Re: Ptolemy Classic [message #366 is a reply to message #306] |
Sat, 24 April 2004 11:56 |
Krzysztof Korcyl
Messages: 7 Registered: April 2004
|
occasional visitor |
From: pb-d-128-141-43-58.cern.ch
|
|
Dear Walter,
I have been modeling computer networks (Ethernet) for more than 6 years. The work is part of the ATLAS (LHC experiment) TDAQ project.
At the very beginning, when I was associate to CERN I was using OPNET - the commercial tool (very powerful but terribly expensive) which was based (and perhaps still is) on C++.
When more Institutes joined the modeling activity and it became clear that the OPNET is too expensive for most of them (including CERN) we decided to find another, cheaper tool - and our choice was Ptolemy. At that time they were at the end of support for the Classic version but we had at least two reasons for staying with C++ and not going for JAVA:
- we could use models from OPNET almost directly in Ptolemy C++
- we found that the JAVA version of Ptolemy was very erratic (at that time) and the overhead added by JAVA was unnecessary. From the very beginning we were using scripts and writing our own code to collect statistics. Later we were using ROOT to plots results.
This is history, but till now we are using (recompile) the old version of the Ptolemy - and it works!
I think that binding to C++ is important and we should stay with it.
I also learned that some of our collaborators are experts in SystemC - which I believe is also based on C++. I hope that at certain moment we could look into possible merge of the Ptolemy C++ code within the SystemC framework.
I hope this explains your question.
cheers,
Krzysztof.
|
|
|