search

DARMA-tasking / LB-analysis-framework

Analysis framework for exploring, testing, and comparing load balancing strategies
Other
4 stars 1 forks source link

#482: Improve and complete JSON writing #483

Closed cwschilly closed 10 months ago

cwschilly commented 11 months ago

Fixes #482

This PR also deprecates the -1 option (to balance all phases). If no phase is specified, it defaults to balancing phase 0.

cwschilly commented 11 months ago

I've found a bug in the code--putting in draft until I fix it

Update: fixed

cwschilly commented 11 months ago

I am reworking the unit test to use multi-phase data generated by vt

cwschilly commented 11 months ago

I'm running into an issue with assigning communication dicts to the correct rank. Originally, I was pairing communications with ranks by checking to see which object the comm is from (communications["from"]["id"]). Then I would compare that id to the ids from rank.get_object_ids(). If they matched, I would write out that communication on that rank.

However, it seems like a lot of these communications are sent from "home": 0, "id": 0, but there isn't an object (entity, in the JSON) that has id: 0 (meaning that there is never a match). Am I misunderstanding something?