Message boards :
Number crunching :
Some WU's give no credits?
Message board moderation
Author | Message |
---|---|
alphax Send message Joined: 17 May 99 Posts: 74 Credit: 1,266,810 RAC: 0 |
I was looking through the results for one of my workstations for kicks, and I noticed quite a few WUs that granted credits to none of the workstations that returned results, for example: http://setiweb.ssl.berkeley.edu/workunit.php?wuid=1273273 I don't really care about the credit itself, but I am curious about: 1) Why this occurs 2) What will happen to the WU in the future 3) Whether this will result in "holes" in the dataset that won't be analyzed, and therefore ET will slip through our search |
mikey Send message Joined: 17 Dec 99 Posts: 4215 Credit: 3,474,603 RAC: 0 |
Never read another message in the groups huh? Seriously, ALOT of messages have been devoted to your vary questions! There ARE NO real hard answers at this time. |
Paul D. Buck Send message Joined: 19 Jul 00 Posts: 3898 Credit: 1,158,042 RAC: 0 |
> I was looking through the results for one of my workstations for kicks, and I > noticed quite a few WUs that granted credits to none of the workstations that > returned results, for example: > > http://setiweb.ssl.berkeley.edu/workunit.php?wuid=1273273 > > I don't really care about the credit itself, but I am curious about: > > 1) Why this occurs > 2) What will happen to the WU in the future > 3) Whether this will result in "holes" in the dataset that won't be analyzed, > and therefore ET will slip through our search I talk about some of these issues in the documentation ... you can look under credit in the Glossary and follow the links around. I also talk about some of the exact situations in the BOINC Web Site Owner's Manual too ... In general, though, work should be issued until there are 3 results that are the "same" (what that means is an open question and we do not have a complete or detailed answer, nor do we really need one). This constitutes a quorum of results. With this quorum established the work unit is "retired" and is not issued any more. Now, there are work units that have something wrong with them and will seemingly not give any credit or will always be involved as an error. The exact boundaries of what that means is also semi-undefined. But, in general, we have work that creates a "valid" result and will allow the granting of credit, others that are so bad that no usable science is available within the result (for example due to RFI - it is defined in the glossary) and will not result in the granting of credit. Other work units are just plain "bad" and cannot be processed. However, the specific example you gave, you can see that the work unit has been issued a number of times with 5 never being returned. Of the three returned results we can infer from the wide spread of credit claimed that the three results could not form a quorum of results because they are not the "same". If you look at the CPU types of the 3 machines you can also see that we have one as an AMD, one unknown type pentium, and a pentium. One more inference/guess is that the WU is of such a nature that it increased the effect of the FPU so that the returned values just don't line up ... I talk about this effect in the bit about Floating Point Numbers (Glossary again ... sorry for the shameless plugs ...) If the quorum cannot be established after a specificed mumber of issues the WU is no longer issued. WIth the data available to the project they can look at doing a number of things with WU of this kind ... issue them to identical type machines to see if that allows a quorum, process the WU on a lab machine to see the in process work, etc. I hope this helps ... |
Bill & Patsy Send message Joined: 6 Apr 01 Posts: 141 Credit: 508,875 RAC: 0 |
Hi, Paul! "... sorry for the shameless plugs ...)" No, not really. Nor should you be sorry! Please, no apologies!!! You are doing great work and great support for the project and its enthusiastic users. You've earned the right to plug!! I enjoy all your posts, and really appreciate all your support for everybody. Plug away!!! --Bill |
Errabee Send message Joined: 24 Jun 01 Posts: 11 Credit: 67,813 RAC: 0 |
> The exact boundaries of what that means is also semi-undefined. But, in > general, we have work that creates a "valid" result and will allow the > granting of credit, others that are so bad that no usable science is available > within the result (for example due to RFI - it is defined in the glossary) and > will not result in the granting of credit. > {Some other interesting stuff left out.} > If the quorum cannot be established after a specificed mumber of issues the WU > is no longer issued. WIth the data available to the project they can look at > doing a number of things with WU of this kind ... issue them to identical type > machines to see if that allows a quorum, process the WU on a lab machine to > see the in process work, etc. > > I hope this helps ... > The reason why no credit is given still eludes me. These results have been computed honestly, and if Berkeley can't do science with them, well that's a pity but not a reason to skip giving credits. In fact, as you put it, they DO some science with these wu's: they analyse why a quorum cannot be established, so that they can specify to which kind of machine these results should be sent in order to receive a scientifically meaningful result. Just my 2 cts. <img src="http://www.seti.nl/images/banners/7.gif" border="0"> <img border="0" src="http://boinc.mundayweb.com/one/stats.php?userID=560" /> |
