Re: kernel thread support - LWP's
(no name) ((no email))
Thu, 15 Jul 1999 15:46:52 -0400
On Thu, Jul 15, 1999 at 03:04:23PM -0400, Benjamin LaHaise wrote:
> On Thu, 15 Jul 1999, Nate Tuck wrote:
>
> > In cases where developer time (or time to market) is at a premium and the
> > problems at hand require some kind of modular parallelism. I have a
> > program on my desk that can run >20 active threads on a one CPU machine.
> > Given the problem(s) we had to solve with it, threads saved us many
> > man-months of development effort and were the right answer. I can imagine
> > that other developers have the same sorts of quandaries (or would if they
> > understood how threads could help them).
>
> I'll agree with you on the cost side of things, but are you certain that
> the model used for the application is correct? That is what I'm
> questioning, not the fact that your app can trigger >20 active threads.
> An event driven system can be worked on by just as many programmers as a
> threads based systems.
I don't think the modularity was referring to had anything to do with
multiple developers. I expect that he had several independent modules,
which could/should execute in parallel, and each of which had some amount
of state. By using multiple threads, it is frequently possible to
implicitly keep track of this state in a thread's stack.
To switch to an event-based model, that implicit state would have to be
explitly saved and restored when processing other events. Depending on
the application, that state could be arbitrarily complex, and
saving/restoring it could be hard to do quickly and correctly. Once
you've written all the save/restore code, you've essentially written your
own special-purpose user-level threading library, and each 'event' is
really just a 'context switch'.
-
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/