Next instalment of questions for the Abacus team.
Abacus will use the AF SDK for data access to a PI Server or Collective, right?
Initial AF SDK v2.5 release won't support PI Buffer Subsystem, right?
Subsequent version of AF SDK will support PI Buffer Subsystem...I hope.
PI Buffer Subsystem cannot buffer & fan to multiple PI Collectives from a single node. (This has got to change...)
How will Abacus, which is running within a platform that supports connections to multiple PI Collectives, support writing calculation results to multiple PI Collectives from a single node running the Abacus 'scheduler'?