Corba is between a factor 5 to 20 slower than ROOT's IPC system.
This is mainly due to the large overhead the ORB has to find out where
what object resides. The communication is hardly ever direct between
two processes, but goes most of the time via the ORB. Further, CORBA only
supports calling objects by reference. That is, you get a pointer to an
object and every time you access the object the call goes over the net
and the operation is executed remotely. ROOT supports object passing
by value , i.e. complete object is transfered from one process to the
other. Once transfered all access to the object are, of course, local.
This is especially interesting when sending objects like histograms
between processes. The ROOT IPC resembles closely the way Java is
doing things with RMI (basically same functionality).
Cheers, Fons.
>
> Dear all,
>
> my problem is not really a root problem,.. but maybe you are willing to
> share your knowledge with me,..
>
> from the root course I attended I somehow remember, that root doesnt use
> corba for netcomunication. I guess the reason will be performance (but I
> am not sure). In this case my question is: what makes corba that
> inefficient? and how does root overcome this problem?
>
> A brief answer with just pointing me out, where the hotspots are would be
> great already.
>
> ThanX in advance!
> Selim
>
>
> S E L I M I S S E V E R
> DESY-F15, Notkestr. 85, 22603 Hamburg, Germany; Tel/Fax: 040 8998-2843/4033
> http://www.physik.uni-dortmund.de/~issevers; selim.issever@desy.de
> Ete kemige burundum, Yunus diye gorundum. Yunus Emre
>
>
-- Org: CERN, European Laboratory for Particle Physics. Mail: 1211 Geneve 23, Switzerland Phone: +41 22 7679248 E-Mail: Fons.Rademakers@cern.ch Fax: +41 22 7677910