create (spawn) your own thread. (Ab)using a platform/runtime built-in threadpool for long running operations could lead to nasty forms of deadlocks etc.
Thread pooling is usually considered better, because the threads are created up front, and used as required. Therefore, if you are using a lot of threads for relatively short tasks, it can be a lot faster. This is because they are saved for future use and are not destroyed and later re-created.
In contrast, if you only need 2-3 threads and they will only be created once, then this will be better. This is because you do not gain from caching existing threads for future use, and you are not creating extra threads which might not be used.
It depends on what you want to execute on the other thread.
For short task it is better to use a thread pool, for long task it may be better to spawn a new thread as it could starve the thread pool for other tasks.
The main difference is that a ThreadPool maintains a set of threads that are already spun-up and available for use, because starting a new thread can be expensive processor-wise.
Note however that even a ThreadPool needs to "spawn" threads... it usually depends on workload - if there is a lot of work to be done, a good threadpool will spin up new threads to handle the load based on configuration and system resources.
There is little extra time required for creating/spawning thread, where as thread poll already contains created threads which are ready to be used.
This answer is a good summary but just in case, here is the link to Wikipedia:
http://en.wikipedia.org/wiki/Thread_pool_pattern
For Multi threaded execution combined with getting return values from the execution, or an easy way to detect that a threadpool has completed, java Callables could be used.
See https://blogs.oracle.com/CoreJavaTechTips/entry/get_netbeans_6 for more info.
Assuming C# and Windows 7 and up...
When you create a thread using new Thread(), you create a managed thread that becomes backed by a native OS thread when you call Start – a one to one relationship. It is important to know only one thread runs on a CPU core at any given time.
An easier way is to call ThreadPool.QueueUserWorkItem (i.e. background thread), which in essence does the same thing, except those background threads aren’t forever tied to a single native thread. The .NET scheduler will simulate multitasking between managed threads on a single native thread. With say 4 cores, you’ll have 4 native threads each running multiple managed threads, determined by .NET. This offers lighter-weight multitasking since switching between managed threads happens within the .NET VM not in the kernel. There is some overhead associated with crossing from user mode to kernel mode, and the .NET scheduler minimizes such crossing.
It may be important to note that heavy multitasking might benefit from pure native OS threads in a well-designed multithreading framework. However, the performance benefits aren’t that much.
With using the ThreadPool, just make sure the minimum worker thread count is high enough or ThreadPool.QueueUserWorkItem will be slower than new Thread(). In a benchmark test looping 512 times calling new Thread() left ThreadPool.QueueUserWorkItem in the dust with default minimums. However, first setting the minimum worker thread count to 512, in this test, made new Thread() and ThreadPool.QueueUserWorkItem perform similarly.
A side effective of setting a high worker thread count is that new Task() (or Task.Factory.StartNew) also performed similarly as new Thread() and ThreadPool.QueueUserWorkItem.