Online data compression in the ALICE O\(^2\) facility

• Main Authors: Richter, Matthias Rudolph, Aamodt, Kenneth, Arsene, Ionut Christian, Bravina, Larissa, Dordic, Olja, Eyyubova, Gyulnara, Hille, Per Thomas, Kolevatov, Rodion, Kværnø, Henning, Lindal, Svein, Løvhøiden, Gunnar, Milosevic, Jovan, Nilsson, Mads Stormo, Nyiri, Agnes, Skaali, Toralf Bernhard, Tveter, Trine Spedstad, Tywoniuk, Konrad, Wikne, Jon Christopher, Zabrodin, Evgeny, Alme, Johan, Bablok, Sebastian Robert, Djuvsland, Øystein, Fehlker, Dominik, Haaland, Øystein Senneset, Huang, Meidana, Kanaki, Kalliopi, Klovning, Arne, Larsen, Dag Toppe, Liu, Lijiao, Nystrand, Joakim, Øvrebekk, Gaute, Pommeresche, Bjørn-Erling, Skjerdal, Kyrre, Ullaland, Kjetil, Wagner, Boris, Helstrup, Håvard, Hetland, Kristin Fanebust, Kileng, Bjarte, Røed, Ketil, Pocheptsov, Timur Anatolievich, Altinpinar, Sedat, Røhrich, Dieter, Adamová, Dagmar, Aggarwal, Madan M, Aglieri Rinella, Gianluca, Ahammed, Zubayer, Ahn, Sang Un, Akindinov, Alexander, Aleksandrov, Dimitry, Alessandro, Bruno, ALICE, Collaboration
• Format: Article
• Language: English

The ALICE Collaboration and the ALICE O2 project have carried out detailed studies for a new online computing facility planned to be deployed for Run 3 of the Large Hadron Collider (LHC) at CERN. Some of the main aspects of the data handling concept are partial reconstruction of raw data organized in so called time frames, and based on that information reduction of the data rate without significant loss in the physics information. A production solution for data compression has been in operation for the ALICE Time Projection Chamber (TPC) in the ALICE High Level Trigger online system since 2011. The solution is based on reconstruction of space points from raw data. These so called clusters are the input for reconstruction of particle trajectories. Clusters are stored instead of raw data after a transformation of required parameters into an optimized format and subsequent lossless data compression techniques. With this approach, a reduction of 4.4 has been achieved on average. For Run 3, not only a significantly higher reduction is required but also improvements in the implementation of the actual algorithms. The innermost operations of the processing loop effectively need to be called up to O \(10^{11}\) /s to cope with the data rate. This can only be achieved in a parallel scheme and makes these operations candidates for optimization. The potential of template programming and static dispatch in a polymorphic implementation has been studied as an alternative to the commonly used dynamic dispatch at runtime. In this contribution we report on the development of a specific programming technique to efficiently combine compile time and runtime domains and present results for the speedup of the algorithm.