An important aspect I learned at the SST user community meeting this week was the relevance of the holistic workflow. I previously learned about the DES lifecycle in sst-core. Now I'm considering a broader view:
- learning about DES. Some users are not familiar with DES
- learning how sst-core provides a framework for DES
- comparing SST to other options to determine whether SST fits the user's desired outcome
- reviewing components and subcomponents available from
sst-elements and either leveraging those or modifying them or creating novel elements to simulate a model of interest.
- specifying the graph of components/subcomponents/links using either (Python driver file) or (ahp_graph) or (single JSON file) or (set of JSON files, one per rank).
- (optional) visualizing of the component graph prior to simulation
- running the simulation using SST
- live logs (for review and for visualization) and debugging during execution
- reviewing logs and stats after the simulation. CSV, HDF5, Elastic
- generating plots (in Excel, in Python, etc)
- documenting the simulation configuration and results for future reproducibility
Documenting the workflow above would be useful to explain on https://sst-simulator.org/sst-docs/
An important aspect I learned at the SST user community meeting this week was the relevance of the holistic workflow. I previously learned about the DES lifecycle in
sst-core. Now I'm considering a broader view:sst-elementsand either leveraging those or modifying them or creating novel elements to simulate a model of interest.Documenting the workflow above would be useful to explain on https://sst-simulator.org/sst-docs/