This workshop goes to more advanced topics and techniques in high-performance computing. Parallel algorithms, more advanced topics in message-passing interface (MPI), parallel tools and numerical libraries, parallel I/O, as well as shared-memory parallelization techniques along (with OpenMP) and combined with MPI are discussed and exemplified.
Agenda
Monday 11 November, 2013
Learning outcome
After the course the participants should have an idea about the best practices in HPC programming, how to scale up applications and optimize them to different platforms.
Pre-requisites
The PATC course "Introduction to Parallel Programming" or equivalent background knowledge together with fluency in Fortran 95 and/or C programming languages will be assumed.
Price: Free of charge
Lecturers: Jussi Enkovaara (CSC), Sami Ilvonen (CSC) and Martti Louhivuori (CSC)
Agenda
Monday 11 November, 2013
- 09:00 - 09:45 Course intro, MPI & OpenMP recap
- 09:45 - 10:30 OpenMp & MPI recap lab
- 10:30 - 10:45 Coffee break
- 10:45 - 11:15 OpenMp & MPI recap lab
- 11:15 - 12:00 Advanced OpenMP
- 12:00 - 13:00 Lunch break
- 13:00 - 13:45 Hybrid programming
- 13:45 - 14:30 Hybrid programming lab
- 14:30 - 14:45 Coffee break
- 14:45 - 15:30 Hybrid programming lab
- 15:30 - 16:15 Demo: Debugging
- 16:15 - 16:30 Wrap-up
- 09:00 - 09:45 Application performance analysis
- 09:45 - 10:30 Performance analysis lab
- 10:30 - 10:45 Coffee break
- 10:45 - 11:15 Advanced MPI I
- 11:15 - 12:00 Advanced MPI lab
- 12:00 - 13:00 Lunch break
- 13:00 - 13:45 Advanced MPI lab
- 13:45 - 14:30 Advanced MPI II
- 14:30 - 14:45 Coffee break
- 14:45 - 15:30 Advanced MPI III
- 15:30 - 16:15 Advanced MPI lab
- 16:15 - 16:30 Wrap-up
- 09:00 - 09:45 Parallel I/O
- 09:45 - 10:30 Parallel I/O lab
- 10:30 - 10:45 Coffee break
- 10:45 - 11:15 Parallel I/O lab
- 11:15 - 12:00 Parallel I/O II
- 12:00 - 13:00 Lunch break
- 13:00 - 13:45 Parallel I/O III
- 13:45 - 14:30 Parallel I/O lab
- 14:30 - 14:45 Coffee break
- 14:45 - 15:30 Parallel I/O lab
- 15:30 - 16:15 HPC libraries
- 16:15 - 16:30 Wrap-up
Learning outcome
After the course the participants should have an idea about the best practices in HPC programming, how to scale up applications and optimize them to different platforms.
Pre-requisites
The PATC course "Introduction to Parallel Programming" or equivalent background knowledge together with fluency in Fortran 95 and/or C programming languages will be assumed.
Price: Free of charge
Lecturers: Jussi Enkovaara (CSC), Sami Ilvonen (CSC) and Martti Louhivuori (CSC)