[Baypiggies] concurrency talk

[Guido van Rossum]

I believe the failure modes and trust models are fundamentally
different. In distributed computing, you have to deal with nodes that
are lying (an important motivation for E, actually) as well as network
partitions where two groups of nodes continue to work but have
fundamentally different views of the world. In parallel computing,
there are no trust issues, and typically a failing CPU either causes
the entire computation to fail (I expect that this is how 2-8 core
systems deal with a single CPU failing -- at best you can disable a
CPU at boot time) or at least is taken out of the computation
permanently.

On 9/17/07, jim stockford <jim at well.com> wrote:
>
>     i'm easily confused (disclaimer), but i've got the idea that
> in-the-future, the problems presented by distributed computing
> and parallel processing may overlap, blur.... what's the truth
> to that idea? (if some, then maybe looking at E has some
> value.)
>
>
> On Sep 17, 2007, at 10:55 AM, Guido van Rossum wrote:
>
> > I was thinking of E as well, but then I realized that E supports
> > *distributed* computing, which isn't quite the same field as
> > *parallel* computing. I'm not sure it's useful to increase the field
> > even more by adding distributed computing -- then you'd have to add
> > MapReduce and many other paradigms as well.
> >
> > On 9/17/07, jim stockford <jim at well.com> wrote:
> >>
> >>     maybe add E to the list (below Erlang)?
> >>
> >>     could the discussoin also include likely new CPU
> >> designs, PCB designs, embedded and desktop and
> >> server systems as well.
> >>
> >>
> >> On Sep 17, 2007, at 2:48 AM, Shannon -jj Behrens wrote:
> >>
> >>> Someone hinted at this idea earlier, but it might be fun to have an
> >>> overview discussion on a bunch of different concurrency techniques.
> >>> We could have a bunch of speakers, each speaking for 15 minutes on a
> >>> specific topic.  For instance:
> >>>
> >>> * Processes
> >>> * Threads (kernel and green)
> >>> * Parallel programming vs. distributed computing
> >>> * IO bound vs. CPU bound
> >>> * Asynchronous
> >>> * Twisted
> >>> * Stackless
> >>> * Actors
> >>> * Erlang
> >>>
> >>> We wouldn't actually try to cover how to use each of these.  Rather,
> >>> the goal would be to explain what it is and what are its advantages
> >>> and limitations.  That'd be a fun talk to give assuming we can keep
> >>> it
> >>> shallow enough to cover everything but deep enough to make sense.
> >>>
> >>> Happy Hacking!
> >>> -jj
> >>>
> >>> --
> >>> http://jjinux.blogspot.com/
> >>> _______________________________________________
> >>> Baypiggies mailing list
> >>> Baypiggies at python.org
> >>> To change your subscription options or unsubscribe:
> >>> http://mail.python.org/mailman/listinfo/baypiggies
> >>>
> >>
> >> _______________________________________________
> >> Baypiggies mailing list
> >> Baypiggies at python.org
> >> To change your subscription options or unsubscribe:
> >> http://mail.python.org/mailman/listinfo/baypiggies
> >>
> >
> >
> > --
> > --Guido van Rossum (home page: http://www.python.org/~guido/)
> >
>
>


-- 
--Guido van Rossum (home page: http://www.python.org/~guido/)