Message boards :
Number crunching :
How many clients are needed.....
Message board moderation
Author | Message |
---|---|
ziggie Send message Joined: 17 Apr 05 Posts: 7 Credit: 279,455 RAC: 0 |
How many clients are needed to return results (single WU) for it to be validated and credits allocated? I have had several 0.0 credit returns and checking those WU's it seems to be if less than 4 results are returned. Looking through some recent pending WU's it looks as though there are going to be a lot of 0.0 credits in the near future.... as most clients working on them had failed downloads. |
mikey Send message Joined: 17 Dec 99 Posts: 4215 Credit: 3,474,603 RAC: 0 |
How many clients are needed to return results (single WU) for it to be validated and credits allocated? At least 3 under the current settings. Berkeley has the ability to adjust it thru programming, but 3 is the number they have settled on. IF the validator gets to the unit when only 3 results have been returned then the top and bottom numbers are thrown out and the middle number is given to all 3 people. If 4 numbers are returned then the top and bottom are still thrown out and the middle 2 are averaged. I do not know what happens if more than 4 come in before validation occurs. If more results come in AFTER validation has occurred then those people are granted the number already given for that unit. If 3 numbers are not reutrned within the 2 week time limit than the unit is sent to mroe people unitl at least 3 results are returned. ONE MORE THING..if the numbers are NOT within a "statistical range", determined by Berkeley and NOT publicized, then the unit is not considered ready to be validated. An example would be computer #1 and 2 returns numbers of 23.54 and 26.75 but computer #3 asks for 56.98 then #3 is out of the 'range" and will not be counted but will be given credit based upon what other numbers come in from other computers. Each unit is initially sent to 4 computers, then after 2 weeks is sent out again, and the process repeats up to 7 times, I think that is the number programmed right now, and then the unit is deemed non crunchable and flagged for berkeley to work on. |
KWSN - MajorKong Send message Joined: 5 Jan 00 Posts: 2892 Credit: 1,499,890 RAC: 0 |
How many clients are needed to return results (single WU) for it to be validated and credits allocated? Sorry to see this Ziggie. It looks like you have been bit by the 'download error' bug on some of your work units. When this happens, it is almost always during an outage at berkeley, and the vast majority of the time, it is caused by some users still using v4.13 (or older) of the BOINC core client. The download bug has been fixed in newer versions, starting at least at v4.19. It is sad that you have been screwed out of getting credit for work you have done by the v4.13 crowd. I have been screwed out of a few work units myself for this reason, as have many other people. This situtation will likely last until Berkeley forces users to use at least a certain BOINC version. This bug got fixed months ago in newer BOINC versions. Getting people to actually USE the newer versions is another matter entirely. https://youtu.be/iY57ErBkFFE #Texit Don't blame me, I voted for Johnson(L) in 2016. Truth is dangerous... especially when it challenges those in power. |
ziggie Send message Joined: 17 Apr 05 Posts: 7 Credit: 279,455 RAC: 0 |
thanxz mikey. they musta upped it to 4 somewhere because I have several 0.0 returns where the results have been returned sucessfully by 3 clients. |
ziggie Send message Joined: 17 Apr 05 Posts: 7 Credit: 279,455 RAC: 0 |
thanxz KWSN. I guess we just have to live with it. It will probably get back to normal again now that u/l d/l seem to be back to normal. |
Jim Baize Send message Joined: 6 May 00 Posts: 758 Credit: 149,536 RAC: 0 |
They send out 4 WU's but only 3 has to be returned valid. If all 4 are returned within the deadline then all 4 get credit. Jim thanxz mikey. they musta upped it to 4 somewhere because I have several 0.0 returns where the results have been returned sucessfully by 3 clients. |
W-K 666 Send message Joined: 18 May 99 Posts: 19062 Credit: 40,757,560 RAC: 67 |
To clear up a few things:
|
Tony D Send message Joined: 20 Mar 04 Posts: 73 Credit: 330,114 RAC: 0 |
To clear up a few things: ...But not always, Andy, as this unit shows... http://setiathome.berkeley.edu/workunit.php?wuid=20833149 I may be getting older but I refuse to grow up! |
Paul D. Buck Send message Joined: 19 Jul 00 Posts: 3898 Credit: 1,158,042 RAC: 0 |
There are a bunch of limits on when to issue and/or not issue work. These are set by the project and we have a table of them (to the extent we know them) in the Wiki ... |
W-K 666 Send message Joined: 18 May 99 Posts: 19062 Credit: 40,757,560 RAC: 67 |
...But not always, Andy, as this unit shows... I was going by this one. Andy |
