How long must I wait for validation for SETI tasks?

Message boards : Number crunching : How long must I wait for validation for SETI tasks?
Message board moderation

To post messages, you must log in.

1 · 2 · 3 · Next

AuthorMessage
Profile George Project Donor
Volunteer tester
Avatar

Send message
Joined: 23 Oct 17
Posts: 222
Credit: 2,597,521
RAC: 13
United States
Message 2014910 - Posted: 10 Oct 2019, 20:13:53 UTC

My computer is waiting for too long (in my eyes) to validate some SETI project tasks. I have 14 "validation pending" tasks, of which some go back into August. How long do I have to wait for validation? Is anyone else having this problem?
George

ID: 2014910 · Report as offensive
Profile Wiggo
Avatar

Send message
Joined: 24 Jan 00
Posts: 34744
Credit: 261,360,520
RAC: 489
Australia
Message 2014913 - Posted: 10 Oct 2019, 20:20:12 UTC

I've got 3330 of them going back to the beginning of July. ;-)

We crunch our work and then we have to wait for our wing people to do the rest.

Cheers.
ID: 2014913 · Report as offensive
Profile Mr. Kevvy Crowdfunding Project Donor*Special Project $250 donor
Volunteer moderator
Volunteer tester
Avatar

Send message
Joined: 15 May 99
Posts: 3776
Credit: 1,114,826,392
RAC: 3,319
Canada
Message 2014914 - Posted: 10 Oct 2019, 20:22:53 UTC - in response to Message 2014910.  
Last modified: 10 Oct 2019, 20:31:25 UTC

The only one from Aug. is this one which is going to time out and get reassigned in two days anyway if the wingperson doesn't complete it. Then the next wingie will have two more months... just something we have to live with.

Edit: 27,604 of them here. This is is why I don't even bother checking them anymore lol.
ID: 2014914 · Report as offensive
Profile George Project Donor
Volunteer tester
Avatar

Send message
Joined: 23 Oct 17
Posts: 222
Credit: 2,597,521
RAC: 13
United States
Message 2014919 - Posted: 10 Oct 2019, 20:47:53 UTC - in response to Message 2014913.  

...and I thought my 14 was bad! How come...? no... Why can't we ask...? no... It should be fair across the board. For me, Milkyway has ZERO pending tasks and Einstein has 22, of which only 8 are from Sept., the oldest is Sept. 17. This isn't fair at all! I understand they need time to verify, but more than a month seems too long. I wonder how long the other projects are taking for verification?
George

ID: 2014919 · Report as offensive
Profile Mr. Kevvy Crowdfunding Project Donor*Special Project $250 donor
Volunteer moderator
Volunteer tester
Avatar

Send message
Joined: 15 May 99
Posts: 3776
Credit: 1,114,826,392
RAC: 3,319
Canada
Message 2014920 - Posted: 10 Oct 2019, 20:53:04 UTC - in response to Message 2014919.  
Last modified: 10 Oct 2019, 20:53:57 UTC

One of the reasons the timeout is so long is that SETI@Home is trying to be as universally accessible of a project as possible, including for those with very slow or barely used computers (ie smartphones or systems that only do work when idle and are powered off when not used for a long time.) If I remember correctly, one of the project's directors/architects did indicate this directly.

I remember the old SETI@Home site when it first got underway that indicated "while you are getting coffee, your computer can be looking for signs of intelligent life..." It was designed and marketed for this casual of a participant.
ID: 2014920 · Report as offensive
Profile Keith Myers Special Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 29 Apr 01
Posts: 13161
Credit: 1,160,866,277
RAC: 1,873
United States
Message 2014921 - Posted: 10 Oct 2019, 20:54:49 UTC

If it bugs you so much about having too many validation waiting tasks, there are other projects that have much shorter deadlines, on the order of two weeks for MilkyWay and Einstein. But that does not guarantee your wingman will return their result in time and it would then go out again to another wingman. Depending on how many task errors are allowed, you might be waiting for as much as 10 months for example here at Seti before your task gets validated. You have to remember that task replication is to assure the validity of the science result.
Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)
ID: 2014921 · Report as offensive
Profile Keith Myers Special Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 29 Apr 01
Posts: 13161
Credit: 1,160,866,277
RAC: 1,873
United States
Message 2014922 - Posted: 10 Oct 2019, 21:02:39 UTC

Depends on the project whether they require task replication. MW does not as an example on some projects. If the MW scheduler has deemed your host of "high integrity" for returning valid science, the task is only sent once to your host so no wingman is required. That is not typical though. The other reason for low pending validations at MW is the fast throughput of tasks on their hosts which are predominately AMD cards. The typical AMD MW host can process its entire allotted cache in only an hour and then spend 10 minutes of dead time before getting more work. The task return rate and turnover rate is very fast at that project, so little likelihood of having many pendings build up.
Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)
ID: 2014922 · Report as offensive
Profile George Project Donor
Volunteer tester
Avatar

Send message
Joined: 23 Oct 17
Posts: 222
Credit: 2,597,521
RAC: 13
United States
Message 2014923 - Posted: 10 Oct 2019, 21:04:48 UTC - in response to Message 2014920.  

