July 24th to August 1st, 2017
Costa Rica National High Technology Center
Dr. Frankiln Chang Díaz building
1.3 km North of the US Embassy
Pavas, San José, Costa Rica
- Shared-memory Programming
- Distributed-memory Programming
- Performance Analysis
The National Advanced Computing Colaboratory (CNCA), the Research Laboratory in Pattern Recognition and Intelligent Systems (PRIS-lab), the Pattern Recognition Group and Automatic Learning (PARMA Group) and CONARE Network organize the Summer School in Computing High Performance (HPC).
The main objective of this event is to train students, researchers and professors in central topics of HPC:
Block 1: Introduction to Programming
Linux programming tools.
Programming language C.
Block 2: High Performance Computing
Shared memory programming (OpenMP).
Accelerator Programming (OpenACC).
Programming in distributed memory (MPI).
Block 3: Advanced Parallel Programming
Registration for each of the blocks is independent. A certificate of participation will be delivered for each block.
The official languages of the Veranillo School at HPC are:
All materials from HPC School are available in the repository https://github.com/CNCA-CeNAT/EV-HPC-2017
Block 1: Introduction to Programming
The Linux Tools workshop will be taught by Ing. Andrés Mora, a PRIS-Lab researcher. The following topics will be addressed:
- File System.
- Environment variables.
- Text Utilities
- Bash principles.
- Network basics: ssh.
Biography: Andrés Mora completed his high school studies at the School of Electrical Engineering of the University of Costa Rica. He is currently carrying out his postgraduate studies in the Academic Master’s Degree in Electrical Engineering through the Graduate Studies System of the University of Costa Rica and working as a teacher in the school where he studies. He is a researcher and active member of the Laboratory of Pattern Recognition and Intelligent Systems (PRIS-Lab), as well as of the Center for Research in Human Movement (CIMOHU). His research delves into bioinformatics, especially in the application of pattern recognition to images of digital microscopes.
Programming with C
The C Programming workshop will be taught by the M.Sc. Ricardo Román, PRIS-Lab researcher. The following topics will be addressed:
- Introduction to programming, problem, algorithm, instance, program
- Constants, variables and data types
- Control structures
- Operators and functions
- Arrangements and pointers
- Memory management
- File management
- Execution threads
- External libraries
Biography: Ricardo Román-Brenes has a bachelor’s degree in Computer Science and Computer Science from the University of Costa Rica and a Master’s Degree in Computer Engineering with an emphasis in Computer Science from the Technological Institute of Costa Rica. He is currently studying for the Academic Doctorate in Engineering of the University of Costa Rica and his graduation work focuses on the application of computational methods for the prediction of chemosensitivity in cancer. Among his interests are operating systems, high performance computing parallel and distributed computer science, visualization and teaching.
Block 2: High Performance Computing
Shared Memory Programming (OpenMP)
The Shared Memory Programming (OpenMP) workshop will be taught by Dr. Esteban Meneses, director of the CNCA. The following topics will be addressed:
- Principles of multi-threaded programming.
- Shared memory systems.
- Parallel programming models.
- Open Multi-processing programming standard (OpenMP).
- Parallel regions and types of variables.
- Critical sections and barriers.
- Parallel cycles and calendaring.
- Other parallel constructions.
Biography: Esteban Meneses is the director of the National Advanced Computing Collaboration (CNCA) of the National Center of High Technology (CeNAT). In addition, he is an associate professor in the Computer School of the Technological Institute of Costa Rica. His research interests revolve around high performance computing: scalable fault tolerance algorithms and scientific programming using parallel objects. Esteban graduated from the doctorate in computer science at the University of Illinois at Urbana-Champaign and worked as a researcher at the Simulation and Modeling Center at the University of Pittsburgh.
Accelerator Programming (OpenACC)
The Programming for Accelerators (OpenACC) workshop will be taught by the M.Sc. Jorge Castro, CNCA researcher. The following topics will be addressed:
- Basic concepts about accelerator programming.
- OpenACC programming standard.
- Code profiling with OpenACC toolkit.
- Directives to express parallelism.
