- About my new algorithm - 1 Update
- cmsg cancel <mjvfee$d2v$2@dont-email.me> - 4 Updates
- My efficient Threadpool engine tutorial - 2 Updates
- Efficient Threadpool engine with priorities was updated to version 2.1 - 1 Update
Ramine <ramine@1.1>: May 25 02:14PM -0700 Hello, You have to know that to design and implement an efficient Threadpool as my efficient Threadpool engine with priorities that is correct and efficient, it is somewhat a hard job, so to facilitate the reasonning about concurrent programming i have used my scalable AMLock around a small portion of the TPThreadPoolThread.Execute() method of my efficient Threadpool engine with priorities to facilitate the reasonning about "correctness", and i think that now that my new algorithm has facilitates the reasonning about correctness and now that i have tested it thoroughly , you can be more confident cause i think that my new algorithm of my efficient Threadpool engine with priorities is correct and stable now and it is also fast. And look at the ThreadPoolExecutor Class of Java, look for example at the awaitTermination() method, it says: --- boolean awaitTermination(long timeout, TimeUnit unit) Blocks until all tasks have completed execution after a shutdown request, or the timeout occurs, or the current thread is interrupted, whichever happens first. -- read more here: https://docs.oracle.com/javase/8/docs/api/java/util/concurrent/ThreadPoolExecutor.html#method.summary Did you notice ? In Java when you wait for the tasks you have to wait for "ALL" the tasks, and that's not efficient , and if you want to use the object from multiple threads i think it will have the same effect, you can avoid some of the problems by using many objects of the ThreadPoolExecutor class but this will take ressources and this will cause more and more context switches and that's bad, i think C# has the same problem, other than that Java and C# don't support priorities, it means that you can not give priorities to tasks/jobs, like high or normal or low, and that's not good for games and other applications where you have to use priorities even if the system is not a realtime system, this is why i have decided to implement my efficient Threadpool engine version 2.1 that supports those characteristics, so that you can create a child object of the Threadpool class that will use the same worker threads and that will wait only for the tasks that you will add with the execute() method , and also my efficient Threadpool engine supports 3 priorities, High and normal and low, that's where my efficient Threadpool engine comes in hand and that's where it's efficient. Hope you will like it. I will talk about an important subject that is software "reliability", you have seen me explaining to you what is that it's efficient and reliable with my efficient Threadpool engine, and you have seen me talking to you about the automaton of my efficient Threadpool engine , this automaton takes care of software "reliability", like in design by contracts that takes care of realiability, i have finally been able to render my efficient Threadpool engine a reliable software by using an automaton that "guides" you by helping you to avoid forbidden transitions that can cause problems like deadlocks and dangerous problems, my automaton has made my efficient threadpool engine a reliable software, i explain more: since i am using 3 methods called execute() that distributes the jobs to the worker threads and i am using also setCounter() method with a boolean as a parameter and using also wait() method that waits for the jobs to finish (this look like a join()) with two parameters, but what i have noticed is that you can call those methods by combining them in a different ways , but this can cause some combinations that are forbidden to be called and that can cause deadlock or dangerous problems, this is why i have wrote an automaton that help you and guide you by using also exception handling to use the right combinations of those methods and there parameters, so like in design by contracts, my automaton has made my efficient Threadpool engine a reliable software. Please read the HTML tutorial inside the zip file to understand how to use the execute() and setCounter() and wait() methods etc. You can download my efficient Threadpool engine from: https://sites.google.com/site/aminer68/threadpool-with-priorities Thank you, Amine Moulay Ramdane. |
bleachbot <bleachbot@httrack.com>: May 25 05:35PM +0200 |
bleachbot <bleachbot@httrack.com>: May 25 06:35PM +0200 |