One of the reasons the timeout is so long is that SETI@Home is trying to be as universally accessible of a project as possible, including for those with very slow or barely used computers (ie smartphones or systems that only do work when idle and are powered off when not used for a long time.) If I remember correctly, one of the project's directors/architects did indicate this directly.

I remember the old SETI@Home site when it first got underway that indicated "while you are getting coffee, your computer can be looking for signs of intelligent life..." It was designed and marketed for this casual of a participant.

Wouldn't it be better if they prioritized the project's tasks reported by how many they have received from individual's computers, the more received the sooner they get validation? I realize that many of the casual users would become a bit disappointed and may drop out of the SETI project, but hey... wouldn't it end up becoming more efficient for the SETI project as a whole?
George

ID: 2014923 · Report as offensive
Profile George Project Donor
Volunteer tester
Avatar

Send message
Joined: 23 Oct 17
Posts: 222
Credit: 2,597,521
RAC: 13
United States
Message 2014925 - Posted: 10 Oct 2019, 21:19:17 UTC - in response to Message 2014921.  

If it bugs you so much about having too many validation waiting tasks, there are other projects that have much shorter deadlines, on the order of two weeks for MilkyWay and Einstein. But that does not guarantee your wingman will return their result in time and it would then go out again to another wingman. Depending on how many task errors are allowed, you might be waiting for as much as 10 months for example here at Seti before your task gets validated. You have to remember that task replication is to assure the validity of the science result.

Keith, it doesn't "bug" me but it does make me wonder... and I had not thought about the replication of tasks, the issues of tasks errors, etc. As I've said before... I have so much to learn. I'm still relatively new at this and I am a semi-casual, semi-serious user of SETI, et. al.

I am taking cues from you, Keith, and many others about the how's, why's and when's that involve SETI and other projects. And I must apologize for my impatient questions about the projects. It really doesn't bug me as much as it sounds when I post a question.
George

ID: 2014925 · Report as offensive
Profile Keith Myers Special Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 29 Apr 01
Posts: 13161
Credit: 1,160,866,277
RAC: 1,873
United States
Message 2014929 - Posted: 10 Oct 2019, 21:33:45 UTC

Little chance of your suggestions being implemented here at Seti. The main project scientist is set in his ways that Seti remain an "everyman" project configured to run on as many possible devices and with all levels of participation from extremely casual, once in a blue moon types to dedicated hyper crunchers. I know that is not going to change, I have no chance of persuading otherwise and just accept the status quo.
Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)
ID: 2014929 · Report as offensive
Profile George Project Donor
Volunteer tester
Avatar

Send message
Joined: 23 Oct 17
Posts: 222
Credit: 2,597,521
RAC: 13
United States
Message 2014931 - Posted: 10 Oct 2019, 22:03:01 UTC - in response to Message 2014929.  

Keith, just out of curiosity, nothing else, would you care to share how many and how long your validation pending is? You are in the top ten of SETI crunchers after all.
George

ID: 2014931 · Report as offensive
Profile Mr. Kevvy Crowdfunding Project Donor*Special Project $250 donor
Volunteer moderator
Volunteer tester
Avatar

Send message
Joined: 15 May 99
Posts: 3776
Credit: 1,114,826,392
RAC: 3,319
Canada
Message 2014944 - Posted: 11 Oct 2019, 0:42:14 UTC - in response to Message 2014923.  
Last modified: 11 Oct 2019, 0:45:54 UTC

Wouldn't it be better if they prioritized the project's tasks reported by how many they have received from individual's computers, the more received the sooner they get validation?


Validation occurs when the "wingperson" participant who receives the same work unit completes and returns it, and the result agrees. It is completely not under the control of the project, as they can take from a little as a minute to as much as two months to return it.

So what the project would have to do is try to send tasks from low-latency hosts only to other low-latency hosts for faster validation. This would be a high added level of complexity and processing at the project level that I think they would prefer to avoid. It wouldn't add anything for us... once enough Pendings have built up you reach a point where they are being validated as fast as you produce new Pendings, so your average daily production (RAC) stabilizes at the same rate as it would without Pendings anyways.
ID: 2014944 · Report as offensive
Profile Joseph Stateson Project Donor
Volunteer tester
Avatar

Send message
Joined: 27 May 99
Posts: 309
Credit: 70,759,933
RAC: 3
United States
Message 2014961 - Posted: 11 Oct 2019, 1:38:28 UTC - in response to Message 2014929.  

