Check this out: https://boinc.bakerlab.org/rosetta/result.php?resultid=545819893

DONE :: 1 starting structures 31323.3 cpu seconds
This process generated 1 decoys from 1 attempts
======================================================
BOINC :: WS_max 5.22678e+008

BOINC :: Watchdog shutting down...
BOINC :: BOINC support services shutting down cleanly ...
called boinc_finish

</stderr_txt>
]]>

Validate state Valid
Claimed credit 171.179661854162
Granted credit 43.8651075559614


App version 3.45 and Boinc 7.0.28 and this task do not seem to agree. 171 vs 43? waste of time! good for science but rotten for morale. My RAC keeps diving even after I have returned loads of results and this does not help any either.

Some models are more compute intensive than others - I just picked another one of yours at random and it got slightly higher granted than claimed:
https://boinc.bakerlab.org/rosetta/result.php?resultid=545133259

---

Hard to make any claim about all RB 11's when you have only sited a single model attempted. Your task happened to hit a heavy model right off the bat, and never did actually complete it before the watchdog stepped in, so it's difficult gauge what that should be worth. This is part of why most credit discussions should point out that "on average" this all works out. You hit a tough one and reported a higher credit claim that what others were seeing with their models. This ticks up the average for these tasks and future reports are awarded some slightly higher credit than they would have if such a tough model had not been reported back.

Always try to keep in mind that the opposite also happens all the time. Your machine crunches a model that happens to be very easy and is awarded more credit than others claim to complete a model, and you are granted more than that model might otherwise be worth, because most of the time that class of tasks is generating models that take longer than yours did.

The system is not perfect. But I would argue it is better than running all of the identical models two or three times just to confirm a given credit claim is within reason or to determine how much credit to award.

---

Rosetta Moderator: Mod.Sense

A think i know probable cause of low credit on this WU type: problems with checkpoints (probably do not work at all in some cases)
I see many times what WUs prom this type reset progress to 0% after each restart(no matter how long they run before restart).
And 3 of them running on my computer right now:
30 min from start
1.5 hr
4.5 hr (btw - target cpu time = 3hr)
and i check working wolders("slots") for this WUs and do not found any checkpoint related files for WU which runs 4.5hr (checkpoints files names start from "chk_" + boinc_checkpoint_count.txt)
This WU rb_11_23_35034_66182__t000__0_C4_SAVE_ALL_OUT_IGNORE_THE_REST_65305_415_0 (not completed yet)

Previous my problem WUs(loosing progress, low credit) names starts from rb_11_23 too. Greg_BE report low CR on rb_11_23... too

While all other rb_11_.. WUs (rb_11_14... rb_11_15... rb_11_24...) work fine for me and i found checpointing files in working folder for this WUs.

Restarts do make it take longer to complete a task, but since the task's CPU seconds also reset, it should not change the credit per CPU second when looking at the CPU seconds the tasks reports with.

Perhaps the tasks are hitting the 5 restarts without progress point though, and that causes them to be cut-off and reported in, regardless of how many models they have completed. So, perhaps that is the commonality to those that report poor credit.

---

Rosetta Moderator: Mod.Sense