Cheating by selecting workunits to crunch?

Message boards : Number crunching : Cheating by selecting workunits to crunch?
Message board moderation

To post messages, you must log in.

1 · 2 · Next

AuthorMessage
Profile michael37
Avatar

Send message
Joined: 23 Jul 99
Posts: 311
Credit: 6,955,447
RAC: 0
United States
Message 598940 - Posted: 5 Jul 2007, 6:49:57 UTC

About a year or so ago, someone (Alex Kan? Alex, if it wasn't your, apologies, my memory is often faulty) hypothesized that some optimized clients will be better for some workunits and worse for some other workunits. That means that credit/hour will be good only for narrow range of workunits. Then, one can hypothetically write a modification to boinc software that will parse a workunit quickly and discard (abort) unwanted units. Obviously, such code would be immoral if not outright cheating.

Is that what happens to this host? I see many workunits aborted for no apparent good reason.

ID: 598940 · Report as offensive
Brian Silvers

Send message
Joined: 11 Jun 99
Posts: 1681
Credit: 492,052
RAC: 0
United States
Message 598943 - Posted: 5 Jul 2007, 6:58:43 UTC - in response to Message 598940.  

About a year or so ago, someone (Alex Kan? Alex, if it wasn't your, apologies, my memory is often faulty) hypothesized that some optimized clients will be better for some workunits and worse for some other workunits. That means that credit/hour will be good only for narrow range of workunits. Then, one can hypothetically write a modification to boinc software that will parse a workunit quickly and discard (abort) unwanted units. Obviously, such code would be immoral if not outright cheating.

Is that what happens to this host? I see many workunits aborted for no apparent good reason.


Version 5.8.17 and higher support project-side aborts. This person is using 5.10.7. Please read the Aborted by Project sticky...
ID: 598943 · Report as offensive
Profile Adrian Taylor
Volunteer tester
Avatar

Send message
Joined: 22 Apr 01
Posts: 95
Credit: 10,933,449
RAC: 0
United Kingdom
Message 598991 - Posted: 5 Jul 2007, 11:04:04 UTC - in response to Message 598940.  

About a year or so ago, someone (Alex Kan? Alex, if it wasn't your, apologies, my memory is often faulty) hypothesized that some optimized clients will be better for some workunits and worse for some other workunits. That means that credit/hour will be good only for narrow range of workunits. Then, one can hypothetically write a modification to boinc software that will parse a workunit quickly and discard (abort) unwanted units. Obviously, such code would be immoral if not outright cheating.

Is that what happens to this host? I see many workunits aborted for no apparent good reason.


I dont think its very productive to use the words immoral and cheating, without even bothering to find out what the issue is.

perhaps you should stop going through other peoples stats looking for wrong-uns, and instead read the message boards over the last few weeks, in which case you may have noticed theres new features in the new boinc app

my machines are open to your scrutiny too, if you'd like to go and check my worthiness to participate in this project :-)

regards

adream
63. (1) (b) "music" includes sounds wholly or predominantly characterised by the emission of a succession of repetitive beats
ID: 598991 · Report as offensive
kittyman Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 9 Jul 00
Posts: 51468
Credit: 1,018,363,574
RAC: 1,004
United States
Message 599032 - Posted: 5 Jul 2007, 13:21:12 UTC - in response to Message 598943.  

About a year or so ago, someone (Alex Kan? Alex, if it wasn't your, apologies, my memory is often faulty) hypothesized that some optimized clients will be better for some workunits and worse for some other workunits. That means that credit/hour will be good only for narrow range of workunits. Then, one can hypothetically write a modification to boinc software that will parse a workunit quickly and discard (abort) unwanted units. Obviously, such code would be immoral if not outright cheating.

Is that what happens to this host? I see many workunits aborted for no apparent good reason.


Version 5.8.17 and higher support project-side aborts. This person is using 5.10.7. Please read the Aborted by Project sticky...


These may indeed by server induced aborts. But this issue does underline the request for Seti to differentiate between aborts by the project server (no longer needed or such) and manual aborts by the user on the client side.
Being an 8 core Xeon based rig, I don't think the RAC is out of line.

