Simulation of the ALICE DAQ System with GALSIM
Paper: 297
Session: B (talk)
Speaker: Jianwei, Yuan, University of Oslo
Keywords: data acquisition systems, simulation, simulation tools, event building, graphics
Jianwei Yuan, Bin Wu, Bernhard Skaali
Department of Physics
University of Oslo,
P.O.Box 1048 Blindern,
0316 Oslo, Norway
Abstract
We present some of the simulation results for the DAQ system of the ALICE
experiment at the Large Hadron Collider (LHC) at CERN.
The simulation is performed by the ALICE DAQ Simulation Program GALSIM.
GALSIM comprises two software packages; 1)the basic ALICE DAQ simulation
program ALSIM and 2) a graphic user interface which dynamically displays
the data transport and synchronization messages in the DAQ system. GALSIM
is written in MODSIM II and has been developed by the ALICE DAQ group at
the Department of Physics, University of Oslo, Norway.
In the paper, we will first introduce the basic ALICE DAQ system and the
simulator GALSIM. We will then concentrate on the simulation of the ALICE
DAQ topology with various parameters. The results we present can be viewed
as an extension of the DAQ simulation shown in the ALICE Technical
Proposal.
Among the important results, we emphasize on the following issues,
1. The effect of message latency. Messages are used by the Event
Destination Manager to signal to a number of Local Data Concentrators
(LDCs) to send sub-events to Global Data Collectors via switch. Messages
with long latency will tremendously slow down the system. The simulation
results under different experimental latencies are illustrated.
2. Balancing the traffic in the Event Building system. For the event
building in a large DAQ system, all the data that belong to the same event
will have to be sent to the same data processor from LDCs. Tokens can be
used to coordinate the data transmission among the LDCs in the system.
However, tokens may influence the system balance in term of buffer usage
and traffic pattern. By GALSIM, the simulator results affected by the
tokens can be monitored and analyzed.
3, The size of the DAQ system versus link speed. The cost of a DAQ system
will depend not only on the price of each device but on the size of the
whole system. However, due to the complexity of DAQ systems, one needs to
investigate the trade-offs between link bandwidth/delay and the number of
data links. We will in our paper explore this issue.