Panel Discussion on Peering HPC and Data-Driven Science: Scalable Cross-Facility and System Workflows
Presenter
DescriptionThe panel session will attempt to foster a synthesis-driven discussion from the speakers from each of the institutions (CSCS, ORNL, and PSI, and participating interested vendors such as NVIDIA and others). Questions along the following lines will enable a back-and-forth dialog from competing interests (computing facilities weighing data-oriented and simulation-oriented campaigns, and observational science facilities with large-scale data needs).
1) How do you prioritize the type of services to couple observational and computational facilities?
2) What is the strategy to assess your need of data and computational resources?
3) How should two facilities bring their experts together to advance the cross-facility coupling?
4) What is the likelihood of the workloads to be dominated by data/AI campaigns for broad subdomains in science?
5) What business models (pay-as-you-go vs. at-scale-upfront) apply to such cross-facility partnerships?
6) What is the preferred role of the commercial sector when they contribute to these (often) purpose-built multi-facility projects?
1) How do you prioritize the type of services to couple observational and computational facilities?
2) What is the strategy to assess your need of data and computational resources?
3) How should two facilities bring their experts together to advance the cross-facility coupling?
4) What is the likelihood of the workloads to be dominated by data/AI campaigns for broad subdomains in science?
5) What business models (pay-as-you-go vs. at-scale-upfront) apply to such cross-facility partnerships?
6) What is the preferred role of the commercial sector when they contribute to these (often) purpose-built multi-facility projects?
TimeWednesday, 7 July 202112:30 - 13:00 CEST
LocationMère Royaume
Session Chair
Event Type
Minisymposium
CS and Math
Emerging Applications
Chemistry and Materials
Physics
Life Sciences
Engineering
