Exascale target (capability/capacity)
For very high frequencies (which affect e.g. smaller buildings and infrastructures) capability is required, which is guaranteed by the extreme scalability of codes involved. In terms of capacity, we need to explore uncertainty. This is done at workflow level.
Main benefits expected from large increase in compute resources
Much higher spatial resolution and frequency. Potentially addressing parts of the uncertainty such as source inversion. Direct reduction in time to solution.
Main challenges in harnessing large compute resources
Mostly resilience and the QoS of the system. We foresee issues related to the scale of very high-frequency runs.
Specific actions towards exascale at present time
The seismic codes are being part of a large-scale benchmark within the ChEESE project.