Parallel file systems provide the storage backend of HPC clusters. Since their characteristics have a large impact on the I/O performance and in consequence on the runtime of compute jobs, it is important to understand them and how to use them efficiently. This applies for both the perspectives of a simple user as well as on developers creating their own codes. The course will give an overview of parallel file systems and parallel-IO. It also covers anti-patterns which result in reduced performance. Furthermore, examples for efficient parallel-IO will be given.
30
Online (BigBlueButton)
This event includes following dates: