Re: [myrinet] Myrinet and Corba

Hi Alan,
First off, I think that I should mention that CORBA developers
will probably react negatively if you suggest CORBA is inherently
slow. An implementation may be slow, and a program using the
CORBA API may be slow, but that does not means CORBA is slow.

One of the most promising ORBs around is TAO:
	http://www.cs.wustl.edu/~schmidt/TAO.html

TAO has a feature called 'pluggable protocols' which allows an
end user to write an adaption layer for their own communication
hardware, or it can use standard techniques like TCP/IP.

Lack of time has regrettably delayed me from experimenting with
this, but if you ask on the TAO mailing list you may find other
people with similar interests.

Regards,
Andrew

-------------
Alan Scheinine wrote:
> 


At an ACM meeting I heard a conversation that someone had ported
ORBacus to Myrinet.  But I did not take note of who did it.
It was said that the conclusion of the person who did the work
was that CORBA was intrinsically slow.  Now, there is Corba-Express
by the group that developed ORBacus so it might be useful to
reconsider implementing a Myrinet communication layer.  Does anyone
know who did is doing or has done work with Myrinet and Corba?
Thanks,
Alan Scheinine  Email: [email protected]

--
Andrew Robbie           Air Operations Simulation Centre
+61 3 9626 7547         Defence Science & Technology Organisation
PGP Fingerprint: 3EF8 0F78 4E76 FDDE E59B  7236 A8E2 EF31 FCD9 6F73