https://boinc.bakerlab.org/rosetta/results.php?hostid=397379&offset=40

As you can see here and in my results, 80% of my work units were reported correctly and with success but I received no credits.

Why that? Can you help me?

Thanks in advance.

---

Your results were returned after their deadline. As you are looking at the tasks tab in the advanced view of the BOINC manager, there is a column that shows the deadline for each work unit. The times shown in your BOINC Manager are adjusted to your local timezone. The date/times shown on the website are UTC.

When a task is downloaded, the deadline to return the completed result is currently 10 days from the download.

---

Rosetta Moderator: Mod.Sense

Your cache size is probably too large.

This can be changed in the "General Preferences" section of your account at the Rosettta site... "Connect To Network About Every ... Days" .

From your stats it appears that you have your cache set to 10 days. Set that to something like "5" to start and then increase it until you have a cache size that you are comfortable with but still allows you to process and return the workunits before the deadline.

---

See Problem with the work buffer on the boinc forum. Look at messages #8107 and #8133. This fellow concludes that for a deadline of 10 days (Rosetta) you need a queue of maybe 4.5 days? See for yourself and judge.

See Problem with the work buffer on the boinc forum. Look at messages #8107 and #8133. This fellow concludes that for a deadline of 10 days (Rosetta) you need a queue of maybe 4.5 days? See for yourself and judge.

Right. The "logic" built into 5.8.* is nuts IMO. You need to reduce the buffer setting in order to increase the buffer. *boggle* When I brought this up with the developers, I was told that the code would not change, and the blame lies with the projects that have too short deadlines.

---

Reno, NV
Team: SETI.USA

Right. The "logic" built into 5.8.* is nuts IMO. You need to reduce the buffer setting in order to increase the buffer. *boggle* When I brought this up with the developers, I was told that the code would not change, and the blame lies with the projects that have too short deadlines.

My own observations are that when I increase the connect interval I download more tasks, and when I decrease the connect interval I get fewer tasks...just like always. What I think is different is that people can no longer ask for very large queues that approach the deadline period. So instead of 10-day queues for Rosetta you will have to settle for something a little less: I think I read that for Rosetta a queue of perhaps around 7.5 days(?) can't remember.

My own observations are that when I increase the connect interval I download more tasks, and when I decrease the connect interval I get fewer tasks...just like always. What I think is different is that people can no longer ask for very large queues that approach the deadline period. So instead of 10-day queues for Rosetta you will have to settle for something a little less: I think I read that for Rosetta a queue of perhaps around 7.5 days(?) can't remember.

Yes, it does work as expected when the deadline is large. However, with short deadlines, you need to keep your setting at least 2-3 lower than the deadline length. Otherwise it ends up doing bizarre things like downloading a lot of WUs, then working them down to near 0, then downloading a lot more. Lather, rinse, repeat. Instead of doing what it should, which is just keeping a relatively large queue, replenishing the WUs as the completed jobs are returned.

---

Reno, NV
Team: SETI.USA