Re: Per-Processor Data Paget

David Lang (dlang@diginsite.com)
Thu, 9 Dec 1999 22:33:17 -0800 (PST)


in that case you still have one thread per connection. I am not saying you
spawn a new thread, re-using an old thread is just as good.

David Lang

On Thu, 9 Dec 1999, Thomas Duffy wrote:

> Date: Thu, 09 Dec 1999 19:07:00 -0800
> From: Thomas Duffy <tduffy@cthulhu.engr.sgi.com>
> Reply-To: tduffy@sgi.com
> To: linux-kernel@vger.rutgers.edu, David Lang <dlang@diginsite.com>
> Subject: Re: Per-Processor Data Paget
>
> David Lang wrote:
> >
> > It depends on if you are doing a CPU intensive program or one that talks
> > to the outside world.
> >
> > for example in a threaded web server you may want to have one thread for
> > each request , instead of one thread per CPU.
>
> This would be *very* expensive to do. More likely, you would want to
> have a pool of threads already spun which would select on several
> descriptors. That way you could take advantage of multiproccessor
> systems and the benefits of threads without the overhead of spawning a
> new thread for every connection which could get huge on a big server.
>
> --
> _
> Tom Duffy __ (_)
> ____ _/,-' __
> / .__) / _ \ |_ |
> \__. \( (_) ) | |
> (____/ \ __/ | |_
> //--, `--'
> ( <> ) Core Linux
> `--'
>

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/