Trac: #430 Profile driven partition scheme for SST

[nmhamster]

Use case from ISCA tutorial - can we provide a way to profile a simulation linkscomponents and then record this information, load in for a future run so that the partition scheme is optimized for the next run.

[jjwilke]

Sounds like a good intern project! Which we don't have this year... @nmhamster @feldergast Do we have any interesting use cases? Something useful enough to be worth a workshop paper? If there's still enough incentive to do this, I don't think it would be too hard.

I would suggest the best way to do this would be to add a flag

--enable-partition-profiling

that counts the number of bytes on each link (-> edge weights) and counts the number of events (->node weights) and dumps a JSON or something. We would then need a Python function

sst.usePartitionerProfile("mydump.json")

Every time a link or component gets instantiated, it goes to JSON, looks up its weight, and writes that to the config graph.