| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Completed Research
Forum: FightAIDS@Home Phase 2
Thread: Valid credit given with only .89 hours? Can a tech take a look at this? |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 15
|
|
| Author |
|
|
uplinger
Former World Community Grid Tech Joined: May 23, 2005 Post Count: 3952 Status: Offline Project Badges: 
|
I am looking into this now. I will be pulling a few examples from the database to see how wide spread of an issue this is,
tmedve, Your first example is a good one as to what should not have happened on granted credit. Thanks, -Uplinger |
||
|
|
Speedy51
Veteran Cruncher New Zealand Joined: Nov 4, 2005 Post Count: 1326 Status: Offline Project Badges: 
|
I am following this topic with interest. I wish I was lucky enough to get a task that ran for that amount of time
---------------------------------------- |
||
|
|
keithhenry
Ace Cruncher Senile old farts of the world ....uh.....uh..... nevermind Joined: Nov 18, 2004 Post Count: 18667 Status: Offline Project Badges: 
|
I am looking into this now. I will be pulling a few examples from the database to see how wide spread of an issue this is, tmedve, Your first example is a good one as to what should not have happened on granted credit. Thanks, -Uplinger Keith, my suspicion is that the science app or the server is telling the client to stop processing a task for whatever reason. If the reason is not something good, it should be marked as an error. Due to the nature of FAH2, if processing of a task is stoppend prior to the end, it can be for a valid reason. I think the validator isn't "adjusting" for that. Thus, when one user stops processing a task at the first iteration and the wingman processing it to the end, you get the most extreme version of the problem. The validator sees the stopping of the task at any iteration prior the end as normal, full processing of the task and, because of thelarge difference in work, the wingman gets treated as an outlier. Even when this happens on the next to last iteration, while the wingman may not get treated as an outlier, they are still getting penalized on credit granted. I suspect that the validator needs to proportionally adjust the amount of work done by a user that does not fully process the task and thus the credit that adjusted amount would be granted in evaluation the wingman's work. Thus, if a user processing ends on the first of ten iterations, ten times the credit they get granted should be what the wingman who fully processes gets compared to in determining their granted credit; ie. eash user's "score" would be multiplied by the number of iterations possible for the wu (10) and divided by the number of iterations completed to determine the credit used for each user in the normal averaging and then the amount granted would be that multiplied by the number of iterations completed and divided by the number of possible iterations. |
||
|
|
tmedve
Senior Cruncher USA Joined: Nov 16, 2004 Post Count: 191 Status: Offline Project Badges:
|
keithhenry, that makes sense to me based on what I saw.
----------------------------------------How prevalent is it that people stop their tasks early? Out of the 15 tasks that I can currently see on my results status page, all of them are for one user (no wingmen). Is it possible then that I could stop one of these tasks and still get credit for the n0% that I completed at the due date? (not that I would do that willfully). Speedy51, those short tasks don't get you very many points, but apparently, you can accumulate a lot of tasks in a short time. It seems that those "short" tasks are not really short, but the system may think of them as short because of the trickles at 10% intervals. If what keithhenry is proposing is correct.  [Edit 2 times, last edit by tmedve at Feb 9, 2016 2:42:29 AM] |
||
|
|
SekeRob
Master Cruncher Joined: Jan 7, 2013 Post Count: 2741 Status: Offline |
You can not stop a task [abort], only the server can instruct to do so if no sufficient progress is being made in the allotted time. Below 70-80% completion these results are considered of no great science value [post uplinger]
|
||
|
|
|