Workload Archive

This page archives the workload logs and problem descriptions presented at JSSPP workshops.

Ultrasound DAG workflows

This workload comes from the 2020 paper "Optimizing Biomedical Ultrasound Workflow Scheduling Using Cluster Simulations" by Marta Jaroš, Dalibor Klusáček and Jiří Jaroš.
It contains real life-based descriptions of computational workflows that are used to simulate ultrasound propagation and its effects in a human body. Detailed description of the workload can be found in the aforementioned paper.
The workload contains two classes of DAGs. The first class (20 DAGs) represents HIFU (High Intensity Focused Ultrasound simulations) and the second class (15 DAGs) represents neurostimulation simulations. The workloads use simple text format (similar to SWF), i.e., one line per job (job = DAG task).
Task dependencies are captured by a dedicated field (Preceding job) which holds the list of all direct predecessors for a given task. If a given job has more than one predecessor in the DAG, then & character is used to concatenate the list of these predecessor IDs. For example, 1&2&3 means that the given job can only start once jobs 1, 2 and 3 are all completed. On the other hand, -1 means that a given task has no direct predecessor.
DAG workload format used in these logs is: ID, Runtime [s], Nodes [#], Memory per Node [kB], Preceding job, DAG name, Task ID, Task name.

Usage and Acknowledgments

The Ultrasound DAG workflows were graciously provided by the Faculty of Information Technology at the Brno University of Technology. If you use this log in your work, please use a similar acknowledgment. To acknowledge authors' work please consider citing the paper that introduced this log:
Marta Jaroš, Dalibor Klusáček and Jiří Jaroš, "Optimizing Biomedical Ultrasound Workflow Scheduling Using Cluster Simulations". In Job Scheduling Strategies for Parallel Processing, May 2020.
 

JSSPP 2021 - the Workshop on Jobs Scheduling Strategies for Parallel Processing. Contact email: jssppw@gmail.com