AstroPulse errors - Reporting

Message boards : Number crunching : AstroPulse errors - Reporting
Message board moderation

To post messages, you must log in.

Previous · 1 . . . 10 · 11 · 12 · 13 · 14 · Next

AuthorMessage
Profile Gustav_and_Padma
Avatar

Send message
Joined: 26 Oct 03
Posts: 16
Credit: 315,654
RAC: 0
United States
Message 832215 - Posted: 19 Nov 2008, 21:27:33 UTC - in response to Message 830174.  

I imagine a similar answer would apply for this one:
http://setiathome.berkeley.edu/workunit.php?wuid=359221696

Right now it looks like we are waiting for a third client to report back. I'm sure you have answeed this question many times before. What I am wondering about, in addition to (1) Who will get how much credit (right now it is zero - not pending - for the first two of us who completed crunching the WU), is (2) Why is there a discrepancy in the first place? Thanks, Padma
ID: 832215 · Report as offensive
Profile Byron S Goodgame
Volunteer tester
Avatar

Send message
Joined: 16 Jan 06
Posts: 1145
Credit: 3,936,993
RAC: 0
United States
Message 832224 - Posted: 19 Nov 2008, 22:07:15 UTC - in response to Message 832215.  
Last modified: 19 Nov 2008, 22:16:05 UTC

I imagine a similar answer would apply for this one:
http://setiathome.berkeley.edu/workunit.php?wuid=359221696

Right now it looks like we are waiting for a third client to report back. I'm sure you have answeed this question many times before. What I am wondering about, in addition to (1) Who will get how much credit (right now it is zero - not pending - for the first two of us who completed crunching the WU), is (2) Why is there a discrepancy in the first place? Thanks, Padma

Found 30 single pulses and 30 repeating pulses, exiting.
Was found on your wingmans WU and not yours, so because of the difference the thrid wingman was sent out. In this instance it seems possible one of you won't get credit. Your version of BOINC could use some updating BTW. As far as who gets how much credit, it depends on who matches the third wingman.
ID: 832224 · Report as offensive
Profile Gustav_and_Padma
Avatar

Send message
Joined: 26 Oct 03
Posts: 16
Credit: 315,654
RAC: 0
United States
Message 832253 - Posted: 20 Nov 2008, 0:19:00 UTC - in response to Message 832224.  

OK, thanks for noticing that. We were getting fun sounding results like the first wingman on this WU before we started having problems on our Vista machine. But everything has been working OK now for months. Or so we think. How do we update our version? Do we need updating on both machines?
ID: 832253 · Report as offensive
Profile Byron S Goodgame
Volunteer tester
Avatar

Send message
Joined: 16 Jan 06
Posts: 1145
Credit: 3,936,993
RAC: 0
United States
Message 832256 - Posted: 20 Nov 2008, 0:25:39 UTC - in response to Message 832253.  
Last modified: 20 Nov 2008, 0:30:17 UTC

OK, thanks for noticing that. We were getting fun sounding results like the first wingman on this WU before we started having problems on our Vista machine. But everything has been working OK now for months. Or so we think. How do we update our version? Do we need updating on both machines?


I think just the one with 5.10.30. Here's the BOINC update page. I'm using the 5.10.45 with Vista, which is the older version but still listed there in case you prefer not to use 6.2.19
ID: 832256 · Report as offensive
Profile Byron S Goodgame
Volunteer tester
Avatar

Send message
Joined: 16 Jan 06
Posts: 1145
Credit: 3,936,993
RAC: 0
United States
Message 832377 - Posted: 20 Nov 2008, 7:59:35 UTC - in response to Message 832224.  
Last modified: 20 Nov 2008, 8:26:46 UTC

Found 30 single pulses and 30 repeating pulses, exiting.
Was found on your wingmans WU and not yours, so because of the difference the thrid wingman was sent out. In this instance it seems possible one of you won't get credit.


Just to follow up that all three did recieve credit but also a question since it seems strange to me. The third wingman returned a result similar to Gustav & Padma. Both wingmen have XP and version 5.10.45. Gustav & Padma has Vista and 5.10.30. Why did the wingman that has a result with "Found 30 single pulses and 30 repeating pulses" recieve credit if it didn't match the other two? I think it's great all three got credit but I don't understand why. I thought I saw a similar situation in the threads somewhere, but I can't find the final result of what happened there or why.

I'm guessing "Found 30 single pulses and 30 repeating pulses" isn't enough in itself to disqualify a result, but is enough to send out another wingman on the task, if the second wingman doesn't return the same result.
ID: 832377 · Report as offensive
Josef W. Segur
Volunteer developer
Volunteer tester

Send message
Joined: 30 Oct 99
Posts: 4504
Credit: 1,414,761
RAC: 0
United States
Message 832488 - Posted: 20 Nov 2008, 18:13:09 UTC - in response to Message 832377.  

...
I'm guessing "Found 30 single pulses and 30 repeating pulses" isn't enough in itself to disqualify a result, but is enough to send out another wingman on the task, if the second wingman doesn't return the same result.

