Please see https://boinc.bakerlab.org/rosetta/workunit.php?wuid=101622736

Claimed 30.53
Granted 4,293.17

Computer 567404 has multiple of these.

I know there can be some diff but 140+ times seems a bit much.

Marcel

---

that computer is also mentioned here:
https://boinc.bakerlab.org/rosetta/forum_forum.php?id=2

---

Here's a link to the info page for that computer.

---

Rosie, Rosie, she's our gal,
If she can't do it, no one shall!

Now I see why my credits granted often is much less than credits claimed!!

---

Now I see why my credits granted often is much less than credits claimed!!

ouch! my single core can not keep up with that machine

---

Now I see why my credits granted often is much less than credits claimed!!

ouch! my single core can not keep up with that machine

Now I see why I saw a granted credit decline.

---

ya, hard to keep up with 10 seconds per model! I've reported this one as well. It is now just a question of the root cause of the disparity. Either someone is trying to get credits for work they are not crunching, or there is a bug somewhere. Since we've seen this now occur on several hosts, and we know the rest of us are all seeing a decline in our granted to claimed ratio, and that we now have a host owned by someone other then anonymous, and these all seem to take about 10 seconds per model (just about same time as it takes to initialize a model) I tend to think these are due to a software change.

Let's HOPE that it is a sign that Rosetta is now so intelligent that it immediately recognizes these initial states are not worth the time to crunch, and that we just have to work out a credit system that is fair to those of us that don't happen to hit 1,000 models in a row that are so bad that they aren't worth crunching through. :) ...it COULD happen!

---

Rosetta Moderator: Mod.Sense

It's likely an issue with the application and the particular work unit(s). I modified the validator to take care of these so let me know if any more come up.

it looks more like 8-11 minutes average per decoy that this machine is getting if my calculations are correct. i randomly sampled a few of the work units on the list of that machine.

---