Brief description: “With LHC@home you can actively contribute to the computing capacity of the Laboratory !” claimed a recent article in the CERN bulletin (http://cds.cern.ch/record/2151943?ln=en). Indeed, the volunteer computing program which was setup for CERN 50th anniversary gradually developed to allow volunteers to help in all the fields of activities : LHC beam simulation, theory, experiments. The history and basic ingredients of the LHC@home project will be developed, and students will be given all the tips needed to become a “CERN ambassador”, stay in touch and even contribute remotely when they will be back home. Speaker's short bio: Claire Adam-Bourdarios is a French physicist working on the ATLAS experiment. She is the current co-ordinator of the Outreach group, which oversees communication, educational programs and various partnerships with artists, scientists from other disciplines and - recently - citizen sciences experts.
The Hadoop ecosystem is the leading opensource platform for distributed storing and processing big data. It is a very popular system for implementing data warehouses and data lakes. Spark has also emerged to be one of the leading engines for data analytics. The Hadoop platform is available at CERN as a central service provided by the IT department. By attending the session, a participant will acquire knowledge of the essential concepts need to benefit from the parallel data processing offered by Spark framework. The session is structures around practical examples and tutorials. Main topics: Architecture overview - work distribution, concepts of a worker and a driver Computing concepts of transformations and actions Data processing APIs - RDD, DataFrame, and SparkSQL
Speaker's short bio: Mathieu Gravey is currently a PhD. student at the Institute of Earth Surface Dynamics, University of Lausanne, Switzerland. He has a computer science background from the Ecole des Mines d'Alès, France. In November 2015, he won the Intel® Modern Code Developer Challenge. During his CERN openlab internship, he is taking part on the implementation of a MPI-based framework to steer NUMA-aware GeantV workload for many-core systems.
Brief description: This lecture will review key aspects of CERN’s computing from 1974 - 2014; four decades that correspond to the speaker’s employment period. Speaker's short bio: Sverre Jarp worked in the IT Department at CERN for over 40 years and held multiple managerial and technical positions promoting advanced but cost-effective, large-scale computing and data management solutions for the Laboratory. Today, as honorary staff at CERN, he retains an unabated interest in several areas, in particular the domain of Big Data combined with High Throughput Computing as well as application scalability based on vector- and parallel-programming. S. Jarp holds a degree in Theoretical Physics from the Norwegian University of Science and Technology (NTNU) in Trondheim.
Brief description: In this talk we will discuss the key internals of modern computing hardware, and their evolution over time. Many recent improvements mean more opportunities but also more work for programmers. Can the promising technologies of the future – from accelerators to highly specialized chips – make life easier again? Speaker's short bio: Andrzej Nowak spent the last decade working on new digital technology at CERN and at Intel. At CERN, Andrzej managed a research lab collaborating with Intel and was part of the CERN openlab Chief Technology Office. There he initiated and ran projects with the private sector (e.g. HP, Google), as well as international research institutes, such as EPFL. Today, Andrzej heads TIK Services, a technology and innovation consultancy, and runs a peer-to-peer finance start-up.
Machine learning has become a hot-topic in the recent years. Together with the rise of distributed computing and GPU's, models can be computed and evaluated faster. In this tutorial the participant will learn how distributed computing frameworks like Apache Spark benefit the analysis of a big data set. We will guide the participant through the complete analysis pipeline using Spark's MLlib (Spark's built-in Machine Learning library); starting with data preparation and feature selection, and ending with model evaluation techniques such as cross-validation.
Welcome to the HNSciCloud Tender Information Day webpage. The event will be live streamed, the webcast will be accessible through this webpage. If you want to know more about HNSciCloud, the PCP methodology and legal contractual aspects, the system architecture & technical requirements, check out the recordings of the presentations made during the HNSciCloud Open Market Consultation (17 March 2016): https://indico.cern.ch/event/507164/timetable/
IT Lightning Talks (ITLT) are short presentations on any topic related to computing technology or to the IT department. See more here: https://twiki.cern.ch/IT/LightningTalks/