Little chance of your suggestions being implemented here at Seti. The main project scientist is set in his ways that Seti remain an "everyman" project configured to run on as many possible devices and with all levels of participation from extremely casual, once in a blue moon types to dedicated hyper crunchers. I know that is not going to change, I have no chance of persuading otherwise and just accept the status quo.


I had similar thoughts during the WOW event in August. I was wondering if the German organizers coordinated with the SETI folks to ensure minimum down time and faster turn around for the participants.

I am guessing that nothing special was done. Looking back at my pending tasks I see a lot of pending that I completed during the WOW event. Before that, the oldest I see is mid July. All the pending wingman ones I checked were CPU bound. Some wingman had timed out and the work unit had not yet been sent back out. Those were like three months out.

My guess, as far as the WOW event goes, that every participant has the same chance of getting a bummer wingman. Guessing also that the WOW score is based on credits earned during the WOW week even if the work had been done days before and submitted 1 second after the start. I actually noticed a huge jump by one setizen just after the start. I enjoyed participating and hope I reduced the backlog of workunits. It would be nice if it was not held during the hottest month of the year of the northern hemisphere.
ID: 2014961 · Report as offensive
Profile Keith Myers Special Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 29 Apr 01
Posts: 13161
Credit: 1,160,866,277
RAC: 1,873
United States
Message 2014973 - Posted: 11 Oct 2019, 2:04:33 UTC - in response to Message 2014931.  
Last modified: 11 Oct 2019, 2:09:34 UTC

Keith, just out of curiosity, nothing else, would you care to share how many and how long your validation pending is? You are in the top ten of SETI crunchers after all.

My current pending validation count for all hosts stands at 22,420 and the oldest validation is a task sent on 6 May and received on 7 May. My latest and 4th wingman is coming due on his deadline on 13
October. So one more possible wingman for this task if the latest bails out. And that would put the task possibly going to around the 1st week of December. So more than half a year before the task gets validated possibly.

[Edit] Counted wrong. Two more errored tasks possible. So the last possible deadline is actually 3rd week of January 2020.
Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)
ID: 2014973 · Report as offensive
Profile Keith Myers Special Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 29 Apr 01
Posts: 13161
Credit: 1,160,866,277
RAC: 1,873
United States
Message 2014974 - Posted: 11 Oct 2019, 2:13:32 UTC - in response to Message 2014961.  

Guessing also that the WOW score is based on credits earned during the WOW week even if the work had been done days before and submitted 1 second after the start. I actually noticed a huge jump by one setizen just after the start.

That is called bunkering or nuclear weapon deployment during the first week of the WOW contest. Lots of people were dropping big A-bombs constantly to goose their credit tally. All in fun and a lot of work gets done trying to one up your competitors.
Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)
ID: 2014974 · Report as offensive
juan BFP Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 16 Mar 07
Posts: 9786
Credit: 572,710,851
RAC: 3,799
Panama
Message 2014975 - Posted: 11 Oct 2019, 3:05:48 UTC - in response to Message 2014974.  

Guessing also that the WOW score is based on credits earned during the WOW week even if the work had been done days before and submitted 1 second after the start. I actually noticed a huge jump by one setizen just after the start.

That is called bunkering or nuclear weapon deployment during the first week of the WOW contest. Lots of people were dropping big A-bombs constantly to goose their credit tally. All in fun and a lot of work gets done trying to one up your competitors.

Not forget the bananas on the rear tail pipes... LOL
ID: 2014975 · Report as offensive
Profile Joseph Stateson Project Donor
Volunteer tester
Avatar

Send message
Joined: 27 May 99
Posts: 309
Credit: 70,759,933
RAC: 3
United States
Message 2015065 - Posted: 11 Oct 2019, 18:00:02 UTC - in response to Message 2014974.  

Guessing also that the WOW score is based on credits earned during the WOW week even if the work had been done days before and submitted 1 second after the start. I actually noticed a huge jump by one setizen just after the start.

That is called bunkering or nuclear weapon deployment during the first week of the WOW contest. Lots of people were dropping big A-bombs constantly to goose their credit tally. All in fun and a lot of work gets done trying to one up your competitors.


Got to thinking about this and decided to run a test. I downloaded 100 SETI (8.22 SoG win) tasks a minutes ago (Oct 11)
Deadlines for the about 100 are 12-2 through 12-11 and eye-balling: most due on 12-3 followed by 12-11 then 12-7.