"Freedom is just Chaos, with better lighting." Alan Dean Foster

ID: 599032 · Report as offensive
Profile Spectrum
Avatar

Send message
Joined: 14 Jun 99
Posts: 468
Credit: 53,129,336
RAC: 0
Australia
Message 599035 - Posted: 5 Jul 2007, 13:30:10 UTC

Sterling_Aug User profile image
Created 22 Jun 2007 21:57:57 UTC
Total Credit 340,575.59
Recent average credit 5,589.42
CPU type GenuineIntel
Intel(R) Xeon(R) CPU X5355 @ 2.66GHz [EM64T Family 6 Model 15 Stepping 7]
Number of CPUs 8
Operating System Microsoft Windows XP Professional x64 Edition
, Service Pack 2, (05.02.3790.00)
Memory 4093.36 MB
Cache 122.07 KB
Measured floating point speed 2533.26 million ops/sec
Measured integer speed 7862.92 million ops/sec
Average upload rate 4.13 KB/sec
Average download rate Unknown
Average turnaround time 3.29 days
Maximum daily WU quota per CPU 100/day
Results 1858

Read your post and checked out the comp in question,,, u don't need to cheat with one of these,,
Wish I could afford one!
ID: 599035 · Report as offensive
Alinator
Volunteer tester

Send message
Joined: 19 Apr 05
Posts: 4178
Credit: 4,647,982
RAC: 0
United States
Message 599037 - Posted: 5 Jul 2007, 13:32:12 UTC - in response to Message 599032.  


These may indeed by server induced aborts. But this issue does underline the request for Seti to differentiate between aborts by the project server (no longer needed or such) and manual aborts by the user on the client side.
Being an 8 core Xeon based rig, I don't think the RAC is out of line.


It does, auto-aborts are 221's, manual aborts are 197's.

Alinator
ID: 599037 · Report as offensive
kittyman Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 9 Jul 00
Posts: 51468
Credit: 1,018,363,574
RAC: 1,004
United States
Message 599043 - Posted: 5 Jul 2007, 13:48:49 UTC - in response to Message 599037.  


These may indeed by server induced aborts. But this issue does underline the request for Seti to differentiate between aborts by the project server (no longer needed or such) and manual aborts by the user on the client side.
Being an 8 core Xeon based rig, I don't think the RAC is out of line.


It does, auto-aborts are 221's, manual aborts are 197's.

Alinator


Well, I think that clears up any question about this computer. I checked a few of the aborts, and they are indeed 221 exit codes.
What I think most users are looking for is that the 'results for computer' page display something other than the usual 'client error' and 'aborted' when there is no client error but rather a cancellation by the project server. Something like 'server cancelled' and 'unneeded' would be more accurate and readily apparent to users who do not know to dig further for the exit codes and what they mean.

"Freedom is just Chaos, with better lighting." Alan Dean Foster

ID: 599043 · Report as offensive
Alinator
Volunteer tester

Send message
Joined: 19 Apr 05
Posts: 4178
Credit: 4,647,982
RAC: 0
United States
Message 599051 - Posted: 5 Jul 2007, 13:57:34 UTC - in response to Message 599043.  
Last modified: 5 Jul 2007, 13:58:23 UTC

Well, I think that clears up any question about this computer. I checked a few of the aborts, and they are indeed 221 exit codes.
What I think most users are looking for is that the 'results for computer' page display something other than the usual 'client error' and 'aborted' when there is no client error but rather a cancellation by the project server. Something like 'server cancelled' and 'unneeded' would be more accurate and readily apparent to users who do not know to dig further for the exit codes and what they mean.


LOL...

Yeah, I hear ya! ;-)

Ever notice how as the versions increase BOINC swings back and forth between CC message tab logging which is either vebose to the point of being mind numblingly useless, to so succinct it's almost bafflingly useless?

Right now, IMO the default is pretty thin, but look out if you enable the detail flags! :-)

Alinator
ID: 599051 · Report as offensive
Brian Silvers

Send message
Joined: 11 Jun 99
Posts: 1681
Credit: 492,052
RAC: 0
United States
Message 599095 - Posted: 5 Jul 2007, 16:03:30 UTC - in response to Message 599043.  


