My Masters thesis is on task graph scheduling for multi core processors. I have developed a new task clustering and scheduling algorithm. I wanted to know about how can I find data to compare it with my work? Me and my supervisor have tried contacting the authors of a few similar works for data but none of them is sharing their data. Is there any benchmark for comparing such types of problems?

  • $\begingroup$ "none of them is sharing their data" -- that is incredibly bad scientific practice, but what can you do... :/ $\endgroup$
    – Raphael
    Jul 4, 2014 at 16:37
  • 1
    $\begingroup$ I don't think the linked "duplicate" will be of any use to OP. task-graph scheduling research generally proposes new heuristics and so you need real task-graphs to show that your heuristic is at all relevant to any real-world problem. In the absence of real task-graphs randomly generated graphs would be better than trying to turn road or airport networks into fake task graphs. Task graphs (generally) are DAGs and have "amounts of work" associated with the nodes and (perhaps) communication costs associated with edges. $\endgroup$ Jul 4, 2014 at 18:40
  • 1
    $\begingroup$ @ResearchStarter, could you describe the needed characteristics of the required graphs, and link us some non-paywalled articles that show the kinds of graphs you are interested in? $\endgroup$ Jul 4, 2014 at 18:47
  • $\begingroup$ I'm personally interested in seeing some task graphs as well. My research is indirectly related. WanderingLogic is right, task graph has specific characteristic. Unlike the "map" graph in the "duplicate" link. Please un-duplicate this question. And for the OP, I suggest you describe the format/meaning of the kind of graphs you want more explicitly to avoid further confusion. $\endgroup$ Jul 4, 2014 at 19:18
  • $\begingroup$ I don't think the duplicate is applicable either, but neither is this question suitable to the site as it stands. It's asking “where can I find …” and that's no what we do here — this is a questions and answers site, not a search engine. Asking how to build test cases would be on-topic, but this looks like a whole research project in itself, so it's too broad. More specific questions along the path of building test cases would be perfectly fine. $\endgroup$ Jul 5, 2014 at 8:52


Browse other questions tagged or ask your own question.