GPU programming using CUDA
10 - 11 February 2020
The focus is to understand the basics of accelerator programming with the CUDA parallel computing platform model.
CUDA is a parallel computing platform and application programming interface (API) model created by NVIDIA. It allows software developers to use a CUDA-enabled graphics processing unit (GPU) for general purpose processing – an approach known as GPGPU. The CUDA platform is a software layer that gives direct access to the GPU’s virtual instruction set and parallel computational elements, for the execution of compute kernels.
The course also contains performance and best practice considerations, e.g., gpu libraries, performance optimizations, tools for debugging and profiling.
After the course the participants should have the basic skills needed for utilizing CUDA and OpenACC in new, or existing (own code) programs.
The course addresses participants who are familiar with the C/C++ or Fortran programming languages and have working experience with the Linux operating system and the use of the command line. Experience with parallel programming or gpu programming (MPI,OpenMP and CUDA) is not required.
Bring your own laptop in order to be able to participate in the training hands on. Hands on work will be done in pairs so if you don’t have a laptop you might work with a colleague.
Course language is English.
The maximum number of participants is 30.
Registrations will be evaluated on a first-come, first-served basis. GRNET is responsible for the selection of the participants on the basis of the training requirements and the technical skills of the candidates. GRNET will also seek to guarantee the maximum possible geographical coverage with the participation of candidates from many countries.
Address: 2nd Floor, 7, Kifisias Av. GR 115 23 Athens
Information on how to reach GRNET headquarters ia available on GRNET website: https://grnet.gr/en/contact-us/
Accommodation options near GRNET can be found at: https://grnet.gr/wp-content/uploads/sites/13/2015/11/Hotels-near-GRNET-en.pdf
ARIS - System Information
ARIS is the name of the Greek supercomputer, deployed and operated by GRNET (Greek Research and Technology Network) in Athens. ARIS consists of 532 computational nodes seperated in four “islands” as listed here:
426 thin nodes: Regular compute nodes without accelerator.
44 gpu nodes: “2 x NVIDIA Tesla k40m” accelerated nodes.
18 phi nodes: “2 x INTEL Xeon Phi 7120p” accelerated nodes.
44 fat nodes: Fat compute nodes have larger number of cores and memory per core than a thin node.
1 ml node: Machine Learning node consisting of 1 server, containing 2 Intel E5-2698v4 processors, 512 GB of central memory and 8 NVIDIA V100 GPU card.
All the nodes are connected via Infiniband network and share 2PB GPFS storage.The infrastructure also has an IBM TS3500 library of maximum storage capacity of about 6 PB. Access to the system is provided by two login nodes.
Dr. Dellis (Male) holds a B.Sc. in Chemistry (1990) and PhD in Computational Chemistry (1995) from the National and Kapodistrian University of Athens, Greece. He has extensive HPC and grid computing experience. He was using HPC systems in computational chemistry research projects on fz-juelich machines (2003-2005). He received an HPC-Europa grant on BSC (2009). In EGEE/EGI projects he acted as application support and VO software manager for SEE VO, grid sites administrator (HG-02, GR-06), NGI_GRNET support staff (2008-2014). In PRACE 1IP/2IP/3IP/4IP/5IP he was involved in benchmarking tasks either as group member or as BCO (2010-2017). Currently he holds the position of “Senior HPC Applications Support Engineer” at GRNET S.A. where he is responsible for activities related to user consultations, porting, optimization and running HPC applications at national and international resources.
Nikolaos Nikoloutsakos holds a diploma of Engineering in Computer Engineering and Informatics (2014) from the University of Patras, Greece. From 2015 he works as software engineer at GRNET S.A. where he is part of the user application support team for the ARIS HPC system. He has been involved in major national and European projects, such as PRACE and EUDAT. His main research interests include parallel programming models, co-processor programming using GPUs and Intel Xeon Phis.
Dr. Ioannis E. Venetis received his PhD in 2006 from the Computer Engineering and Informatics Department at the University of Patras, Greece. Currently he teaches "Parallel Processing" and "Software and Programming for High Performance Systems" at the same Department. He has participated in numerous research projects in the area of Parallel Computing. His main research interests include parallel programming models, run-time systems for supporting such models, co-processor programming (especially using GPUs and the Intel Xeon Phi) and parallelization of computationally demanding applications.
GRNET – National Infrastructures for Research and Technology, is the national network, cloud computing and IT e-Infrastructure and services provider. It supports hundreds of thousands of users in the key areas of Research, Education, Health and Culture.
GRNET provides an integrated environment of cutting-edge technologies integrating a country-wide dark fiber network, data centers, a high performance computing system and Internet, cloud computing, high-performance computing, authentication and authorization services, security services, as well as audio, voice and video services.
GRNET scientific and advisory duties address the areas of information technology, digital technologies, communications, e-government, new technologies and their applications, research and development, education, as well as the promotion of Digital Transformation.
Through international partnerships and the coordination of EC co-funded projects, it creates opportunities for know-how development and exploitation, and contributes, in a decisive manner, to the development of Research and Science in Greece and abroad.
National Infrastructures for Research and Technology – Networking Research and Education