Previous education and training events

Primary tabs

    The course will take place in Barcelona Supercomputing Center, within the UPC Campus Nord premises. Vertex Building VS208 :
    13 Feb 2019 - 09:00 to 15 Feb 2019 - 18:00
    The objetive of this course is to give a panorama on the use of hpc-based computational mechanics in Engineering and Environment through the projects BSC are carrying on. This panorama includes the basics of what is behind the main tools: computational mechanics and parallelization. The training is delivered in collaboration with the center of excellence CompBioMed. A number of fixed amount travel grants will be offered to a selected number of participants for covering part of their travelling expenses.   To learn about how to apply to a CompBioMed travel grant please go to Further information.  
    Barcelona - UPC/BSC Campus Building A6. Classroom A6204 (second floor of Building A6, Aula A6204). :
    12 Feb 2019 - 10:00 to 13 Feb 2019 - 18:00
    The tutorial will cover the main aspects of the Vivado HLS optimizations, and the OmpSs@FPGA approach. There will be a number of lab sessions based on HLS Xilinx samples for coding and optimization:        Matrix multiplication        RGB – YUV filter        Discrete Cosine Transform (DCT) The tutorial will cover the main aspects of the Vivado HLS optimizations, and the OmpSs@FPGA approach.  Requirements and other informations:  A USB Hard disk will be used with all the toolchain needed to perform the  hands on. Please, test that your computer will be able to boot from a bootable  Hard Disk.   Zedboards will be used in the tutorial...
    Sala d'actes de la FiB (Campus Nord) :
    5 Feb 2019 - 11:00
    Abstract: At least 70 human RNA-binding proteins contain a prion-like domain (PrLD). PrLDs are low complexity domains which resemble in composition the infectious yeast prions. Mutations in PrLDs are associated to the onset of many neurodegenerative conditions, such as Amyotrophic Lateral Sclerosis (ALS). PrLDs are able to populate multiple physical states: diffuse, liquid de-mixed, insoluble amyloid. Pathological mutations affect these equilibria in ways we cannot yet fully understand, or predict. The TAR DNA binding protein TDP-43 contains a 140 aa long PrLD and forms cytoplasmic aggregates in most cases of ALS. We use Deep Mutational Scanning to understand how sequence determines the toxicity of TDP-43 in a yeast model. I will present the first "genotype-to-phenotype" map of...
    Benedetta Bolognesi, Junior Group Leader, IBEC
    The course will take place in Barcelona Supercomputing Center, within the UPC Campus Nord premises. Vertex Building VS208 :
    5 Feb 2019 - 09:30 to 8 Feb 2019 - 13:00
    The course brings together key information technologies used in manipulating, storing, and analysing data including: the basic tools for statistical analysis, techniques for parallel processing, tools for access to unstructured data, storage solutions.
    Sala d´actes de la FIB, Campus Nord. :
    31 Jan 2019 - 11:00
    To access the presentation, please click here Abstract: Decomposition is a basic approach to speed-up solving sparse systems on parallel-architectures. Though classic task of a matrix (graph) decomposition is known to be hard. A new kind of decomposition into clans is offered which complexity is linear in the number of nonzero elements. The technique is applicable for solving sparse systems over arbitrary rings (fields) with sign. Analysis of collections of matrices for real-life application shows that lots of matrices are decomposable into clans. A minimal clan size of the decomposition restricts the granulation of the technique. The system decomposition into clans is represented by a weighted graph. The task of constructing a sequence of systems of lower dimension is called an...
    Dmitry A. Zaitsev, Professor of Computer Science at Vistula University, Warsaw.
      The course will take place in Barcelona Supercomputing Center, within the UPC Campus Nord premises. Vertex Bulding, Room VS208     :
    29 Jan 2019 - 09:30 to 30 Jan 2019 - 16:30
    COMPSs is a programming model which is able to exploit the inherent concurrency of sequential applications and execute them in a transparent manner to the application developer in distributed computing platform. This is achieved by annotating part of the codes as tasks, and building at execution a task-dependence graph based on the actual data used consumed/produced by the tasks. The COMPSs runtime is able to schedule the tasks in the computing nodes and take into account facts like data locality and the different nature of the computing nodes in case of heterogeneous platforms. Additionally, recently COMPSs has been enhanced with the possibility of coordinating Web Services as part of the applications. COMPSs supports Java, C/C++ and Python as programming languages.
    Sala d'actes de la FiB (Campus Nord) :
    24 Jan 2019 - 11:00
    To see the presentation please click here ABSTRACT: One of the most fundamental design challenges in any interconnection network is that of routing deadlocks. A deadlock is a cyclic dependence between buffers that renders forward progress impossible. Deadlocks are a necessary evil and almost every on-chip/HPC network today avoids it either via routing restrictions across physical channels (Dally's Theory) or with at least one escape virtual channel (Duato's Theory). This ensures that a cyclic dependence between buffers is never created in the first place. Moreover, each solution is tied to a specific topology, requiring an updated policy if the topology were to change. Alternately, solutions have also been proposed to reserve certain resources (buffers) and allocate them...
    Paul V. Gratz is an Associate Professor in the department of Electrical and Computer Engineering at Texas A&M University, currently visiting the University of Edinburgh on sabbatical.
    Campus Nord, C6 Building, Room E-106   :
    21 Jan 2019 - 09:00 to 25 Jan 2019 - 18:00
    Objectives: VI-HPS Tuning Workshops are the major training vehicle where around 40 participants receive instruction and guidance to become familiar with VI-HPS programming tools suite. During the training, we will explain the functionality of each tool and how they can be used. We will also give them hands-on assistance on how to apply VI-HPS tools to their own parallel application codes, along with advice for potential corrections and optimizations. Additionally, feedback to tools developers helps direct tools development to user needs, as well as improve tool documentation and ease of use. For further information please check the VI-HPS website
    Sala d'actes de la FiB (Campus Nord) :
    15 Jan 2019 - 11:00
    To download the presentation please click here Abstract: Reuse predictors predict whether a block will be used again before it is evicted from the cache. We propose to use several features associated to an access as input to our predictors, the goal is to correlate past behavior with future accesses. The higher accuracy obtained by our predictors yield superior overall performance. This talk will present two techniques that apply perceptron learning to cache management. For our Perceptron Predictor we propose using perceptrons to predict cache line reuse. Perceptron trains weights selected by hashes of multiple features, including the PC of the memory access instruction, other recent PCs, and two different shifts of the tag of the referenced block. The Multiperspective Reuse...
    Elvira Teran, Assistant Professor in the School of Engineering at Texas A&M International University.
    Sala de juntes, B6 building :
    21 Dec 2018 - 11:00
    Abstract Steadily growing amounts of sensor data is being generated and is typically processed by Cloud based servers leading to highly unbalanced systems with most of the computations being performed at the back-end. Such large-scale systems are characterized by rocketing energy budgets of the back-end, bandwidth hungry interconnect infrastructures and many end-users exposed to disturbing application response delays. We will discuss how data-movement centric, massive-scale smart systems can be integrally developed, built and orchestrated. We show that adding customizable compute capabilities at the edge will enable truly application aware systems able to perform all of the required data processing and movements within affordable energy budgets at all levels. Short bio Georgi...
    Georgi Gaydadjiev, Director of Maxeler IoT-Labs BV in the Netherlands and Honorary Visiting Professor at Imperial College in London, UK.