I have the opposite problem. I don't know why I am crunching this wu when the two others have completed and been awarded credit. I've had this chap running for 50+ hours now? It is 97% complete.
I reckon one of the first two over-ran their deadline, triggering the re-send, but then returned it only a little bit late. If you hadn't started crunching, you might then have had a 'Redundant Result - cancelled by server' instruction (not sure if Einstein has implemented this yet), and you could have skipped it and moved on to the next one. But once you've started, BOINC lets you crunch on to the end, and claim credit - as I see you've now done.
Yes, it finished shortly after I noticed it. I had always thought a late result was timed out before the resend. Guess it just wasted a few cycles, sure I got credit, but it would have been better doing something fresh!
Wave upon wave of demented avengers march cheerfully out of obscurity into the dream.
a single user is NEVER assigned a WU as its own quorum partner.
It can happen if you have more than one PC attached that you're sent the same result twice, on different hosts. Or at least, that's an option that can be enabled by the project. Don't know if it's enabled or disabled here.
No idea, I can't remember it ever happening to me but that doesn't mean a thing. You would probably have to have as many boxes as Dr. Bruce Allen to have a halfway realistic chance of that event.
Actually, it is possible to have the same host get two results for the same WU. I had it happen once on Leiden when I first attached there.
However AFAIK, EAH doesn't permit that case, or allow two hosts from the same User ID to crunch any given WU. You can get different WU's from the same template frequency on multiple hosts on the same User ID though. I had that happen earlier this month.
My question is related to wingmen, but actually the opposite:
I have found in the past that certain other crunchers show up fairly often as my wingman. For awhile it was none other than Bruce Allen, but lately it has been one Anonymous. No complaints, but the way -- he's a better wingman, but I wonder why certain wingmen keep showing up on my results page. I thought wingmen were random. Any thoughts?
This is due to the fact that EAH doesn't just 'shotgun' the datapacks to any host which comes looking for work. They are initially sent to a limited set of hosts, and that set will work through all WU's for that datapack set. That's why you get teamed up with the same hosts a lot from time to time.
If the scheduler determines that adequate progress isn't being made on the datapack set, then it will issue them to new hosts to speed things up.
[…] I wonder why certain wingmen keep showing up on my results page. I thought wingmen were random. Any thoughts?
As you may have noticed, this project sends out data to be processed in the form of large files containing ‘fodder’ for numerous WUs; once a host has downloaded one, it’ll usually be assigned a batch of WUs that use the same data-file, so all that has to be downloaded is an ‘address list’ that tells the app where to look for each task. Since the servers try to minimize the number of hosts that get sent a given ‘master’ file, the pool of hosts crunching tasks from that batch will be quite small.
Since WUs from a particular data-file all have similar names (and credit value), you can identify batches in your results pretty easily. You should find that your frequent quorum partners are each associated with such a batch, because of the master file both your hosts have downloaded.
RE: I have the opposite
)
I reckon one of the first two over-ran their deadline, triggering the re-send, but then returned it only a little bit late. If you hadn't started crunching, you might then have had a 'Redundant Result - cancelled by server' instruction (not sure if Einstein has implemented this yet), and you could have skipped it and moved on to the next one. But once you've started, BOINC lets you crunch on to the end, and claim credit - as I see you've now done.
Yes, it finished shortly
)
Yes, it finished shortly after I noticed it. I had always thought a late result was timed out before the resend. Guess it just wasted a few cycles, sure I got credit, but it would have been better doing something fresh!
Wave upon wave of demented avengers march cheerfully out of obscurity into the dream.
RE: a single user is NEVER
)
It can happen if you have more than one PC attached that you're sent the same result twice, on different hosts. Or at least, that's an option that can be enabled by the project. Don't know if it's enabled or disabled here.
No idea, I can't remember it
)
No idea, I can't remember it ever happening to me but that doesn't mean a thing. You would probably have to have as many boxes as Dr. Bruce Allen to have a halfway realistic chance of that event.
Actually, it is possible to
)
Actually, it is possible to have the same host get two results for the same WU. I had it happen once on Leiden when I first attached there.
However AFAIK, EAH doesn't permit that case, or allow two hosts from the same User ID to crunch any given WU. You can get different WU's from the same template frequency on multiple hosts on the same User ID though. I had that happen earlier this month.
Alinator
My question is related to
)
My question is related to wingmen, but actually the opposite:
I have found in the past that certain other crunchers show up fairly often as my wingman. For awhile it was none other than Bruce Allen, but lately it has been one Anonymous. No complaints, but the way -- he's a better wingman, but I wonder why certain wingmen keep showing up on my results page. I thought wingmen were random. Any thoughts?
(Click for detailed stats)
This is due to the fact that
)
This is due to the fact that EAH doesn't just 'shotgun' the datapacks to any host which comes looking for work. They are initially sent to a limited set of hosts, and that set will work through all WU's for that datapack set. That's why you get teamed up with the same hosts a lot from time to time.
If the scheduler determines that adequate progress isn't being made on the datapack set, then it will issue them to new hosts to speed things up.
HTH,
Alinator
RE: […] I wonder why
)
As you may have noticed, this project sends out data to be processed in the form of large files containing ‘fodder’ for numerous WUs; once a host has downloaded one, it’ll usually be assigned a batch of WUs that use the same data-file, so all that has to be downloaded is an ‘address list’ that tells the app where to look for each task. Since the servers try to minimize the number of hosts that get sent a given ‘master’ file, the pool of hosts crunching tasks from that batch will be quite small.
Since WUs from a particular data-file all have similar names (and credit value), you can identify batches in your results pretty easily. You should find that your frequent quorum partners are each associated with such a batch, because of the master file both your hosts have downloaded.