This course will be delivered as an ONLINE COURSE for remote participation due to the COVID-19 measures enforced by most European governments.
Description
This course is part of the first OpenMP training organized by the LUMI User Support Team (LUST) and supported by CSC – IT Center for Science Ltd (Finland) and CSC Autumn of HPC 2021 (five modules program on teaching essential skills in parallel programming for modern GPU-accelerated supercomputers) under PRACE Training Centres activity.
This training aims to help the users to port their code to LUMI, the European pre-exascale supercomputer, that will achieve its high computing power thanks to a large number of nodes with AMD GPUs.
This course address the use of OpenMP for programming co-processors such as GPUs. It focuses on how to get the best out of OpenMP in terms of performance by exploring the implications of possible OpenMP parallelization strategies. Advanced topics such as asynchronous execution, interoperability with CUDA/HIP and the use of multiple GPUs are covered.
The course consists of lectures and hands-on exercises. Participants will be provided with training account on the LUMI Early Access Platform.
For more information please contact LUMI User Support Team.
Learning outcomes
After the course the participants will be able to use OpenMP offloading to GPU: manage efficiently the transfer of data to/from the accelerator, distribute optimally the work among different tasks and synchronize them to minimize waiting times, as also how to integrate OpenMP with HIP(CUDA). They will be able to port non-parallel code to OpenMP offloading or modify an existing one.
Prerequisites and content level
The participants are assumed to have working knowledge of Fortran and/or C/C++ programming languages. Knowledge of OpenMP tasking model and experience with OpenMP is strongly required.
The content level of the course is broken down as: beginner's - 25%, intermediate - 25%, advanced - 50%, community-targeted content - 0%.
Agenda
Day 1, Tuesday 9.11
-
11:00 – 13:00 Introduction to OpenMP offloading (with 15 min coffee-break)
13:00 - 14:00 Lunch
-
14:00 – 17:00 Hands-on and Q&A (with 15 min coffee-break)
Day 2, Wednesday 10.11
-
11:00 – 13:00 Asynchronous offloading and HIP/CUDA interoperability (with 15 min coffee-break)
13:00 - 14:00 Lunch
-
14:00 – 17:00 Hands-on and Q&A (with 15 min coffee-break)
Lecturers
Dr.-Ing. MICHAEL KLEMM is part of the HPC Center of Excellence at AMD. His focus is on High Performance & Throughput Computing Enabling. He obtained an M.Sc. in Computer Science in 2003. He received a Doctor of Engineering degree (Dr.-Ing.) in Computer Science from the Friedrich-Alexander-University Erlangen-Nuremberg, Germany, in 2008. His research focus was on compilers and runtime optimizations for distributed systems. His areas of interest include compiler construction, design of programming languages, parallel programming, and performance analysis and tuning. He is CEO of the OpenMP Language Committee, where he is also leading the group on the OpenMP Error Model efforts.
Dr. CHRISTIAN TERBOVEN is a senior scientist and leads the HPC group at RWTH Aachen University. His research interests center around Parallel Programming and related Software Engineering aspects. Dr. Terboven has been involved in the Analysis, Tuning and Parallelization of several large-scale simulation codes for various architectures. He is responsible for several research projects in the area of programming models and approaches to improve the productivity and efficiency of modern HPC systems. He is further co-author of the new book “Using OpenMP – The Next Step“
Organizers
NICOLINO LO GULLO (LUMI User Support Team), ORIAN LOUANT (LUMI User Support Team)
For course information please contact LUMI User Support Team
Language: English
Price: Free of charge (2 training days)
REGISTRATION is OBLIGATORY since the details to access the online course will be provided to the registered and accepted attendees only. If you have registered to this course and you are not able to attend, please CANCEL your registration in advance by sending an email to patc@csc.fi
