World Community Grid - View Thread - If your device stops work fetch when using % of CPU time of BOINC preference!

There's a confirmed bug that if a client is set to "Use at most xx% of CPU time" to less than 100%, work fetch might not operate properly. If you have it set to less than 100%, try the maximum value and see if the problem goes away. When it does, let us know the exact version and also how much there is buffered, or if there are even idle processors. The latest 7.1.17 test client has a fix for this.

We have a new version to test. It includes the following changes since
7.1.15:
* Work fetch and scheduler changes

There was a bug which prevented work-fetch for a project if the CPU
throttling preference was not 100%. Try various values and make sure
the will still fetch work when it is running low.

ALSO, those who contribute to NCI projects such as WUprop [a Non-CPU-Intense project tracking what work a client is processing producing great device performance stats at 0,1% of 1 CPU thread], it seemed that sometimes after completion of a WU, the client would not fetch a new task for another hour or multiples of that. The fix was checked in after 7.1.17 test release was published... it will be 7.1.18 or later before the solution shows.

client: fix bug that delayed work fetch from non-CPU-intensive projects

We were waiting until there was no task for the project
before asking for another task.
We should have been waiting until there was no in-progress task.

N.B. CPU time throttling goes into the work fetch equation too in terms of buffering. If you set it to 50%, with a buffer value of 1.0 days, only half a day should be buffered. If you still get 1 full day of work, something is wrong, because then you have 2 days of calendar time computing cached.

edit: typo