- Directives to express data movement.
- Directives to express optimizations.
Biography: Jorge Castro has a bachelor’s degree in Computer Science and Computer Science from the University of Costa Rica and an Academic Master’s Degree in Electrical Engineering with an emphasis in Digital Systems from the University of Costa Rica. Among his interests are digital signal processing, bioacoustics, machine learning and high performance computing with OpenACC. He is currently working on accelerating with OpenACC the method of canceling ambient noise in recordings of vocalizations of manatees antillanos, developed in his master’s thesis.
Distributed Memory Programming (MPI)
The Distributed Memory Programming (MPI) workshop will be taught by Mr. Diego Jiménez, a CNCA researcher. The following topics will be addressed:
- Principles on parallel programming models, distributed memory systems, SPMD.
- The standard of Message Step programming, history, general description.
- Basic definitions: communicators, ranges, and messages.
- Point-to-point messages, including sending and receiving without blocking.
- Generalities of collective communication operations, using broadcast and reduction.
Biography: Diego Jiménez Vargas, a computer engineer graduated from the Technological Institute of Costa Rica with a bachelor’s degree, is part of the work team of the National Advanced Computing Collaboration, where he conducts research in high-performance scientific computing. Currently, he is involved in a project in the area of computational neuroscience where he contributes his knowledge in parallel programming for the construction of a computational cluster, based on reconfigurable hardware platforms, for the simulation of biologically significant neural networks. On the other hand, it conducts studies on the communication patterns present in large-scale MPI applications and their mapping to common interconnection topologies in high-performance computing. His research interests focus on parallel programming and high performance computing, as well as applied scientific computing in the area of neuroscience.
Block 3: Advanced Parallel Programming
Advanced Parallel Programming
The Advanced Parallel Programming workshop will be taught by the M.Sc. Ivan Girotto and M.Sc. Marlon Brenes The following topics will be addressed:
- Introduction to the ICTP and HPC related activities.
- Overview on modern computer architectures systems for HPC: the software crisis.
- Parallel thinking.
- Best practices of parallel programming for HPC.
- Exponential integrators using matrix functions: Krylov subspace methods and Chebyshev expansion – The HPC approach.
Biography: Ivan Girotto is appointed at the Abdus Salam, International Centre for Theoretical Physics as High-Performance Computing (HPC) Application Specialist. With more than ten years of experience Ivan Girotto has previously served, as computational scientist, at national supercomputing centres such as CINECA, and later at the Irish Centre for High-End Computing (ICHEC). Today at ICTP, Ivan is co-director and lecturer in number of international workshops and advanced schools as well as in other programs for education and training on parallel programming for scientific computing and HPC (i.e., the joint SISSA/ICTP Master program in HPC). His main activity at the ICTP remains to conduit and support the world-wide community to efficient research production on platforms for high-performance computing. Ivan Girotto is also one of the Principal Investigators of the Europen Centre of Excellence MaX – Materials design at the Exascale.
Biography: Marlon Brenes is a physicist and a chemical engineer from Universidad de Costa Rica. After graduating he moved to Trieste, Italy; in order to take the joint SISSA/ICTP Master in HPC program under the ICTP fellowship, where he graduated with honors. He is currently a visiting scientist at the Condensed Matter and Statistical Physics section of the ICTP and his research interests are out-of-equilibrium physics of complex quantum systems, quantum entanglement and thermodynamics; and HPC-related approaches towards the simulation of quantum dynamics. He’s starting his PhD in Physics at Trinity College Dublin in September 2017 supported by the Royal Society of London URF program.
Speaker: Mariano José Sánchez Bontempo
Summary of the talk: The use of advanced networks has been enhanced in the world, but … what are advanced networks? They are very high speed communication networks that allow higher education and research communities to develop new knowledge, strengthening working groups and collaboration around the world. These communication and collaboration networks also allow the deployment of many projects and initiatives that favor all of humanity.