bleachbot <bleachbot@httrack.com>: May 25 06:54PM +0200 |
bleachbot <bleachbot@httrack.com>: May 25 08:12PM +0200 |
Ramine <ramine@1.1>: May 25 12:36PM -0700 Hello, Here is my efficient Threadpool engine tutorial On a multicore system, your goal is to spread the work efficiently among many cores so that it does executes simultaneously. And performance gain should be directly related to how many cores you have. So, a quad core system should be able to get the work done 4 times faster than a single core system. A 16-core platform should be 4-times faster than a quad-core system, and 16-times faster than a single core... That's where my efficient Threadpool is useful , it spreads the work efficiently among many cores. Threadpool (and Threadpool with priorities) consist of a fast concurrent FIFO queue, so when you call ThreadPool.execute() , your work item get queued in thethe fast concurrent queue. The worker threads pick them out in a First In First Out order (i.e., FIFO order), and execute them. The following have been added to the efficient Threadpool engine: * The worker threads enters in a wait state when there is no job in the concurrent FIFO queues - for more efficiency - * You can distribute your jobs to the worker threads and call any method with the threadpool's execute() method. * You can wait for the jobs to finish with the wait() method. My Threadpool allows load balancing, and also minimize contention. Threadpool is very easy to use, let's look now at an example in Object Pascal... program test; uses {$IFDEF Delphi} cmem, {$ENDIF} ThreadPool,sysutils,syncobjs; {$I defines.inc} type TMyThread = class (TThreadPoolThread) //procedure ProcessRequest(obj: Pointer); override; procedure MyProc1(obj: Pointer); procedure MyProc2(obj: Pointer); end; var myobj:TMyThread; TP: TThreadPool; obj:pointer; cs: TCriticalSection; tp1:TThreadPoolC; procedure TMyThread.MyProc1(obj: Pointer); begin cs.enter; writeln('This is MyProc1 with parameter: ',integer(obj)); cs.leave; end; procedure TMyThread.MyProc2(obj: Pointer); begin cs.enter; writeln('This is MyProc2 with parameter: ',integer(obj)); cs.leave; end; begin myobj:=TMyThread.create; cs:=TCriticalSection.create; TP := TThreadPool.Create(4, TMyThread); // 4 workers threads. tp1:=TP.CreateThreadPoolC(true); obj:=pointer(1); TP1.execute(myobj.myproc1,pointer(obj)); obj:=pointer(2); TP1.execute(myobj.myproc2,pointer(obj)); TP1.setCounter() ; TP1.wait(INFINITE); readln; TP1.free; TP.Terminate; TP.Free; end. Let us look at the first line... uses {$IFDEF Delphi} cmem, {$ENDIF} cmem is required for Delphi to use TBB memory manager (from Intel) , this will allow delphi memory manager to scale linearely... Note: FPC doesn't need cmem, cause it scales linearely with threads... ThreadPool: is our threadpool unit .. syncobjs: contains all the sychronizations stuff like CriticalSections, Events etc.. After that we have the following lines: type TMyThread = class (TThreadPoolThread) //procedure ProcessRequest(obj: Pointer); override; procedure MyProc1(obj: Pointer); procedure MyProc2(obj: Pointer); end; We declare a TMyThread that ineherit from TThreadPoolThread, and we declare our two methods MyProc1 and MyProc2 that we want to be executed by our threadpool's worker threads. Each method has an obj as a paramater. In the main body we create a TMyThread object like this: myobj:=TMyThread.create; and after that we create a TThreadPool object with 4 workers threads like this: TP := TThreadPool.Create(4, TMyThread); // 4 workers threads. After that you create ThreadPoolC object so that you can distribute your jobs to the ThreadPool by writing this: tp1:=TP.CreateThreadPoolC(true); If you pass true to CreateThreadPoolC() , wait() will wait for all the jobs to finish, if you pass false to CreateThreadPoolC(), wait() will not wait for the jobs to finish. After that we distribute to our worker threads the methods to be executed , we do it by calling the Threadpool's execute() method and we pass it myobj.myproc1 and myobj.myproc2 with there parameters:. After that you have to call setCounter() and wait() to wait for all the threads.. TP.1.setCounter(); TP1.wait(INFINITE); You have to call setCounter() , it is mandatory, the first parameter of wait() is the time to wait in milliseconds, if you set it to INFINITE, it will wait for the threads to finish. After that you have to free TP1 and TP objects. As you see, Threadpool (and threadpool with priority) is very easy to use... Let's look now at an example of a