From the Big Bang to Massive Data Flow: Parallel Computing in High Energy Physics Experiments

Abstract

Tracking detectors in high-energy physics experiments produce hundreds of megabytes of data at a rate of several hundred Hz. Processing this data at a bandwidth of 10-20 Gbyte/sec requires parallel computing. Reducing the huge data rate to a manageable amount by realtime data compression and pattern recognition techniques is the prime task. Clustered SMP (Symmetric Multi-Processor) nodes, based on off-the-shelf PCs and connected by a high bandwidth, low latency network, provide the necessary computing power. Such a system can easily be interfaced to the front-end electronics of the detectors via the internal PCI-bus. Data compression techniques like vector quantization and data modeling and fast transformations like conformal mapping or the adaptive, generalized Hough-transform for feature extraction are the methods of choice.KeywordsLarge Hadron ColliderData CompressionTracking DetectorTime Projection ChamberTrack SegmentThese keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.