Hi,

I recently read this article: 

http://www.reuters.com/newsArticle.jhtml?type=scienceNews&storyID=3711894

I do not know much about parallel computing. But does this article
mean that since serial chip are becoming very fast, parallel computing
has hard days ahead?

Thanks.
Sharon

[c.p. moderator:
Keywords from the URL: optical DSP (special purpose).
There is insufficient technical detail in the article for substantive
evaluation on this special purpose development.  It gives an unqualified,
undetailed speed and some vague dimensions.  As such, I will keep
your cross post to the unmoderated pvm and mpi subgroups, but others
who wish to comment should look hard at keeping the full list of
newsgroups.

On my browser, despite the scienceNews in the URL, I also see a tab on
Finance section.  The reader needs to be aware of the URL's market speak.
To comment on your specific question:
The future is parallel at various level: the bit-word level, the
thread level, the heavier process level, and more.  The parallel
will subsume more serial processors (gangs, waves, or grids), but there will
still remain big difficult problems.  Still lots of work room.i

Years ago from /usr/games/fortune:
Bradley's Bromide:
        If computers get too powerful, we can organize them into a
        committee -- that will do them in.
]

I don't think so. The new chip has just arrived. Let's see how well it 
does in the real world.
Moreover it's a specialized DSP chip. A Linux cluster build from 
ordinary off the shelf components should be comparatively cheap.

Sharon wrote:
> Hi,
> 
> I recently read this article: 
> 
> http://www.reuters.com/newsArticle.jhtml?type=scienceNews&storyID=3711894
> 
> I do not know much about parallel computing. But does this article
> mean that since serial chip are becoming very fast, parallel computing
> has hard days ahead?
> 
> Thanks.
> Sharon
> 
> [c.p. moderator:
> Keywords from the URL: optical DSP (special purpose).
> There is insufficient technical detail in the article for substantive
> evaluation on this special purpose development.  It gives an unqualified,
> undetailed speed and some vague dimensions.  As such, I will keep
> your cross post to the unmoderated pvm and mpi subgroups, but others
> who wish to comment should look hard at keeping the full list of
> newsgroups.
> 
> On my browser, despite the scienceNews in the URL, I also see a tab on
> Finance section.  The reader needs to be aware of the URL's market speak.
> To comment on your specific question:
> The future is parallel at various level: the bit-word level, the
> thread level, the heavier process level, and more.  The parallel
> will subsume more serial processors (gangs, waves, or grids), but there will
> still remain big difficult problems.  Still lots of work room.i
> 
> Years ago from /usr/games/fortune:
> Bradley's Bromide:
>         If computers get too powerful, we can organize them into a
>         committee -- that will do them in.
> ]
>