Message boards :
Number crunching :
Claimed/Granted Credit For Workunits Explained
Message board moderation
Author | Message |
---|---|
Gordak Send message Joined: 22 May 03 Posts: 4 Credit: 11,045 RAC: 0 |
I thought this would help out those getting started. Please correct me if you discover any mistakes. Claimed Credit http://boinc-doc.net/site-common/glossary/letter-c.php#claimed-credit Formula: claimed credit = ([Your CPU's measured floating point speed]+[Your CPU's measured integer speed]) * [Your CPU's time to complete workunit in seconds] / 1728000 (See SETI@Home's benchmark of your computer's processor speed as these are the values that are used in computing claimed credit.) Granted Credit http://boinc-doc.net/site-common/glossary/letter-g.php#granted-credit SETI@Home uses BOiNC's credit granting method #3: (Remove the highest and lowest values and take the average of the claimed credit values of the remaining results.) Detailed Explaination on How SETI@Home Grants Credit SETI@Home issues the same workunit to four computers (on average). The highest and lowest claimed credit values of the computed results are thrown out, and the average claimed credit of the remaining (usually two) workunit results are awarded to all computers involved in that workunit's calculation. This is to prevent exceptional computers ("latest "hot-rod" processor" or "ancient legacy processor") from drastically altering the credit granted (for better or worse). Sometimes, only three workunit results are returned. In which case, the claimed credit of the computer with neither the highest nor lowest score (the "middle computer") is granted to all computers involved in that workunit's calculation. If less than three workunit results are returned, perhaps because of problems with the workunit's results from some machines, uploading/downloading difficulties, or the workunit issued to target computers expired, SETI@Home may issue more copies of the same workunit (usually two more). Three conditions must be met before credit is granted (I don't know of any others).
|
Bruno G. Olsen & ESEA @ greenholt Send message Joined: 15 May 99 Posts: 875 Credit: 4,386,984 RAC: 0 |
Unless we look at boinc as a whole, and take variations of different projects into account - I found no mistakes ;) Only a small clarification (as this is intended for newbies ;)): "At least three, non-flawed workunit results are successfully returned" is a two part process - the actual upload of the result, and then the result must be reported. You can upload a result before it's deadline and report it after it's deadline which means the result was returned after the deadline - which leeds to no credit no matter what other conditions are met. (as far as I understand it, anyway) |
Ingleside Send message Joined: 4 Feb 03 Posts: 1546 Credit: 15,832,022 RAC: 13 |
> SETI@Home issues the same workunit to four computers (on average). The highest > and lowest claimed credit values of the computed results are thrown out, and > the average claimed credit of the remaining (usually two) workunit results are > awarded to all computers involved in that workunit's calculation. Not correct... If nothing is backlogged, SETI@home for BOINC works like this: 1; Immediately when 3 "success"-results is reported, the wu is marked as "need_validate" by the Transitioner. 2; If Validator finds a pair of results to be "similar enough", the wu passes validation. If can't find a pair, is marked "no consensus yet" and must wait for another "success"-result being reported. If found a pair, tries to validate all of the "success"-results. Any failing validation is marked "invalid". 3; The Validator counts up how many of the results passed validation in #2, and awards credit after this formula: a; If only 2 passed validation, grant lowest claimed to all. b; If 3 or more, remove highest and lowest claimed and average the rest. 4; Any later reported results is tried validated, and if passes validation gets the same credit as in #3. No re-calculation of crediting is done. So, as long as nothing is backlogged, the normal in SETI@home is the 3 "success"-results reported 1st decides the credit, and the middle claimed of these 3 is the credit for all. Only if something is backlogged, or if failed validation due to "no consensus yet", is 4 or more results used when deciding crediting. In rare ocassions, one or more of the results is too different from the rest, so 3a is used when deciding crediting in SETI@home. |
Paul D. Buck Send message Joined: 19 Jul 00 Posts: 3898 Credit: 1,158,042 RAC: 0 |
> If nothing is backlogged, SETI@home for BOINC works like this: > > 1; Immediately when 3 "success"-results is reported, the wu is marked as > "need_validate" by the Transitioner. > 2; If Validator finds a pair of results to be "similar enough", the wu passes > validation. > If can't find a pair, is marked "no consensus yet" and must wait for another > "success"-result being reported. > If found a pair, tries to validate all of the "success"-results. Any failing > validation is marked "invalid". > 3; The Validator counts up how many of the results passed validation in #2, > and awards credit after this formula: > a; If only 2 passed validation, grant lowest claimed to all. > b; If 3 or more, remove highest and lowest claimed and average the rest. > > 4; Any later reported results is tried validated, and if passes validation > gets the same credit as in #3. No re-calculation of crediting is done. > > > So, as long as nothing is backlogged, the normal in SETI@home is the 3 > "success"-results reported 1st decides the credit, and the middle claimed of > these 3 is the credit for all. > Only if something is backlogged, or if failed validation due to "no consensus > yet", is 4 or more results used when deciding crediting. > > In rare ocassions, one or more of the results is too different from the rest, > so 3a is used when deciding crediting in SETI@home. There are some minor "tweaks" in the works for handling the boundary condition when 0 credit is claimed, but, otherwise, will not (should not?) change this explanation ... I think I may steal this also to add to the "credit" definitions in the glossary ... |
©2024 University of California
SETI@home and Astropulse are funded by grants from the National Science Foundation, NASA, and donations from SETI@home volunteers. AstroPulse is funded in part by the NSF through grant AST-0307956.