Overview
Examples
Screenshots
Comparisons
Applications
Download
Documentation
Tutorials
Bazaar
Status & Roadmap
FAQ
Authors & License
Forums
Funding Ultimate++
Search on this site
Search in forums












SourceForge.net Logo
Home » U++ Library support » U++ MT-multithreading and servers » CoWork question
CoWork question [message #17459] Fri, 15 August 2008 23:10 Go to next message
Infausto is currently offline  Infausto
Messages: 28
Registered: June 2008
Promising Member
I'm very noob using UPP but i think this platform is really amazing.
I wrote this post because i have a doubt about CoWork; Its CoWork some kind of IOCP implementation?.
As you can know I/O completion port is an asynchronous mechanism provided directly by the OS kernel to manage I/O request. This technique y very efficient to manipulate thousands of clients in an client-server architecture.
How it works this mechanism is very similar to the work of CoWork (only a few threads to handle many operations), so for this reason i have this doubt.

thx in advance.
Re: CoWork question [message #17587 is a reply to message #17459] Thu, 21 August 2008 16:58 Go to previous messageGo to next message
mirek is currently offline  mirek
Messages: 11715
Registered: November 2005
Ultimate Member
Infausto wrote on Fri, 15 August 2008 17:10

I'm very noob using UPP but i think this platform is really amazing.
I wrote this post because i have a doubt about CoWork; Its CoWork some kind of IOCP implementation?.
As you can know I/O completion port is an asynchronous mechanism provided directly by the OS kernel to manage I/O request. This technique y very efficient to manipulate thousands of clients in an client-server architecture.
How it works this mechanism is very similar to the work of CoWork (only a few threads to handle many operations), so for this reason i have this doubt.

thx in advance.


Not really. It is basically a loop parallelizer.

If you are performing loop and each iteration is independent (or mostly independent) from other, you can use CoWork to spread the work across all CPU cores.

In fact, in processing intensive code, it is often enought to parallelize a couple of loops where it is possible and you get app using multicore CPU nicely.

Note that CoWork does not solve sharing problem - if you share any data between loop iterations, you have to synchronize access as usual (using Mutex etc...)

Mirek

Mirek
Re: CoWork question [message #17619 is a reply to message #17587] Fri, 22 August 2008 20:22 Go to previous message
Infausto is currently offline  Infausto
Messages: 28
Registered: June 2008
Promising Member
ohh, ok, i completly wrong about this issue. So CoWork is similar to OpenMP, right?.

I want to write a IOCP implementation using UPP, but is very difficult to me make it without the proper socket documentation. So i have to wait until this docs are available.

bye.
Previous Topic: Socket documentation
Next Topic: two threads make error
Goto Forum:
  


Current Time: Sat Feb 23 20:31:35 CET 2019

Total time taken to generate the page: 0.01763 seconds