These may indeed by server induced aborts. But this issue does underline the request for Seti to differentiate between aborts by the project server (no longer needed or such) and manual aborts by the user on the client side.
Being an 8 core Xeon based rig, I don't think the RAC is out of line.


It does, auto-aborts are 221's, manual aborts are 197's.

Alinator


Well, I think that clears up any question about this computer. I checked a few of the aborts, and they are indeed 221 exit codes.
What I think most users are looking for is that the 'results for computer' page display something other than the usual 'client error' and 'aborted' when there is no client error but rather a cancellation by the project server. Something like 'server cancelled' and 'unneeded' would be more accurate and readily apparent to users who do not know to dig further for the exit codes and what they mean.


You're preaching to the choir... ;-)

It's all wonderful that there is a different code, but codes are meaningless without their definition. Lots of people won't go hunting for the definition and just assume stuff...
ID: 599095 · Report as offensive
Profile Philadelphia
Volunteer tester
Avatar

Send message
Joined: 12 Feb 07
Posts: 1590
Credit: 399,688
RAC: 0
United States
Message 599102 - Posted: 5 Jul 2007, 16:15:11 UTC - in response to Message 599095.  


These may indeed by server induced aborts. But this issue does underline the request for Seti to differentiate between aborts by the project server (no longer needed or such) and manual aborts by the user on the client side.
Being an 8 core Xeon based rig, I don't think the RAC is out of line.


It does, auto-aborts are 221's, manual aborts are 197's.

Alinator


Well, I think that clears up any question about this computer. I checked a few of the aborts, and they are indeed 221 exit codes.
What I think most users are looking for is that the 'results for computer' page display something other than the usual 'client error' and 'aborted' when there is no client error but rather a cancellation by the project server. Something like 'server cancelled' and 'unneeded' would be more accurate and readily apparent to users who do not know to dig further for the exit codes and what they mean.


You're preaching to the choir... ;-)

It's all wonderful that there is a different code, but codes are meaningless without their definition. Lots of people won't go hunting for the definition and just assume stuff...


Maybe it would be helpful if there was a sticky of the exit codes.

ID: 599102 · Report as offensive
Alinator
Volunteer tester

Send message
Joined: 19 Apr 05
Posts: 4178
Credit: 4,647,982
RAC: 0
United States
Message 599109 - Posted: 5 Jul 2007, 16:26:52 UTC

ID: 599109 · Report as offensive
Brian Silvers

Send message
Joined: 11 Jun 99
Posts: 1681
Credit: 492,052
RAC: 0
United States
Message 599114 - Posted: 5 Jul 2007, 16:37:26 UTC - in response to Message 599102.  


Maybe it would be helpful if there was a sticky of the exit codes.


The answer to this vexing question was already covered in the sticky that I mentioned, yet the OP apparently never read it...

A semi-humorous example of sticky failure on another project, LHC, has this from the project administrator about the question of when the project will have work:


Also please note, and this is to everyone, if I see another thread on this topic I will delete it, can you not read the sticky at the top of the forum?


A stickied post in the forum, currently the second from the top, is titled:

Please note: this project rarely has work

ID: 599114 · Report as offensive
Profile Philadelphia
Volunteer tester
Avatar

Send message
Joined: 12 Feb 07
Posts: 1590
Credit: 399,688
RAC: 0
United States
Message 599137 - Posted: 5 Jul 2007, 17:43:15 UTC

Thanks.

Actually what I was referring to was a sticky on the "Number crunching" page. I think there are ~5 stickies now.

My thought, maybe not a good one, lol, was that it was visible to everyone who looks at the page.

No biggie though.


ID: 599137 · Report as offensive
Alinator
Volunteer tester

Send message
Joined: 19 Apr 05
Posts: 4178
Credit: 4,647,982
RAC: 0
United States
Message 599140 - Posted: 5 Jul 2007, 17:50:14 UTC

LOL...

