Register

Abstract:

Researchers from a variety of disciplines in natural sciences are becoming increasingly interested in High-Performance Computing (HPC). However, factors such as lack of know-how and apparently steep learning curve can hinder new users from moving into HPC. Adopting a workflow to run jobs, monitor their status, and postprocess their outputs can be challenging for new users or for those who have limited experience.

This webinar will try to alleviate this issue by illustrating common steps typically required to run jobs in HPC environments. Using a simple yet real hydrological modelling example, the demonstration consists of running a parallel FORTRAN application using R, as well as retrieving and summarizing its results in the Cedar Cluster.

The presentation will demonstrate a practical workflow that encompasses:

1. preparing submission scripts,
2. testing parallel-computation jobs before submission,
3. debugging parallel-computation jobs that return an error, and
4. monitoring single or multiple jobs running at a time. Finally, a summary of the resources utilized to run a computationally-intensive research on the Cedar cluster will be presented.