Re[2]: [Insight-users] Streaming

[Miller, James V (Research)]

I admit the options are intriguing, particularly given the number of 
linux clusters and computation grids being assembled (we have several
of both on site).

Whether ITK adopts these capabilities is dependent on the community as 
a whole.  We need to balance the benefit with the complexity.  If a large
percentage of the community wants these types of capabilities, then we 
will certainly address this issue.  But if only a small community wants
these types of capabilities, I fear the complexity involved will outweigh
the benefit. This is not to say that is not going to happen.  Just that 
we need to scope the cost/benefit.

As an example, the research community is more likely to embrace running 
algorithms across a cluster.  The clinical community, however, is more 
geared towards shared memory configurations. (Personally, I want everything
to run my laptop. But I am probably an outlier here.)

The developer community for ITK is larger (by construction) than the 
developer community for VTK. Also, the types of algorithms in ITK 
(for instance, segmentation algorithms) are more complicated than the 
algorithms in VTK. These issues drive us to keep things as simple as 
possible under the hood.

Note, however, that since one can mix VTK and ITK, it would be possible
to used VTK's capability to bridge across a network to other cpus and still
use ITK in the pipeline housed on each cpu.
 

> -----Original Message-----
> From: Sebastian Fiorentini [mailto:sebasf@lncc.br]
> Sent: Wednesday, May 07, 2003 9:56 AM
> To: Miller, James V (Research)
> Cc: insight-users@public.kitware.com
> Subject: Re[2]: [Insight-users] Streaming
> 
> 
> MJVR> This level of complexity has not been added to ITK.
> MJVR> Supporting this feature requires adding complexity to the
> MJVR> underlying pipeline mechanism.  I hesitate to make the 
> MJVR> pipeline more complicated.
> But there is a future plan? Because, many interesting ideas can be
> applied (modern technologies have opened a new range of architechtures
> for parallel working). What do you think about that?
> 
> MJVR> I think I need to find a few "pipeline apprentices" so
> MJVR> that more people in the community understand the inner 
> MJVR> workings of the pipeline.
> Maybe a good discusion about this item can help in develop a good and
> robust architechture.
> 
> 
> -- 
> Best regards,
>  Sebastian                            mailto:sebasf@lncc.br
> ---
> Visit SkullyDoo OpenSource project (3D image segmentation and 
> visualization software) http://www.skullydoo.com.ar
>