It is actually possible, though unlikely, that a result which exited early will be strongly similar to one which didn't, and not even need a third opinion. The validator makes allowances for the minor differences which can be expected from apps for different platforms compiled with different compilers, one of those allowances is that only signals which are at least 1% above threshold are compared. If the wingman's host only found 29 of one type so had to run the full set of calculations, there's a reasonable chance the different one was less than 1% above threshold.

For the weakly similar match needed for the early exit to be granted credit when a third host has been called in, more than half the signals need to match and that's fairly likely. Most of the early exits for AP seem to run a considerable time, perhaps indicating a gradual accumulation of signals which are only slightly above threshold.
                                                               Joe
ID: 832488 · Report as offensive
Profile Byron S Goodgame
Volunteer tester
Avatar

Send message
Joined: 16 Jan 06
Posts: 1145
Credit: 3,936,993
RAC: 0
United States
Message 832563 - Posted: 20 Nov 2008, 22:18:36 UTC - in response to Message 832488.  

It is actually possible, though unlikely, that a result which exited early will be strongly similar to one which didn't, and not even need a third opinion. The validator makes allowances for the minor differences which can be expected from apps for different platforms compiled with different compilers, one of those allowances is that only signals which are at least 1% above threshold are compared. If the wingman's host only found 29 of one type so had to run the full set of calculations, there's a reasonable chance the different one was less than 1% above threshold.

For the weakly similar match needed for the early exit to be granted credit when a third host has been called in, more than half the signals need to match and that's fairly likely. Most of the early exits for AP seem to run a considerable time, perhaps indicating a gradual accumulation of signals which are only slightly above threshold.
                                                               Joe


TY Josef

ID: 832563 · Report as offensive
michael
Volunteer tester

Send message
Joined: 19 Sep 03
Posts: 21
Credit: 183,890
RAC: 0
United Kingdom
Message 832568 - Posted: 20 Nov 2008, 22:35:19 UTC
Last modified: 20 Nov 2008, 22:36:04 UTC

I have just downloaded An AP 5.00 WU which is running along side three AP 4.36 will this cause any problems or just let them run.

Michael.
ID: 832568 · Report as offensive
Josef W. Segur
Volunteer developer
Volunteer tester

Send message
Joined: 30 Oct 99
Posts: 4504
Credit: 1,414,761
RAC: 0
United States
Message 832574 - Posted: 20 Nov 2008, 22:56:11 UTC - in response to Message 832568.  

I have just downloaded An AP 5.00 WU which is running along side three AP 4.36 will this cause any problems or just let them run.

Michael.

That's how it should be on a quad, let them run. The radar removal which 5.00 does shouldn't even slow it down significantly.
                                                                Joe
ID: 832574 · Report as offensive
Profile Gustav_and_Padma
Avatar

Send message
Joined: 26 Oct 03
Posts: 16
Credit: 315,654
RAC: 0
United States
Message 834019 - Posted: 24 Nov 2008, 19:43:24 UTC - in response to Message 832488.  

So is the number 30 some kind of threshhold for pulses and reapeating pulses? For those of us who don't understand the science, it would be interesting to know, hence motivating for us to dedicate our computers to crunching this data. The question really is, why don't the AP results show us in a few characters how many pulses and repeating pulses our computers found per WU, even though they might be twenty-something or less -- perhaps not 'exciting' per se, but maybe worth a mention of 'moderately interesting', 'mildly remarkable', or some other message decipherable with average human skills? -smiles- Gustav
ID: 834019 · Report as offensive
Richard Haselgrove Project Donor
Volunteer tester

Send message
Joined: 4 Jul 99
Posts: 14659
Credit: 200,643,578
RAC: 874
United Kingdom
Message 834023 - Posted: 24 Nov 2008, 19:57:38 UTC - in response to Message 834019.  

The question really is, why don't the AP results show us in a few characters how many pulses and repeating pulses our computers found per WU...

As it happens, this is one of the unsung improvements in the new second-generation optimised AP application:

single pulses: 4
repetitive pulses: 30

By all means feel free to ask the Berkeley developers to include similar information in the stock application.
ID: 834023 · Report as offensive
Chilcotin

Send message
Joined: 1 Jun 99
Posts: 4
Credit: 3,281,561
RAC: 1
Canada
Message 835032 - Posted: 28 Nov 2008, 3:46:07 UTC

Task completed successfully:

http://setiathome.berkeley.edu/workunit.php?wuid=365752248

but zero credit assigned for me or my wingman.
ID: 835032 · Report as offensive
Profile Leaps-from-Shadows
Volunteer tester
Avatar

Send message
Joined: 11 Aug 08
Posts: 323
Credit: 259,220
RAC: 0
United States
Message 835036 - Posted: 28 Nov 2008, 3:52:10 UTC - in response to Message 835032.  

Task completed successfully:

http://setiathome.berkeley.edu/workunit.php?wuid=365752248

but zero credit assigned for me or my wingman.

If you look at the page you posted, it shows that the work unit has been sent to another computer to crunch. Evidently the two computers that have completed it already didn't agree on the results.

