21-25 May 2012
New York City, NY, USA
US/Eastern timezone

The ALICE DAQ Detector Algorithms framework

May 24, 2012, 1:30 PM
4h 45m
Rosenthal Pavilion (10th floor) (Kimmel Center)

Rosenthal Pavilion (10th floor)

Kimmel Center

Poster Online Computing (track 1) Poster Session


Sylvain Chapeland (CERN)


ALICE (A Large Ion Collider Experiment) is the heavy-ion detector studying the physics of strongly interacting matter and the quark-gluon plasma at the CERN LHC (Large Hadron Collider). The 18 ALICE sub-detectors are regularly calibrated in order to achieve most accurate physics measurements. Some of these procedures are done online in the DAQ (Data Acquisition System) so that calibration results can be directly used for detector electronics configuration before physics data taking, at run time for online event monitoring, and offline for data analysis. A framework was designed to collect statistics and compute calibration parameters, and has been used in production since 2008. This paper focuses on the recent features developed to benefit from the multi-cores architecture of CPUs, and to optimize the processing power available for the calibration tasks. It involves some C++ base classes to effectively implement detector specific code, with independent processing of events in parallel threads and aggregation of partial results. We present benchmarks showing the performance improvements, and some results of investigations conducted with CUDA and GPUs to push the speed-up further. The Detector Algorithm (DA) framework provides utility interfaces for handling of input and output (configuration, monitored physics data, results, logging), and self-documentation of the produced executable. New algorithms are created quickly by inheritance of base functionality and implementation of few ad-hoc virtual members, while the framework features are kept expandable thanks to the isolation of the detector calibration code. The DA control system also handles unexpected processes behavior, logs execution status, and collects performance statistics.

Primary author


Adriana Telesca (CERN) Alexandru Grigore (Polytechnic University of Bucharest (RO)) Mr Barthelemy von Haller (CERN) Mr Bartolomeu Andre Rodrigues Fernandes Rabacal (Instituto Superior Tecnico (IST)) Csaba Soos (CERN) Ervin Denes (Hungarian Academy of Sciences (HU)) Filippo Costa (CERN) Mr Franco Carena (CERN) Giuseppe Simonetti (Universita e INFN (IT)) Mr Pierre Vande Vyvre (CERN) Roberto Divia (CERN) Ulrich Fuchs (CERN) Mr Vasco Chibante Barroso (CERN) Wisla Carena (CERN)

Presentation materials