Using of GPUs to enhance CMS event detection: Current challenges and future prospective

摘要

The Compact Muon Solenoid (CMS) is a physics experiment held at the European Organization for Nuclear Research (CERN). CMS main role is to detect the events generated by colliding particles in the Large Hadron Collider (LHC). There are around 40 million collisions per second in the LHC, each collision generates billions of events. From each collision, CMS hardware and software events triggers are responsible of selecting the most interesting events with 1 kHz acceptance rate. Future upgrades to the detector will increase the number of selected events by at least seven times. This poses a major computing challenge. This paper discusses this challenge by exploring the current CMS data acquisition and triggering system and future upgrades. Its objective is to locate an opportunity to improve the current system. After going through some of the literature that covers the topic, the paper explores the use of General Purpose Graphics Processing Units (GPGPUS) and Parallel Computing techniques as an opportunity to overcome this challenge.