[omniORB] Flick CORBA IDL compiler

[Sai-Lai Lo]

>>>>> Richard Heycock writes:

> Has anyone been looking at feasibility of using the flick idl compiler
> from Utah University (http://www.cs.utah.edu/projects/flux/flick/)?

Don't know. Might be interesting to look into.

> It optimises the marshalling process so that transfers are between
> 2 and 17 times faster. There home page clearly states that it is NOT
> a drop in replacement for other idl compilers but I wondered if any
> of the ideas could be implemented in omniORBs idl compiler. Or if
> work could be done with Utah to aid the process of making flick
> work with omniORB.

> Comments?

=46rom our measurements, the time to marshal data arguments is only a min=
or
contributor to the overhead. The bulk of the time is spent somewhere else=
=2E
Usually, the time spend in the kernel contributes to half of the null ech=
o
time. If we get rid of the protocol stack in the kernel, like we have don=
e =

when running omniORB2 over SCI between two pentium pro linux, we are
getting 152 micro second null echo. I estimates that around a 100 is used=
 in
the runtime, the rest is the latency of the hardware. At this performance=

level, any saving in data marshalling time becomes significant. With good=

old TCP/IP, the gain in data marhsalling may not be noticeably.

Sai-Lai


-- =

Dr. Sai-Lai Lo                          |       Research Scientist
                                        |
E-mail:         S.Lo@orl.co.uk          |       Olivetti & Oracle Researc=
h Lab
                                        |       24a Trumpington Street
Tel:            +44 223 343000          |       Cambridge CB2 1QA
Fax:            +44 223 313542          |       ENGLAND