Minutes of Online Meeting
7th October, 1998
Present:
Jean-Pierre Duffey, Phlippe Gras, John Harvey, Beat Jost, Pere
Mato, Hans Muller, Tatsuya Nakada, Wolfgang Tejessy, Jose Toledo, Markus Tujula, Mike
Koratzinos
Agenda:
- Report on ATM Workshop
J.-P. Dufey
- Status of Readout Unit Project J. Toledo
- Simulation of Myrinet
M. Tujula
- OPC Project
P. Gras
- AOB
1) Report on ATM Workshop (slides)
Jean-Pierre Dufey gave a report on the Esone Workshop on
Applications of ATM in High Energy Physics which was held in Paris on the 21st
and 22nd of September. After an introduction of ATM in general by the Forum
representative, the main domains of interest for use of ATMs was Lans and Wans, fast
controls (for accelerators) and DAQ. Some 50 people attended with about 25 presentations
given.
For LANS the use of ATM is shrinking to backbones and fast real-time applications with
predictions being that in 5 years time we will see 20% ATM and 80% Ethernet (of which 20%
will be Gigabit). For Desktop there is no need for, and therefore little interest in, ATM.
Concerning protocols, "IP is leading the dance".
Bob Dobinson reported on the Gigabit Ethernet project in CERN for the ATLAS level 2
trigger. Comparisons were given between Gigabit Ethernet and Fast Ethernet and a summary
of results shown.
HEP experiments may be an interesting customer for switch providers. A comparison shows
that the French Telecom will need 200 Gigabits/s, which is in the same range as, e.g.,
Atlas needs.
2) Status of
Readout Project (slides)
A report on the status of the LHCb read-out unit was given by
Jose Toledo. After a short overview, the discussion centred on the limitations at the
input side and the role of the dual port DPM memories. The bottleneck here is the memory
bus leading to the FPGA sub-event builder. A 9unit VME test module has will be built to
allow intensive testing. On the output side the design questions are strongly related to
the final network choice. More details are given in draft 5.10.98 of the paper LHCb
Readout Unit.
3) Simulation of Myrinet
First results of simulation studies for this network were presented by Markus
Tujula. Using Foresight, a 4 x 4 composite Myrinet switch was simulated.
Source-to-destination routing was compared with packet routing under various conditions
(straight versus retarded, length variations,
) and the results presented in tabular
form. Hardware testing will start next week.
4) OPC Project (slides)
The start-up of a project to evaluate the potentialities of OLE
for Process Control, OPC, in the controls area was introduced by Philippe Gras. OPC looks
like an interesting technology specially for commercial hardware because most
manufacturers will provide OPC servers with their modules.
4) AOB
There was no other business.
Minutes Prepared by Wolfgang Tejessy