Aardvark Send message Joined: 9 Sep 99 Posts: 44 Credit: 353,365 RAC: 0 |
> I was looking through the results for one of my workstations for kicks, and I > noticed quite a few WUs that granted credits to none of the workstations that > returned results, for example: > > http://setiweb.ssl.berkeley.edu/workunit.php?wuid=1273273 > > I don't really care about the credit itself, but I am curious about: > > 1) Why this occurs > 2) What will happen to the WU in the future > 3) Whether this will result in "holes" in the dataset that won't be analyzed, > and therefore ET will slip through our search > > Or worse yet: Other people on same WU getting credit when you don't for instance http://setiweb.ssl.berkeley.edu/workunit.php?wuid=1273273 ... Mine is there at the bottom: claimed credit of 33.83, successful etc. Perhaps I completed it too quickly? -Aardvark |
RossM Send message Joined: 5 Apr 02 Posts: 37 Credit: 36,921 RAC: 0 |
From looking at my results it seems that there is a recurring theme regarding WU's which have been processed and completed by 3 different people then granted no credit. If one of those results has processed with a different Boinc client version it seems to have granted no credits. I don't know for sure if this is a problem just with some WU's or all of them. I suspect it's the former however. |
RossM Send message Joined: 5 Apr 02 Posts: 37 Credit: 36,921 RAC: 0 |
> Or worse yet: Other people on same WU getting credit when you don't for > instance http://setiweb.ssl.berkeley.edu/workunit.php?wuid=1273273 ... Mine is > there at the bottom: claimed credit of 33.83, successful etc. Perhaps I > completed it too quickly? > Nobody seemed have got credit for that WU. |
Benher Send message Joined: 25 Jul 99 Posts: 517 Credit: 465,152 RAC: 0 |
> > Nobody seemed have got credit for that WU. > If you see 3 valid results returned...and still no credit... then results haven't been checked against each other by validator yet. And like so many other WUs...eventually it will. You are correct. Nobody granted credit so far for this WU although 3 non-error results returned (as of this posting time). Reason: (see above) |
Heffed Send message Joined: 19 Mar 02 Posts: 1856 Credit: 40,736 RAC: 0 |
> If you see 3 valid results returned...and still no credit... > then results haven't been checked against each other by validator yet. > > And like so many other WUs...eventually it will. > > You are correct. Nobody granted credit so far for this WU although 3 non-error > results returned (as of this posting time). > Reason: (see above) Sorry to burst your bubble, but you are incorrect. If there is a number in the granted credit column, the WU has been validated and that is what you get. If the number is zero, then most the WU most likely has "check skipped" status. If the WU has not yet been validated, it will say "pending" in the granted credit column. |
Cary Stone Send message Joined: 11 Jan 02 Posts: 1 Credit: 337,452 RAC: 0 |
Just an observation. In looking at my own results, it appears that any WU with 8 or more result entries automatically gets 0 credit even if there are 3 successful results. In these cases, the Validate State is Check Skipped indicating that the results were not even sent to the Validator. |
Bruno G. Olsen & ESEA @ greenholt Send message Joined: 15 May 99 Posts: 875 Credit: 4,386,984 RAC: 0 |
|
Aardvark Send message Joined: 9 Sep 99 Posts: 44 Credit: 353,365 RAC: 0 |
> > > Or worse yet: Other people on same WU getting credit when you don't for > > instance http://setiweb.ssl.berkeley.edu/workunit.php?wuid=1273273 ... > Mine is > > there at the bottom: claimed credit of 33.83, successful etc. Perhaps I > > completed it too quickly? > > > > Nobody seemed have got credit for that WU. > > <img> src="http://boinc.mundayweb.com/seti2/stats.php?userID=651&trans=off"> > > Oops, I must have pasted the wrong address. I have a bunch for this machine which the others received credit but I didn't. Such as: http://setiweb.ssl.berkeley.edu/workunit.php?wuid=1980274 My total credit for the machine after 2.5 weeks of processing is 0.44. :( -Aardvark |
drinking12many Send message Joined: 27 Mar 01 Posts: 19 Credit: 7,591,427 RAC: 0 |
> > > >> > > > Oops, I must have pasted the wrong address. I have a bunch for this machine > which the others received credit but I didn't. Such as: > http://setiweb.ssl.berkeley.edu/workunit.php?wuid=1980274 > > My total credit for the machine after 2.5 weeks of processing is 0.44. :( >\ You will get credit for some eventually i went back to the 20th of sept and i still have some as pending i prob could have went back further and found more but by my 6000th result i was tired of hitting next 20 <img src=http://seti2.mundayweb.com/stats.php?userID=131&trans=off> |
