Maximum number of threads allowed in Windows

By default, a thread stack needs to reserve 1 MB of memory space.
However, a process has only 2 GB of memory available. Therefore, in theory, a process can open up to 2048 threads.
However, the memory certainly cannot be used as a thread stack completely, so the actual number is smaller than this value.
You can also modify the default stack size during connection to reduce the size so that you can open more threads.
For example, change the default stack size to 512 KB, so that up to 4096 threads can be opened theoretically.

Even if the physical memory is larger, the number of threads that can be started in a process is limited by 2 GB.
For example, your machine has 64 GB physical memory, but the memory space of each process is 4 GB, of which the user mode is 2 GB.

The number of threads that can be started on the same machine is also limited by the memory. Every thread object must use non-page memory instead of page memory, which is also limited. When non-page memory is exhausted, the thread cannot be created.

If the physical memory is very large, the maximum number of threads that can be run on the same machine will increase.

Write a file in WindowsProgramA process fork generates about 2000 threads and then exits abnormally. Why?

This problem occurs because in Windows 32-bit system, a process can use a maximum of 2 GB of virtual memory, and the default thread stack stacksize of a thread is 1024 K (1 m ), in this way, when the number of threads approaches 2000, 2000*1024 K = 2G (approximately), memory resources are equivalent to depletion.

 

Msdn Original:

"The number of threads a process can create is limited by the available virtual memory. By default, every thread has one megabyte of stack space. Therefore, you can create at most 2,028 threads.If you reduce the default stack size, you can create more threads.However, your application will have better performance if you create one thread per processor and build queues of requests for which the application maintains the context information. A thread wocould process all requests in a queue before processing requests in the next queue."

 

How can we break through the 2000 limits?

You can reduce the stack stacksize by modifying the createthread parameter. For example

# DefineMax_threads 50000
 

DWORD winapi threadproc (lpvoid lpparam ){
 

 
While(1 ){
 

Sleep (100000 );
 

 
}
 

Return0;
 

 
}
 

 
IntMain (){
 

DWORD dwthreadid [max_threads];
 

 
Handle hthread [max_threads];
 

 
For(IntI = 0; I <max_threads; ++ I)
 

{
 

 
Hthread [I] = createthread (0, 64, threadproc, 0, stack_size_param_is_a_reservation, & dwthreadid [I]);
 

 
If(0 = hthread [I])
 

{
 

 
Dword e = getlasterror ();
 

Printf ("% D \ r \ n", E );
 

 
Break;
 

}
 

 
}
 

 
Threadproc (0 );
 

}
 

 

Server programming

If your server program is designed to create a thread for a client connection request, there will be 2000 limits (when the number of Hardware Memory and CPU is fixed ). The suggestions are as follows:

The "one thread per client" model is well-known not to scale beyond a dozen clients or so. if you're going to be handling more than that has clients simultaneously, you should move to a model where instead of dedicating a thread to a client, you instead allocate an object. (someday I'll muse on the duality between threads and objects .) windows provides I/O Completion Ports andThread PoolTo help you convert from a thread-based model to a work-item-based model.

1. Serve serving clients with each thread, and use nonblocking I/O and Level-triggered Readiness notification2. serve serving clients with each thread, and use nonblocking I/O and readiness Change Icationication3. serve your clients with each server thread, and use asynchronous I/O