ziggie Send message Joined: 17 Apr 05 Posts: 7 Credit: 279,455 RAC: 0 |
Andy[/quote] ...But not always, Andy, as this unit shows... http://setiathome.berkeley.edu/workunit.php?wuid=20833149 [/quote] Tony D: that is exactly what I am getting. 3 results returned sucessfully but 0.0 credit granted. here are 2 of my personal examples: http://setiweb.ssl.berkeley.edu/workunit.php?wuid=20944227 http://setiweb.ssl.berkeley.edu/workunit.php?wuid=20638514 |
barbarossa Send message Joined: 4 Sep 99 Posts: 1294 Credit: 6,629,998 RAC: 3 |
Same with me. I've got quite a number of these. They all have something in common: One of the three claims is significantly higher or lower than the other two. This could result in the result being ruled out by the system (I read something of that kind today in a thread). And then there's but two remaining which makes no quorum. Sigh... back to crunching. The day comes when all will be well. :-)= Greybeard All about BOINC: BOINC-Wiki (by Paul D. Buck) |
John McLeod VII Send message Joined: 15 Jul 99 Posts: 24806 Credit: 790,712 RAC: 0 |
ONE MORE THING..if the numbers are NOT within a "statistical range", determined by Berkeley and NOT publicized, then the unit is not considered ready to be validated. An example would be computer #1 and 2 returns numbers of 23.54 and 26.75 but computer #3 asks for 56.98 then #3 is out of the 'range" and will not be counted but will be given credit based upon what other numbers come in from other computers. Each unit is initially sent to 4 computers, then after 2 weeks is sent out again, and the process repeats up to 7 times, I think that is the number programmed right now, and then the unit is deemed non crunchable and flagged for berkeley to work on. It is the validity of the science that is being checked, not the credit request. The credit requests can be 0.00, 95.7 and 150.3, and if the science is valid, everyone gets 95.7. On the other hand all can request 25.5 credits, and if the science does not validate, no credit will be granted - more WUs will be sent out. BOINC WIKI |
W-K 666 Send message Joined: 18 May 99 Posts: 19062 Credit: 40,757,560 RAC: 67 |
Just so people can see what is going on. I had a unit, see previous post, that had six d/l errors but gave credit other people were not so lucky. So here is a unit that again has 6 d/l errors, the too many errors indicator, but with 3 sucessful returns also says pending. This One . Andy |
Tony D Send message Joined: 20 Mar 04 Posts: 73 Credit: 330,114 RAC: 0 |
Just so people can see what is going on. I had a unit, see previous post, that had six d/l errors but gave credit other people were not so lucky. So here is a unit that again has 6 d/l errors, the too many errors indicator, but with 3 sucessful returns also says pending. This One . That one now says 0.00 credit. There does not seem to be a pattern. Or am I missing something? |
W-K 666 Send message Joined: 18 May 99 Posts: 19062 Credit: 40,757,560 RAC: 67 |
Before my result arrive at berkeley it just said, Too Many Errors, and granted credit 0.0. After the line below err msg changed to pending. If you look at earlier posts, just after the outage about 1st july, I had similar unit which eventually was granted credit, which I hope will happen this time. But several other people have said it their experience they, in a similar situation, examples given, they receive the big fat '0'. Andy |
W-K 666 Send message Joined: 18 May 99 Posts: 19062 Credit: 40,757,560 RAC: 67 |
My hopes were justified, the unit has now been granted credit. But it does still say Too many errors. Andy |
mikey Send message Joined: 17 Dec 99 Posts: 4215 Credit: 3,474,603 RAC: 0 |
edited |
mikey Send message Joined: 17 Dec 99 Posts: 4215 Credit: 3,474,603 RAC: 0 |
ONE MORE THING..if the numbers are NOT within a "statistical range", determined by Berkeley and NOT publicized, then the unit is not considered ready to be validated. An example would be computer #1 and 2 returns numbers of 23.54 and 26.75 but computer #3 asks for 56.98 then #3 is out of the 'range" and will not be counted but will be given credit based upon what other numbers come in from other computers. Each unit is initially sent to 4 computers, then after 2 weeks is sent out again, and the process repeats up to 7 times, I think that is the number programmed right now, and then the unit is deemed non crunchable and flagged for berkeley to work on. How do you do that then? I thought the way the "good Science" was determined was by different computers coming up with similar results for the same workunit. |
Jim Baize Send message Joined: 6 May 00 Posts: 758 Credit: 149,536 RAC: 0 |
How do you do that then? I thought the way the "good Science" was determined was by different computers coming up with similar results for the same workunit. I'm sure there is a really good explaination of this on the BOINC Wiki |
©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.