I have five systems in part dedicated to Rosetta - all AMD Phenom CPU's ranging from 2 to 8 gigabytes in memory size.

All running Linux, all executing Rosetta Mini 2.11 as the application.

Free disk or memory does not seem to be a player in this issue.

Several times a day I will note a task getting to about 95% complete with about 10 to 11 minutes left to run.

At this point the clock stops counting down and progress increments very slowly. Top shows them still drawing their full allotment of CPU cycles.

(but not at 100% as if they were in a tight loop)

Since I have never aborted one of these tasks I believe that they go on to finish eventually. But they are sometimes hung at this point for hours.

Thanks

The runtime preference is configurable and defaults to 3 hours. For most types of work, this means several models will be produced in that time and that the completion will be roughly inline with the preference. If tasks require more then the typical time per model, or if the runtime preference is set to less then 3 hours, it is not uncommon for tasks to task longer then the preference.

The estimated time to completion is based upon the history of your machine, and the runtime preference, but without knowledge of the specific task's characteristics. There really isn't a great way to know in advance if the initial estimate will be correct or not, so it counts down normally assuming the task will complete at approximately the estimated time, but once the time remaining gets below 10 minutes, it is increasingly likely that the task will not complete at the estimated time. And so the program adjusts the estimate, essentially making it count down slower and slower.

It's just an estimate. The behavior is normal. Otherwise estimated time remaining would become negative, and that would be even more confusing.

---

Rosetta Moderator: Mod.Sense