Numerical resources for HPC course
- Jean-Matthieu Etancelin, LMAP, IPRA
- Matthieu Haefele, LMAP, CNRS, IPRA
Syllabus:
- Chapter1 : Introduction : Context, history of computing, Parallel computing and hardware architecture
- Chapter2 : Profiling and serial optimisation in Python
- Chapter3 : Message Passing Interface (MPI) for distributed memory parallel computing
- Chapter4 : Python packages for shared memory parallel computing (Numba and/or Pythran)
Evaluation
- Lab on Chapter 1 and 2, 3h on November 5
- Lab on Chapter 3, 3h on January 14
- Mini-project: code, 3 pages report + 10 min individual interview, February 19
- Final grade : average of 3 exams
Practical details
- Python3 programming
- 36 hours for course and practical sessions
- Remote access for homework and projects
Expected skills
- Implement a parallel algorithm
- Execute a parallel program on a supercomputer
- Perform a profiling of a parallel application
- Analyse the parallel performances of a program
Additional resources:
- Usage of Bilbo cluster
- Linux command line:
- command line tutorial : https://tutorials.ubuntu.com/tutorial/command-line-for-beginners#0
- command line course : https://ryanstutorials.net/linuxtutorial/
Matplotlib configuration file on bilbo cluster:
mkdir -p ~/.config/matplotlib && cp /home/user/j/jmetancelin/Public/matplotlibrc ~/.config/matplotlib/matplotlibrc
- Enseignant: Jean-Matthieu Etancelin
- Enseignant: Matthieu Haefele