Yur assumption is wrong, at least for the default. Usually the reporting is done in batches (keeping the deadline, anyway), in order to keep the server load somewhat limited. If you absolutely need to report immediately, you can do that manually, or you can set it in the config files. However, this is not recommended, otherwise we will see a heavy overload of servers if everyone does that.

Regards,
Martin

There's many reasons for reporting, the list is http://www.boinc-wiki.info/Ready_to_Report

The most common reason for reporting is to get more work. If the DCF isn't constantly oscillating but all tasks take the estimated time to run, anyone running a single BOINC-project would most of the time have one work-request per task, meaning if example each task takes 6 hours to run, on average it will be "ready to report" for 3 hours. The more cores a computer has the more likely it's not following this exactly, so you'll report multiple tasks at once.

Still, a good rule-of-thumb is, by letting BOINC-client decide when to report, there'll be one scheduler-request per task. If you instead uses <report_results_immediately>, there will be 2 scheduler-requests per task...


As for changing the default client-behaviour, this has zero priority, since it's highly undesirable to do any changes that will unnessessarily increase the server-load. The database is already the bottleneck in some instances, increasing the database-load will just make things worse. As for being "unimportant", if the database gets too busy handling all the reports from all the users that "needs" to report "immediately", it won't have enough time to add newly-generated work, meaning users won't get any work, and they'll sit idle instead, or get work from another BOINC-project instead.

Well, that's for not being english, my interpretation of your text was along the lines of "database operation being expensive is unimportant".