Well, we can't take all the sport out of BOINC'ing! :-)

Alinator
ID: 599140 · Report as offensive
Profile Clyde C. Phillips, III

Send message
Joined: 2 Aug 00
Posts: 1851
Credit: 5,955,047
RAC: 0
United States
Message 599142 - Posted: 5 Jul 2007, 17:53:34 UTC

It looked like he aborted 30s and 60s in favor of doing 30s. If his machine bears any resemblance to mine and to many others, he's cutting his own throat by aborting. In my and many other cases 60s do better than 30s. The really funny thing is that he's aborting both 30s and 60s. I don't know why.
ID: 599142 · Report as offensive
Profile Philadelphia
Volunteer tester
Avatar

Send message
Joined: 12 Feb 07
Posts: 1590
Credit: 399,688
RAC: 0
United States
Message 599145 - Posted: 5 Jul 2007, 17:55:33 UTC - in response to Message 599140.  

LOL...

Well, we can't take all the sport out of BOINC'ing! :-)

Alinator


I didn't even know the FAQ section was there, lol.

ID: 599145 · Report as offensive
Profile LTDInvestments
Volunteer tester
Avatar

Send message
Joined: 2 Aug 99
Posts: 14
Credit: 4,592,515
RAC: 42
United States
Message 599148 - Posted: 5 Jul 2007, 17:58:43 UTC - in response to Message 599142.  
Last modified: 5 Jul 2007, 17:59:42 UTC

It looked like he aborted 30s and 60s in favor of doing 30s. If his machine bears any resemblance to mine and to many others, he's cutting his own throat by aborting. In my and many other cases 60s do better than 30s. The really funny thing is that he's aborting both 30s and 60s. I don't know why.


I looked at about a dozen of them and all the aborts come after qourum is meet.


ID: 599148 · Report as offensive
Alinator
Volunteer tester

Send message
Joined: 19 Apr 05
Posts: 4178
Credit: 4,647,982
RAC: 0
United States
Message 599154 - Posted: 5 Jul 2007, 18:10:09 UTC - in response to Message 599148.  
Last modified: 5 Jul 2007, 18:11:41 UTC


I looked at about a dozen of them and all the aborts come after qourum is meet.



Agreed, all the ones I looked at were 221 auto-aborts. No CWCP'ing here that I could see.

Also, The main reason you see a lot 60 something credit results draw the 221 is if you have your cache set to a large value, when you draw a bunch 'shorties' the scheduler puts the long running ones on the backburner in order to clear the earlier deadline ones. In a lot of case this gives your wingmen a much better chance of beating your host back to forum the quorum and make your result a candidate for a 221.

Alinator
ID: 599154 · Report as offensive
Brian Silvers

Send message
Joined: 11 Jun 99
Posts: 1681
Credit: 492,052
RAC: 0
United States
Message 599161 - Posted: 5 Jul 2007, 18:25:38 UTC - in response to Message 599137.  


My thought, maybe not a good one, lol, was that it was visible to everyone who looks at the page.


@Philadelphia - Yeah, I know what you thought, and it's not a bad thought, it's just that if the result pages (both views) had better explanations, it would go a lot further towards both explaining it to the curious and for quelling the "fears" of those wanting to "protect" their stat standings...

IMO, YMMV, etc, etc, etc...
ID: 599161 · Report as offensive
Profile Philadelphia
Volunteer tester
Avatar

Send message
Joined: 12 Feb 07
Posts: 1590
Credit: 399,688
RAC: 0
United States
Message 599171 - Posted: 5 Jul 2007, 18:40:45 UTC - in response to Message 599161.  


My thought, maybe not a good one, lol, was that it was visible to everyone who looks at the page.


@Philadelphia - Yeah, I know what you thought, and it's not a bad thought, it's just that if the result pages (both views) had better explanations, it would go a lot further towards both explaining it to the curious and for quelling the "fears" of those wanting to "protect" their stat standings...

IMO, YMMV, etc, etc, etc...


Agreed, they are short and cripted.

ID: 599171 · Report as offensive
1 · 2 · Next

Message boards : Number crunching : Cheating by selecting workunits to crunch?


 
©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.