threefold validation

Message boards : Number crunching : threefold validation
Message board moderation

To post messages, you must log in.

AuthorMessage
atlov

Send message
Joined: 11 Aug 12
Posts: 35
Credit: 32,718,664
RAC: 34
Germany
Message 1699540 - Posted: 8 Jul 2015, 16:04:58 UTC

Hi folks!

I noticed WU 1833335409 received 3 valid results, although the two hosts 7261736 and 6782743 supplied results within their deadlines, so at first glance I don't see a reason why to send the WU out to a third host. I'm not quite sure, but this WU could have been in my validation inconclusive folder for some time, so maybe the first two results differed above some numerical threshold and the third result validated both by being inbetween the other two? Any comments on this?
ID: 1699540 · Report as offensive
Rasputin42
Volunteer tester

Send message
Joined: 25 Jul 08
Posts: 412
Credit: 5,834,661
RAC: 0
United States
Message 1699543 - Posted: 8 Jul 2015, 16:14:55 UTC

I have something similar.
It is marked validated in my account, but a third user is still processing it.
WHY?
WUhttp://setiathome.berkeley.edu/workunit.php?wuid=1761966073
ID: 1699543 · Report as offensive
Profile Donald L. Johnson
Avatar

Send message
Joined: 5 Aug 02
Posts: 8240
Credit: 14,654,533
RAC: 20
United States
Message 1699544 - Posted: 8 Jul 2015, 16:15:20 UTC - in response to Message 1699540.  

Hi folks!

I noticed WU 1833335409 received 3 valid results, although the two hosts 7261736 and 6782743 supplied results within their deadlines, so at first glance I don't see a reason why to send the WU out to a third host. I'm not quite sure, but this WU could have been in my validation inconclusive folder for some time, so maybe the first two results differed above some numerical threshold and the third result validated both by being inbetween the other two? Any comments on this?

Yes, that appears to be the case. While all 3 valid results had 30 and 30, the lists of signals/strengths are different for the first two, so the WU was ruled "Inconclusive", and sent out for a 3rd. The 3rd errored, and so generated a 4th. The 4th appears to have matched the 1st result, so the 1st became the canonical result entered into the AP science database, but the 2nd was within the tolerances for "weakly similar", so all 3 got credit.
Donald
Infernal Optimist / Submariner, retired
ID: 1699544 · Report as offensive
atlov

Send message
Joined: 11 Aug 12
Posts: 35
Credit: 32,718,664
RAC: 34
Germany
Message 1699546 - Posted: 8 Jul 2015, 16:17:51 UTC - in response to Message 1699543.  

Host 7027212 returned its result after the deadline, so the WU was sent out to the third host.
ID: 1699546 · Report as offensive
Rasputin42
Volunteer tester

Send message
Joined: 25 Jul 08
Posts: 412
Credit: 5,834,661
RAC: 0
United States
Message 1699548 - Posted: 8 Jul 2015, 16:20:49 UTC

A computer only knows ones and zeros.
How can there be ANY difference in the results at all?(unless there is a computing error)
ID: 1699548 · Report as offensive
Profile Donald L. Johnson
Avatar

Send message
Joined: 5 Aug 02
Posts: 8240
Credit: 14,654,533
RAC: 20
United States
Message 1699549 - Posted: 8 Jul 2015, 16:21:11 UTC - in response to Message 1699543.  
Last modified: 8 Jul 2015, 16:43:05 UTC

I have something similar.
It is marked validated in my account, but a third user is still processing it.
WHY?
WUhttp://setiathome.berkeley.edu/workunit.php?wuid=1761966073

See my response to your previous question about this Workunit in this thread.
Donald
Infernal Optimist / Submariner, retired
ID: 1699549 · Report as offensive
Rasputin42
Volunteer tester

Send message
Joined: 25 Jul 08
Posts: 412
Credit: 5,834,661
RAC: 0
United States
Message 1699550 - Posted: 8 Jul 2015, 16:28:29 UTC

1st: Why did the user get credit, if the result was submitted after the deadline.