Allan Taylor Send message Joined: 31 Jan 00 Posts: 32 Credit: 270,259 RAC: 0 |
If you look at the result of that work unit for your machine, there is an error associated with it. Your result comes back with invalid and the following error message: 4.06 SETI@Home Informational message -9 result_overflow NOTE: The number of results detected exceeds the storage space allocated. Maybe check the other results you are getting no credit for. I don't really know what the message means, but maybe someone else with more experience can tell you. |
Aardvark Send message Joined: 9 Sep 99 Posts: 44 Credit: 353,365 RAC: 0 |
> If you look at the result of that work unit for your machine, there is an > error associated with it. Your result comes back with invalid and the > following error message: > > 4.06 > > SETI@Home Informational message -9 result_overflow > NOTE: The number of results detected exceeds the storage space allocated. > > > > Maybe check the other results you are getting no credit for. I don't really > know what the message means, but maybe someone else with more experience can > tell you. > I have a bunch for that machine with the same problem. They all seem to be invalid. I'm using a G5 optimized version... perhaps whoever compiled it did it badly. What a bummer. Out of 500 WU only 1 has received credit. -Aardvark |
Ron Roe Send message Joined: 28 Feb 02 Posts: 156 Credit: 24,124 RAC: 0 |
> If you look at the result of that work unit for your machine, there is an > error associated with it. Your result comes back with invalid and the > following error message: > > 4.06 > > SETI@Home Informational message -9 result_overflow > NOTE: The number of results detected exceeds the storage space allocated. > > > > Maybe check the other results you are getting no credit for. I don't really > know what the message means, but maybe someone else with more experience can > tell you. > These type of msgs are for wu's that have too much RFI(radio frequency interference) in them and they exit within a very short time but do give you sometimes a tiny bit of credit. These are similar to the quickies on SETI Classic. |
Paul D. Buck Send message Joined: 19 Jul 00 Posts: 3898 Credit: 1,158,042 RAC: 0 |
> These type of msgs are for wu's that have too much RFI(radio frequency > interference) in them and they exit within a very short time but do give you > sometimes a tiny bit of credit. > > These are similar to the quickies on SETI Classic. And there are pictures in the Glossary ... And this time I am not sorry for the shameless plugs ... Actually, I do have some material ported from the old SETI@Home site with slight improvments to spelling and so forth that talk about these issues ... look in the SETI@Home area ... or follow the links... if you still cannot find them let me know and I will get you direct links ... but I think roaming around helps you see things that you might not otherwise see if I give you a direct link ... :) And Bill, Thanks! :) |
Allan Taylor Send message Joined: 31 Jan 00 Posts: 32 Credit: 270,259 RAC: 0 |
I'm not sure that this error would relate to RFI. The other two completed results were both validated and granted credit. Only the one result from Aardvark shows the error message. I would assume (and it is just an assumption) that if there was RFI that all the work units would come back as invalid. If he is having the same problem across multiple work units, wouldn't this mean the problem is more specific to his setup? |
Paul D. Buck Send message Joined: 19 Jul 00 Posts: 3898 Credit: 1,158,042 RAC: 0 |
> I'm not sure that this error would relate to RFI. The other two completed > results were both validated and granted credit. Only the one result from > Aardvark shows the error message. I would assume (and it is just an > assumption) that if there was RFI that all the work units would come back as > invalid. If he is having the same problem across multiple work units, wouldn't > this mean the problem is more specific to his setup? In general, the -9 error indicates that the work unit has a lot of noise ... and I will be honest that I did not look at the results... (bad Paul!) So you are right ... If the others were granted credit than the errors are a result of something else. In this case, my next guess would be problems with the computer. So, yes, in this case we seem to have another possible cause. Possibilities: 1) Overheating 2) Bad memory 3) aggressive over-clocking 4) bad luck 5) evil deamons ... By the way, I did find that some of the links from the RFI part in the glossary did not have all the best links so, I did add some to the pages and I am uploading them now. In essence, there are examples of what the displays look like, explanations on the processing, screen save and what all those wavy lines mean ... etc. So, the new pages should be available soon ... |
©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.