بحث بعنوان The Second Level Trigger of the ATLAS Experiment at CERN’s LHC

بحث بعنوان
The Second Level Trigger of the ATLAS Experiment at CERN’s LHC
A. dos Anjos, M. Abolins, S. Armstrong, J. T. Baines, M. Barisonzi, H. P. Beck, C. P. Bee, M. Beretta, M. Biglietti,
R. Blair, A. Bogaerts, V. Boisvert, M. Bosman, H. Boterenbrood, D. Botterill, S. Brandt, B. Caron, P. Casado,
G. Cataldi, D. Cavalli, M. Cervetto, M. Ciobotaru, G. Comune, A. Corso-Radu, E. Palencia Cortezon, R. Cranfield,
G. Crone, J. Dawson, B. Di Girolamo, A. Di Mattia, M. Diaz Gomez, R. Dobinson, J. Drohan, N. Ellis, M. Elsing,
B. Epp, Y. Ermoline, F. Etienne, S. Falciano, A. Farilla, M. L. Ferrer, D. Francis, S. Gadomski, S. Gameiro,
S. George, V. Ghete, P. Golonka, S. González, B. Gorini, B. Green, M. Grothe, M. Gruwe, S. Haas, C. Haeberli,
Y. Hasegawa, R. Hauser, C. Hinkelbein, R. Hughes-Jones, P. Jansweijer, M. Joos, A. Kaczmarska, K. Karr,
A. Khomich, G. Kieft, E. Knezo, N. Konstantinidis, K. Korcyl, W. Krasny, A. Kugel, A. Lankford, G. Lehmann,
M. LeVine, W. Li, W. Liu, A. Lowe, L. Luminari, T. Maeno, M. Losada Maia, L. Mapelli, B. Martin, R. McLaren,
C. Meessen, C. Meirosu, A. G. Mello, G. Merino, A. Misiejuk, R. Mommsen, P. Morettini, G. Mornacchi, E. Moyse,
M. Müller, Y. Nagasaka, A. Nairz, K. Nakayoshi, A. Negri, N. Nikitin, A. Nisati, C. Padilla, I. Papadopoulos,
F. Parodi, V. Perez-Reale, J. Petersen, J. L. Pinfold, P. Pinto, G. Polesello, B. Pope, D. Prigent, Z. Qian, S. Resconi,
S. Rosati, D. A. Scannicchio, C. Schiavi, J. Schlereth, T. Schoerner-Sadenius, E. Segura, J. M. Seixas, T. Shears,
M. Shimojima, S. Sivoklokov, M. Smizanska, R. Soluk, R. Spiwoks, S. Stancu, C. Stanescu, J. Strong, S. Tapprogge,
L. Tremblet, F. Touchard, V. Vercesi, V. Vermeulen, A. Watson, T. Wengler, P. Werner, S. Wheeler, F. J. Wickens,
W. Wiedenmann, M. Wielers, Y. Yasu, M. Yu, H. Zobernig, and M. Zurek
Abstract—The ATLAS trigger reduces the rate of interesting
events to be recorded for off-line analysis in three successive levels
from 40 MHz to 100 kHz, 2 kHz and 200 Hz. The
high level triggers and data acquisition system are designed to
profit from commodity computing and networking components to
achieve the required performance. In this paper, we discuss data
flow aspects of the design of the second level trigger (LVL2) and
present results of performance measurements.
Index Terms—ATLAS, cluster, data acquisition, high-energy
physics, network, object oriented, performance, testbed, trigger.
IV. CONCLUSION
We have presented a LVL2 system that uses a farm of conventional PCs running ESS guided by results from the LVL1
trigger. Standard Gigabit Ethernet is used to connect the individual processors to the readout system. All measurements were
made using a connection-less protocol, UDP, or raw Ethernet.
Although not explicitly shown in the measurements, the performance differences were minimal.
It has been shown that the I/O capacity of each node is largely
sufficient. The CPU resources devoted to RoI data collection
use a modest fraction of the allocated time budget for a large
range of RoI sizes. Though concentration of multiple ROBs in
one ROS (resulting in the transfer of larger data slices if the
ROB to detector mapping is optimized) is more efficient, even
an architecture with a concentration factor of one would still
give acceptable results.
Event parallelism using multiple threads is effective, allowing the use of multi CPU nodes to save on cost, cooling and
floor space. Additional threads are necessary to compensate
the readout latency.
The scalability has been measured for a system with a small
number of nodes. It has been shown that, with unrealistically
high data rates into each processor and a small number of ROSs,
LVL2 rates as high as 70 kHz can be obtained with a relatively
modest fall of scaling.
كلمة سر فك الضغط : books-world.net
The Unzip Password : books-world.net