2nd: If a result is validated, why is any ongoing calculation of this task not canceled? It is already validated, so why crunch it again?
ID: 1699550 · Report as offensive
Profile Donald L. Johnson
Avatar

Send message
Joined: 5 Aug 02
Posts: 8240
Credit: 14,654,533
RAC: 20
United States
Message 1699551 - Posted: 8 Jul 2015, 16:29:08 UTC - in response to Message 1699548.  

A computer only knows ones and zeros.
How can there be ANY difference in the results at all?(unless there is a computing error)

The applications written to process Seti@Home Tasks on different platforms (CPU, GPU, different Operating systems) have some differences based on hardware and OS characteristics, and so sometimes produce slightly different results for the same Task/data set. Jason Gee and Joe Segur have written much better explanations than I can give as to how and why this is. We seem to see it most often when a CPU is paired with a GPU, or when one of the Hosts runs a stock application while the other runs Anonymous Platform. It's just how it is.
Donald
Infernal Optimist / Submariner, retired
ID: 1699551 · Report as offensive
Profile HAL9000
Volunteer tester
Avatar

Send message
Joined: 11 Sep 99
Posts: 6534
Credit: 196,805,888
RAC: 57
United States
Message 1699553 - Posted: 8 Jul 2015, 16:29:28 UTC - in response to Message 1699540.  
Last modified: 8 Jul 2015, 16:31:10 UTC

Hi folks!

I noticed WU 1833335409 received 3 valid results, although the two hosts 7261736 and 6782743 supplied results within their deadlines, so at first glance I don't see a reason why to send the WU out to a third host. I'm not quite sure, but this WU could have been in my validation inconclusive folder for some time, so maybe the first two results differed above some numerical threshold and the third result validated both by being inbetween the other two? Any comments on this?

It is likely that the first two failed to validate because they were not "strongly similar", but with the third result they all validated due to being "weakly similar".

With the various apps/hardware we do get different rounding & precision values. So this tends to occur from time to time.
AstroPulse v7 Windows x86 rev 2721 Nvidia
AstroPulse v7 Windows x86 rev 2737 Nvidia
AstroPulse v7 Windows x64 rev 2692 CPU
SETI@home classic workunits: 93,865 CPU time: 863,447 hours
Join the [url=http://tinyurl.com/8y46zvu]BP6/VP6 User Group[
ID: 1699553 · Report as offensive
Profile Donald L. Johnson
Avatar

Send message
Joined: 5 Aug 02
Posts: 8240
Credit: 14,654,533
RAC: 20
United States
Message 1699554 - Posted: 8 Jul 2015, 16:36:26 UTC - in response to Message 1699550.  
Last modified: 8 Jul 2015, 16:41:43 UTC

1st: Why did the user get credit, if the result was submitted after the deadline.

I answered that question in the previous thread.

2nd: If a result is validated, why is any ongoing calculation of this task not canceled? It is already validated, so why crunch it again?

Probably because the folks who wrote the server software decided that it was not worth the effort to add the extra code (and server processing time/database searches) to send out a notice and cancel the 3rd (or 4th, or....) Task. And other than it takes longer for you to see the Task drop off the BOINC database and your Tasks page, what harm does it do? If this sort of thing happened a LOT, it might make sense, but apparently it only happens once in a blue moon......
Donald
Infernal Optimist / Submariner, retired
ID: 1699554 · Report as offensive
Profile Mike Special Project $75 donor
Volunteer tester
Avatar

Send message
Joined: 17 Feb 01
Posts: 34253
Credit: 79,922,639
RAC: 80
Germany
Message 1699555 - Posted: 8 Jul 2015, 16:40:53 UTC

Its a known issue that GPU apps act different than CPU app on 30/30 after signals are found.
Joe can explain better.


With each crime and every kindness we birth our future.
ID: 1699555 · Report as offensive

Message boards : Number crunching : threefold validation


 
©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.