Speaker
Dr
Timm Morten Steinbeck
(FIAS - Institute for Computer Science-Johann-Wolfgang-Goethe Uni)
Description
ALICE (A Large Ion Collider Experiment) is a dedicated heavy ion
experiment at the Large Hadron Collider (LHC) of CERN. In order to
reduce the amount of data written to mass storage a High Level Trigger
(HLT) completely analyses every event triggered by the preceding trigger
stages to be able to select only the most promising events for
storage. In addition to this event selection the HLT is also used to
provide reconstructed events for a live online display in the main
ALICE control room as well as online data compression. The HLT is a
Linux cluster consisting of more than 200 nodes with 8-24 CPU cores
and 16 to 48 GB RAM each. In addition to these processing capabilities
the HLT can also make use of FPGA co-processing in its front-end
receiver cards, even before the data from the detector enters a
computer's memory and 35 nVidia graphics cards which can perform
online tracking. These capabilities are in particular important for
the large amounts of data encountered during the heavy ion collisions,
e.g. like in the first heavy ion period at the end of 2010. All HLT
software reconstruction is embedded into ALICE's offline data analysis
framework for testing and verification. Some parts of the HLT
processing software even directly use offline software for online
analysis. Other parts are highly optimized for most efficient online
running like the TPC tracking which shows speed improvements compared
to its offline equivalent of a factor of more than 3 (pp) and 8 (heavy
ion) respectively. We will present experiences from the previous
running since the first LHC collisions, in particular of the first
heavy ion collisions, as well as the current configuration and setup of
the HLT for the 2011 running period.
Author
Dr
Timm Morten Steinbeck
(FIAS - Institute for Computer Science-Johann-Wolfgang-Goethe Uni)