Eventually, those computers that completed the unit with correct results will get credit for it.
Cruiser
Gateway GT5692 L-f-S Edition
-Phenom X4 9650 CPU
-4GB 667MHz DDR2 RAM
-500GB SATA HD
-Vista x64 SP1
-BOINC 6.2.19 32-bit client
-SSE3 optimized 32-bit apps
ID: 835036 · Report as offensive
Chilcotin

Send message
Joined: 1 Jun 99
Posts: 4
Credit: 3,281,561
RAC: 1
Canada
Message 835161 - Posted: 28 Nov 2008, 14:28:29 UTC - in response to Message 835036.  

Thank you.

An error or status message to that effect would be helpful. Did I miss one ?
ID: 835161 · Report as offensive
samuel7
Volunteer tester

Send message
Joined: 2 Jan 00
Posts: 47
Credit: 2,194,240
RAC: 0
Finland
Message 835168 - Posted: 28 Nov 2008, 14:44:16 UTC - in response to Message 835032.  

Task completed successfully:

http://setiathome.berkeley.edu/workunit.php?wuid=365752248

but zero credit assigned for me or my wingman.

Made the link clickable.

This one is interesting (and potentially frustrating) since two v4 results didn't agree and the tiebreaker will likely be crunched with v5.00. Could well be that yet another tiebreaker is needed and then the v4 results will be deemed invalid. Hope not.

And as L-f-S said, in Astropulse the only clue is the reissue. Validate state should be 'Checked, but no consensus yet,' but it's broken in Astropulse and says 'Valid'.
ID: 835168 · Report as offensive
Profile tullio
Volunteer tester

Send message
Joined: 9 Apr 04
Posts: 8797
Credit: 2,930,782
RAC: 1
Italy
Message 835172 - Posted: 28 Nov 2008, 15:11:15 UTC - in response to Message 835168.  

Judging from the stderr.txt one of the results was obtained with a stock Windows app and a second with an optimized Windows app. Then it is not true that only Linux optimized apps. like the one I am using, need a third wingman to be validated.
Tullio
ID: 835172 · Report as offensive
Profile Leaps-from-Shadows
Volunteer tester
Avatar

Send message
Joined: 11 Aug 08
Posts: 323
Credit: 259,220
RAC: 0
United States
Message 835181 - Posted: 28 Nov 2008, 16:39:32 UTC - in response to Message 835172.  

Then it is not true that only Linux optimized apps. like the one I am using, need a third wingman to be validated.

That's never been true. Some people (like me) are lucky and rarely have to wait for a third cruncher to validate.

Others seem to always be paired with a computer that doesn't agree with their result. It could be different version numbers or one gets an error and the other doesn't, or other factors. We don't really know every reason why, it just happens.
Cruiser
Gateway GT5692 L-f-S Edition
-Phenom X4 9650 CPU
-4GB 667MHz DDR2 RAM
-500GB SATA HD
-Vista x64 SP1
-BOINC 6.2.19 32-bit client
-SSE3 optimized 32-bit apps
ID: 835181 · Report as offensive
geoff

Send message
Joined: 25 Apr 00
Posts: 123
Credit: 34,100,351
RAC: 18
United Kingdom
Message 835193 - Posted: 28 Nov 2008, 17:40:37 UTC
Last modified: 28 Nov 2008, 17:42:29 UTC

Checking up on the status of my AP work units I have 8 pending crunched with 4.36 waiting for crunching by wingmen by possibly v5.0.

3 crunched AP wus have received 0 and have been issued to a 3rd wingman maybe using v5.0.

I have waiting in cache 8 AP work units assigned to use 4.36 and all these have already been crunched by wingmen using v5.0. Is it possible to change these wus to be crunched by v5.0 or do I abort them so they are reissued? I have suspended these for the present time.

I have also in cache 19 AP wus waiting to be crunched by v5.0
ID: 835193 · Report as offensive
Profile jason_gee
Volunteer developer
Volunteer tester
Avatar

Send message
Joined: 24 Nov 06
Posts: 7489
Credit: 91,093,184
RAC: 0
Australia
Message 835199 - Posted: 28 Nov 2008, 17:58:17 UTC - in response to Message 835193.  

Hi geoff,

If these are not started yet, you can alter your app_info.xml to make them use the version 5 app instead, just copy the v5 filename in place of the earlier versions filename.

Be warned doing this with a WU in progress (even partially completed but suspended) will dump the WU.

Jason

"Living by the wisdom of computer science doesn't sound so bad after all. And unlike most advice, it's backed up by proofs." -- Algorithms to live by: The computer science of human decisions.
ID: 835199 · Report as offensive
geoff

Send message
Joined: 25 Apr 00
Posts: 123
Credit: 34,100,351
RAC: 18
United Kingdom
Message 835213 - Posted: 28 Nov 2008, 19:19:04 UTC

Hi Jason

Many thanks, I thought it should be possible, I have 3 AP wus unfinished so will let them complete with probably 0 credit, then change the app_info to crunch 5 remaining wus that are suspended.

Geoff
ID: 835213 · Report as offensive
Previous · 1 . . . 10 · 11 · 12 · 13 · 14 · Next

Message boards : Number crunching : AstroPulse errors - Reporting


 
©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.