Threadpool with priority:. program test; uses {$IFDEF Delphi} cmem, {$ENDIF} PThreadPool,sysutils,syncobjs; {$I defines.inc} type TMyThread = class (TPThreadPoolThread) //procedure ProcessRequest(obj: Pointer); override; procedure MyProc1(obj: Pointer); procedure MyProc2(obj: Pointer); end; var myobj:TMyThread; TP: TPThreadPool; obj:pointer; cs:TCriticalSection; tp1:TPThreadPoolC; procedure TMyThread.MyProc1(obj: Pointer); begin cs.enter; writeln('This is MyProc1 with parameter: ',integer(obj)); cs.leave; end; procedure TMyThread.MyProc2(obj: Pointer); begin cs.enter; writeln('This is MyProc2 with parameter: ',integer(obj)); cs.leave; end; begin myobj:=TMyThread.create; cs:=TCriticalSection.create; TP := TPThreadPool.Create(4,TMyThread, 20); // 4 workers threads and 2^20 items for each queue. tp1:=TP.CreateThreadPoolC(true); obj:=pointer(1); TP1.execute(myobj.myproc1,pointer(obj),NORMAL_PRIORITY); obj:=pointer(2); TP1.execute(myobj.myproc2,pointer(obj),NORMAL_PRIORITY); TP1.setCounter(); TP1.wait(INIFINITE); readln; TP.Terminate; TP.Free; end. As you have noticed, this is almost the same as threadpool.. You use PThreadPool - P for priority - rather than Threadpool TPThreadPoolThread rather than TThreadPoolThread TPThreadPoolC rather than TThreadPoolC TPThreadPool.Create rather than TThreadPool.Create and as you have noticed in TP.execute(myobj.myproc1,pointer(obj),NORMAL_PRIORITY) we are using priorities. You can give the following priorities to jobs: LOW_PRIORITY NORMAL_PRIORITY HIGH_PRIORITY That's all. You can download threadpool (and threadpool with priority) from: https://sites.google.com/site/aminer68/ Sincerely, Amine Moulay Ramdane. |
Ramine <ramine@1.1>: May 25 12:55PM -0700 Hello, Here is the new updated tutorial: On a multicore system, your goal is to spread the work efficiently among many cores so that it does executes simultaneously. And performance gain should be directly related to how many cores you have. So, a quad core system should be able to get the work done 4 times faster than a single core system. A 16-core platform should be 4-times faster than a quad-core system, and 16-times faster than a single core... That's where my efficient Threadpool is useful , it spreads the work efficiently among many cores. Threadpool (and Threadpool with priorities) consist of a fast concurrent FIFO queue and a number of worker threads that you have to start, so when you call ThreadPool.execute() , your work item get queued in the fast concurrent queue. The worker threads pick them out in a First In First Out order (i.e., FIFO order), and execute them. The following have been added to the efficient Threadpool engine: * The worker threads enters in a wait state when there is no job in the concurrent FIFO queues - for more efficiency - * You can distribute your jobs to the worker threads and call any method with the threadpool's execute() method. * You can wait for the jobs to finish with the wait() method. My Threadpool allows load balancing, and also minimize contention. Threadpool is very easy to use, let's look now at an example in Object Pascal... program test; uses {$IFDEF Delphi} cmem, {$ENDIF} ThreadPool,sysutils,syncobjs; {$I defines.inc} type TMyThread = class (TThreadPoolThread) //procedure ProcessRequest(obj: Pointer); override; procedure MyProc1(obj: Pointer); procedure MyProc2(obj: Pointer); end; var myobj:TMyThread; TP: TThreadPool; obj:pointer; cs: TCriticalSection; tp1:TThreadPoolC; procedure TMyThread.MyProc1(obj: Pointer); begin cs.enter; writeln('This is MyProc1 with parameter: ',integer(obj)); cs.leave; end; procedure TMyThread.MyProc2(obj: Pointer); begin cs.enter; writeln('This is MyProc2 with parameter: ',integer(obj)); cs.leave; end; begin myobj:=TMyThread.create; cs:=TCriticalSection.create; TP := TThreadPool.Create(4, TMyThread); // 4 workers threads. tp1:=TP.CreateThreadPoolC(true); obj:=pointer(1); TP1.execute(myobj.myproc1,pointer(obj)); obj:=pointer(2); TP1.execute(myobj.myproc2,pointer(obj)); TP1.setCounter() ; TP1.wait(INFINITE); readln; TP1.free; TP.Terminate; TP.Free; end. Let us look at the first line... uses {$IFDEF Delphi} cmem, {$ENDIF} cmem is required for Delphi to use TBB memory manager (from Intel) , this will allow delphi memory manager to scale linearely... Note: FPC doesn't need cmem, cause it scales linearely with threads... ThreadPool: is our threadpool unit .. syncobjs: contains all the sychronizations stuff like CriticalSections, Events etc.. After that we have the following lines: type TMyThread = class (TThreadPoolThread) //procedure ProcessRequest(obj: Pointer); override; procedure MyProc1(obj: Pointer); procedure MyProc2(obj: Pointer); end; We declare a TMyThread that ineherit from TThreadPoolThread, and we declare our two methods MyProc1 and MyProc2 that we want to be executed by our threadpool's worker threads. Each method has an obj as a paramater. In the main body we create a TMyThread object like this: myobj:=TMyThread.create; and after that we create a TThreadPool object with 4 workers threads like this: TP := TThreadPool.Create(4, TMyThread); // 4 workers threads. After that you create ThreadPoolC object so that you can distribute your jobs to the ThreadPool by writing this: tp1:=TP.CreateThreadPoolC(true); If you pass true to CreateThreadPoolC() , wait() will wait for all the jobs to finish, if you pass false to CreateThreadPoolC(), wait() will not wait for the jobs to finish. After that we distribute to our worker threads the methods to be executed , we do it by calling the Threadpool's execute() method and we pass it myobj.myproc1 and myobj.myproc2 with there parameters:. After that you have to call setCounter() and wait() to wait for all the jobs.. TP.1.setCounter(); TP1.wait(INFINITE); You have to call setCounter() , it is mandatory, the first parameter of wait() is the time to wait in milliseconds, if you set it to INFINITE, it will wait for the jobs to finish. After that you have to free TP1 and TP objects. As you see, Threadpool (and threadpool with priority) is very easy to use... Let's look now at an example of a Threadpool with priority:. program test; uses {$IFDEF Delphi} cmem, {$ENDIF} PThreadPool,sysutils,syncobjs; {$I defines.inc} type TMyThread = class (TPThreadPoolThread) //procedure ProcessRequest(obj: Pointer); override; procedure MyProc1(obj: Pointer); procedure MyProc2(obj: Pointer); end; var myobj:TMyThread; TP: TPThreadPool; obj:pointer; cs:TCriticalSection; tp1:TPThreadPoolC; procedure TMyThread.MyProc1(obj: Pointer); begin cs.enter; writeln('This is MyProc1 with parameter: ',integer(obj)); cs.leave; end; procedure TMyThread.MyProc2(obj: Pointer); begin cs.enter; writeln('This is MyProc2 with parameter: ',integer(obj)); cs.leave; end; begin myobj:=TMyThread.create; cs:=TCriticalSection.create; TP := TPThreadPool.Create(4,TMyThread, 20); // 4 workers threads and 2^20 items for each queue. tp1:=TP.CreateThreadPoolC(true); obj:=pointer(1); TP1.execute(myobj.myproc1,pointer(obj),NORMAL_PRIORITY); obj:=pointer(2); TP1.execute(myobj.myproc2,pointer(obj),NORMAL_PRIORITY); TP1.setCounter(); TP1.wait(INIFINITE); readln; TP.Terminate; TP.Free; end. As you have noticed, this is almost the same as threadpool.. You use PThreadPool - P for priority - rather than Threadpool TPThreadPoolThread rather than TThreadPoolThread TPThreadPoolC rather than TThreadPoolC TPThreadPool.Create rather than TThreadPool.Create and as you have noticed in TP.execute(myobj.myproc1,pointer(obj),NORMAL_PRIORITY) we are using priorities. You can give the following priorities to jobs: LOW_PRIORITY NORMAL_PRIORITY HIGH_PRIORITY That's all. You can download threadpool (and threadpool with priority) from: https://sites.google.com/site/aminer68/ Sincerely, Amine Moulay Ramdane. |
Ramine <ramine@1.1>: May 25 11:37AM -0700 Hello, My efficient Threadpool engine with priorities was updated to version 2.1, i have changed the concurrent FIFO queue of my efficient Threadpool engine with priorities, now the concurrent FIFO queue is waitfree on the push() and lockfree on the pop(), so it's really fast, and i have also changed the algorithm of my efficient Threadpool engine and now i have tested it thoroughly and i think that it's a stable and a correct algorithm and it's fast, i have also updated the HTML tutorial inside the zipfile, please look inside the tutorial to learn how to use my efficient Threadpool engine. I have also updated my Threadpool engine to version 2.1.. You can download my efficient Threadpool engine with priorities version 2.1 from: https://sites.google.com/site/aminer68/threadpool-with-priorities And you can download my Threadpool engine version 2.1 from: https://sites.google.com/site/aminer68/threadpool Thank you, Amine Moulay Ramdane. |
You received this digest because you're subscribed to updates for this group. You can change your settings on the group membership page. To unsubscribe from this group and stop receiving emails from it send an email to comp.programming.threads+unsubscribe@googlegroups.com. |
No comments:
Post a Comment