Bunkering up oct:20 +nov:30 +dec:3 maybe 50 days to bunker
Cuda90 linux anon-platform takes 2 minutes per work unit. With six gtx1060 get about 14000 credits per hour

----conservative calculation ---

Assuming the limit of 900 work units per linux system that does not upload results:

With 6 GPUs at 2 minutes per WU, the 900 limit will be reached in 2 * 900 / 6 or 5 hours. After 5 hours I boot another new linux system.
For a total of 50 days I will need 50 * 24 / 5 or about 240 flash drives of 16 GB each, maybe 8 if minimum ubuntu system

In credits that gives 900 work units for 240 systems averaging 77 credits per WU or 16,632,000 credits
That just 2x over what I accumulated duringt WOW
https://www.seti-germany.de/Wow/stats/users/2019
The 900 limit can be bumped up by spoofing the number of GPUs.

1. Bunkering: 50 days
2. 240 nukes dropped
3. Banana up the tailpipe: Abort all workunits that expire before the start of the WOW event.

Lemme know if I left something out!
ID: 2015065 · Report as offensive
juan BFP Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 16 Mar 07
Posts: 9786
Credit: 572,710,851
RAC: 3,799
Panama
Message 2015077 - Posted: 11 Oct 2019, 19:15:25 UTC - in response to Message 2015065.  

In theory not. But in the real world....

- Manage such large number of hosts (240) is not practical and there are a high risk of crash a lot of them.

- The credit numbers are lower since a lot of the WU will not be validated until the WOW event ends.

- I could do bunkering, but i'm against those who abort the WU after the end of the event. That adds extra work load to the servers.

You could look in my hosts list and see how rely bunkering must works in the real world.
Each nuke (a spoofed host) gives you about 400-500K credit for the event.
As you could see i not abort any WU after the end of the event.
And i still have after about 1 1/2 months a lot of WU waiting to validation, and some will take 2 or 3 months.

But for do that you need a lot of babysitting and carefully manage your hosts.
Something i could tell you, I done with few and gives a lot of work, not imagine handling 240 hosts instead.
ID: 2015077 · Report as offensive
Profile Keith Myers Special Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 29 Apr 01
Posts: 13161
Credit: 1,160,866,277
RAC: 1,873
United States
Message 2015078 - Posted: 11 Oct 2019, 19:15:59 UTC - in response to Message 2015065.  

Guessing also that the WOW score is based on credits earned during the WOW week even if the work had been done days before and submitted 1 second after the start. I actually noticed a huge jump by one setizen just after the start.

That is called bunkering or nuclear weapon deployment during the first week of the WOW contest. Lots of people were dropping big A-bombs constantly to goose their credit tally. All in fun and a lot of work gets done trying to one up your competitors.


Got to thinking about this and decided to run a test. I downloaded 100 SETI (8.22 SoG win) tasks a minutes ago (Oct 11)
Deadlines for the about 100 are 12-2 through 12-11 and eye-balling: most due on 12-3 followed by 12-11 then 12-7.

Bunkering up oct:20 +nov:30 +dec:3 maybe 50 days to bunker
Cuda90 linux anon-platform takes 2 minutes per work unit. With six gtx1060 get about 14000 credits per hour

----conservative calculation ---

Assuming the limit of 900 work units per linux system that does not upload results:

With 6 GPUs at 2 minutes per WU, the 900 limit will be reached in 2 * 900 / 6 or 5 hours. After 5 hours I boot another new linux system.
For a total of 50 days I will need 50 * 24 / 5 or about 240 flash drives of 16 GB each, maybe 8 if minimum ubuntu system

In credits that gives 900 work units for 240 systems averaging 77 credits per WU or 16,632,000 credits
That just 2x over what I accumulated duringt WOW
https://www.seti-germany.de/Wow/stats/users/2019
The 900 limit can be bumped up by spoofing the number of GPUs.

1. Bunkering: 50 days
2. 240 nukes dropped
3. Banana up the tailpipe: Abort all workunits that expire before the start of the WOW event.

Lemme know if I left something out!

No, you got the gist of it correct. Thanks for the head's up of what to expect next year. Juan you have a new MAD competitor. LOL.
Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)
ID: 2015078 · Report as offensive
juan BFP Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 16 Mar 07
Posts: 9786
Credit: 572,710,851
RAC: 3,799
Panama
Message 2015139 - Posted: 12 Oct 2019, 1:22:46 UTC - in response to Message 2015078.  
Last modified: 12 Oct 2019, 1:26:23 UTC

Juan you have a new MAD competitor. LOL.

I think before the next WOW event i need to put my experimental 16 x 3070 mining auto bunkering host running the multi spoofed new client working or i will be doomed.
Will add a couple of: just in case. LOL
ID: 2015139 · Report as offensive
1 · 2 · 3 · Next

Message boards : Number crunching : How long must I wait for validation for SETI tasks?


 
©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.