|
Phase1G /
Phase1GTeleconParticipants:
NotesSubmissionWe discussed pros and cons of submitting to EGPGV on Friday vs waiting. Our consensus was to wait. We view EGPGV as a good opportunity, but want this to be fully baked before we submit. We don't have a firm view for where to submit, but decided to nominally shoot for SC16. We don't know if SC16 would actually take the paper, but we like the notion of being finished in one month. We can discuss further as the paper takes form. WorkflowsOur call today mostly focused on scenarios with multiple instances of an in situ program/systems working in coordination, which we started to call workflows. As an example: - A: "tight coupling" with the simulation code where data triage occurs - B: "loose coupling" on dedicated resources where vis/analysis occurs on the triaged data - (and the triaged data sent from A to B) This example can get more complicated, with C, D, and so on. One concern is that our proposed axes don't capture this scenarios. A proposal was to consider A and B as separate in situ systems. Therefore, a description of the combined A+B system would be a 6x2 matrix, with further description of the workflow combining them. An alternate proposal was to consider them jointly, and to have a spectrum of options for each axis. (We didn't seem to favor this proposal, as it didn't seem to capture as much detail.) A final complicating factor was what constitutes an instance (i.e., an A, B, C, or D). One view is that that we should only count instances where vis/analysis is being done. Another view is that anything that touches or moves data should be included. We all wanted to think about this more, but I thought I would send this out to the group so others can start mulling on it as well. |