Biography: Mariano José Sánchez Bontempo is a graduate of the Bachelor of Computer Engineering of the Technological of Costa Rica, a graduate of the Bachelor of Philosophy and Humanities of the Catholic University of Costa Rica and a student of the Bachelor of Information Technology for Business Management of the University Costa Rican Latina. Currently, he is the Scientific Coordinator of the National Research and Education Network of Costa Rica: RedCONARE. In addition, he serves as Executive Director of RedCONARE before the Latin American Cooperation of Advanced Networks (CLARA) and member of the fiscal commission of said organization for the period 2016-2018.
Estimation of the Performance of a Computational Platform (use case: HPC Platform)
Speakers: Jaime Rodriguez / Jaime Güell
Summary of the talk: The performance of a computing platform is the basis for the prediction and preparation of future improvements. The talk explains the fundamentals that Intel uses to determine performance in an optimal way on a platform such as an HPC cluster and provide foundations for future projections of its performance in different configurations, with different variations and different use cases .
Jaime Rodriguez Quesada
Biography: He graduated in Computer Engineering from the Technological Institute of Costa Rica in 2006, completed studies in professional masters in Computing in 2010 at the University of Costa Rica. He has 12 years of experience in Intel Components of Costa Rica where he worked performing test execution automation for marginal analysis of specifically processor circuits, and also developed software to facilitate the process of debugging the processor graphics component. He is currently working on the characterization and performance analysis of Intel processors and co-develops a solution in Hadoop to facilitate searches for test execution traces for architectural studies.
Jaime Güell Romero
Biography: Graduated from Tecnológico de Monterrey, México, in 2012, from the Electronic Technologies Engineering degree. He has 4 years of experience in Intel Components of Costa Rica, where he worked for 1 year as a process engineer for validation tests on the Intel processor platform. He also developed validation tests for the graphic components of the processor for two years. Currently working on characterization and analysis of performance and power of Intel processors
Participation is free. There are no tuition costs associated with participating in this school for those affiliated to CONARE institutions.
An email must be sent to firstname.lastname@example.org requesting the application form. On Thursday, July 20, 2017, people accepted at HPC School will be notified.
The maximum quota is 40 participants.
To be a student, professor, or researcher at any public university (UCR, TEC, UNA, UNED, UTN), from CONARE or any of its ascribed programs: CeNAT, PEN and SINAES.
Have an intermediate English level (reading and listening). Some presentations and exercises may be in English.
Have intermediate programming skills (C/C++ desired, Python acceptable) and intermediate Linux handling.
The National Advanced Computing Collaborative (CNCA) is a multidisciplinary space where scientific discovery is accelerated through the use of an advanced computing infrastructure. This infrastructure includes not only specialized and updated hardware, but also a set of efficient applications and trained personnel to take advantage of all this technology. This allows the CNCA to work in the central dimensions of development of research projects, training and service provision.
The Research Laboratory in Pattern Recognition and Intelligent Systems (PRIS-Lab) is part of the School of Electrical Engineering of the University of Costa Rica. It was founded in 2012 by Dr. rer. nat. Francisco Siles Canales as a core of transdisciplinary training. This laboratory seeks the generation of scientific research and technological innovation, in pursuit of integration between the academic community, the government, the productive sector, civil society and the environment, for the improvement of people’s quality of life. .
The objective of the Pattern Recognition and Machine Learning Group (PARMA Group) is to conduct research and development of solutions based on machine learning, in conjunction with other institutions
tion In addition to stimulating the organization of training activities for students and teachers, and professionals in the area (signal processing, machine learning, data mining). Some of the projects developed in the area of biomedicine involve the segmentation and tracking of cells in glioblastoma tissues, automatic prostate histology analysis and automatic blood flow analysis.
RedCONARE is the National Education and Research Network (RNIE) of Costa Rica, which provides Public Universities (UCR, TEC, UNA, UNED) with a telecommunications infrastructure and communication services, such as: eduroam, Mconf, LA Reference, the collaborative. etc. The NRENs or Advanced Networks are a space that has the community of higher education and research in the world to enhance their knowledge and contributions to humanity. In Costa Rica, RedCONARE has been positioning itself as a joint research and collaboration space between the entities that comprise it.