Welcome to MilkyWay@home

Please check this host

Message boards : Number crunching : Please check this host
Message board moderation

To post messages, you must log in.

1 · 2 · Next

AuthorMessage
Profile DaveSun
Avatar

Send message
Joined: 10 Nov 07
Posts: 28
Credit: 2,549,231
RAC: 0
Message 556 - Posted: 24 Nov 2007, 19:24:32 UTC

Checkout this host for the last few days it's claimed/granted credit jumped from 1.xx to 25x.xx per result with no increase in crunch time.

Even without a quorum requirement there should be a check for this type of over claim.
ID: 556 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile agony

Send message
Joined: 24 Oct 07
Posts: 22
Credit: 130,021
RAC: 0
Message 557 - Posted: 24 Nov 2007, 19:56:45 UTC

just one more from the cheaters gang. check his cpu benchmark. maybe he brought a cpu from the year 2070 or something ;). best way to get rid of them is to have a quorom. fits niceley to almost all top hosts in this project all have silly benchmarks.

ID: 557 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Crystallize
Avatar

Send message
Joined: 12 Nov 07
Posts: 31
Credit: 123,621
RAC: 0
Message 558 - Posted: 24 Nov 2007, 19:59:32 UTC

Hm, yes, these is as far as I can see an obvious cheat attempt ????

or is it a error of some kind ???

I never get more than 4-8 CS per WU, and it shouldn't be more either with such a short WU completion time and certainly not over 200CS for every WU !
ID: 558 · Rating: -1 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile rebirther
Avatar

Send message
Joined: 28 Aug 07
Posts: 52
Credit: 8,353,747
RAC: 0
Message 559 - Posted: 24 Nov 2007, 20:14:12 UTC - in response to Message 558.  

Hm, yes, these is as far as I can see an obvious cheat attempt ????

or is it a error of some kind ???

I never get more than 4-8 CS per WU, and it shouldn't be more either with such a short WU completion time and certainly not over 200CS for every WU !


I have contacted the admin earlier, but crystallize, you are also using an unofficial 5.9.0 version with higher benchmarks!
ID: 559 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Crystallize
Avatar

Send message
Joined: 12 Nov 07
Posts: 31
Credit: 123,621
RAC: 0
Message 560 - Posted: 24 Nov 2007, 20:36:32 UTC - in response to Message 559.  

Hm, yes, these is as far as I can see an obvious cheat attempt ????

or is it a error of some kind ???

I never get more than 4-8 CS per WU, and it shouldn't be more either with such a short WU completion time and certainly not over 200CS for every WU !


I have contacted the admin earlier, but crystallize, you are also using an unofficial 5.9.0 version with higher benchmarks!


Hm, I'm using our teams (TSWB's) BOINC application, with instant WU reporting, as far as I know it doesn't have higher benchmarks than the original BOINC !
ID: 560 · Rating: -1 · rate: Rate + / Rate - Report as offensive     Reply Quote
Irishgeezah
Avatar

Send message
Joined: 10 Nov 07
Posts: 37
Credit: 11,855,733
RAC: 0
Message 561 - Posted: 24 Nov 2007, 21:30:27 UTC

This is another host that possibly is using a dodgy client http://milkyway.cs.rpi.edu/milkyway/show_host_detail.php?hostid=1314


ID: 561 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile agony

Send message
Joined: 24 Oct 07
Posts: 22
Credit: 130,021
RAC: 0
Message 563 - Posted: 24 Nov 2007, 22:43:21 UTC

out of sudden. the cheat hosts all have hidden computers now lol

ID: 563 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
zombie67 [MM]
Avatar

Send message
Joined: 29 Aug 07
Posts: 115
Credit: 501,600,397
RAC: 5,019
Message 564 - Posted: 24 Nov 2007, 23:12:13 UTC - in response to Message 557.  

best way to get rid of them is to have a quorom.


I disagree. That also has the side effect of reducing the total crunching by 50% or more. Just slows down the project. Increasing the quorum should be done only if the science requires it.

A better way to deal with this kind of thing is fixed credits, or some sort of step counting.

ID: 564 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Crystallize
Avatar

Send message
Joined: 12 Nov 07
Posts: 31
Credit: 123,621
RAC: 0
Message 566 - Posted: 25 Nov 2007, 9:44:37 UTC - in response to Message 563.  
Last modified: 25 Nov 2007, 9:52:23 UTC

out of sudden. the cheat hosts all have hidden computers now lol



Yes, I'll ask the team on our forum about this subject.
I don't want to take any more sh*t about it before I've investigate this more throughout ...

http://www.tswb.org/Forums/tabid/173/mid/504/threadid/11478/scope/posts/Default.aspx#11478
ID: 566 · Rating: -1 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile rebirther
Avatar

Send message
Joined: 28 Aug 07
Posts: 52
Credit: 8,353,747
RAC: 0
Message 567 - Posted: 25 Nov 2007, 10:03:51 UTC - in response to Message 564.  

best way to get rid of them is to have a quorom.


I disagree. That also has the side effect of reducing the total crunching by 50% or more. Just slows down the project. Increasing the quorum should be done only if the science requires it.

A better way to deal with this kind of thing is fixed credits, or some sort of step counting.


Yes, also excluding 5.5.0/5.9.0 clients and perhaps an upper credit cap, reducing the credits of these high cheated hosts by factor 50-100, but as I have seen fixed credits would be wonderful, running times are all the same.
ID: 567 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile nutcase

Send message
Joined: 25 Nov 07
Posts: 11
Credit: 40,758,862
RAC: 0
Message 568 - Posted: 25 Nov 2007, 11:01:01 UTC
Last modified: 25 Nov 2007, 11:07:30 UTC

The High Benchmarks could be causing a problem if that is how the Core client figures out scores. That is an Old version of BOINC and was known to cause weird results. They did not cheat but should update their Boinc client to a newer version.

You guys should have looked at the returned results and would have found out Either the core client has a problem or the WU's are causing problems.

Every WU I checked showed memory leaks in them. this can only be caused by the core client or by the WU's. This would not be caused by the Boinc Client.

ID: 568 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Jord
Avatar

Send message
Joined: 30 Aug 07
Posts: 125
Credit: 207,206
RAC: 0
Message 569 - Posted: 25 Nov 2007, 11:24:15 UTC - in response to Message 568.  

The High Benchmarks could be causing a problem if that is how the Core client figures out scores. That is an Old version of BOINC and was known to cause weird results. They did not cheat but should update their Boinc client to a newer version.

Version 5.10.20 had no problems with correctly benchmarking a CPU.
The BOINC versions that possibly gave weird outcomes due to benchmark inconsistencies were 5.8.17, 5.10.1 and 5.10.6

None of the above versions gave integer benchmark claims in the way the two clients in this thread have them. So either someone took the core client code and adjusted how the benchmarks should be done before compiling it as a 5.10.20 client (easily done by changing the version number before compiling), or they've changed their values in client_state.xml


Jord.

The BOINC FAQ Service.
ID: 569 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Paul
Avatar

Send message
Joined: 28 Aug 07
Posts: 1
Credit: 25,038,876
RAC: 0
Message 570 - Posted: 25 Nov 2007, 14:07:35 UTC - in response to Message 568.  

The High Benchmarks could be causing a problem if that is how the Core client figures out scores. That is an Old version of BOINC and was known to cause weird results. They did not cheat but should update their Boinc client to a newer version.

You guys should have looked at the returned results and would have found out Either the core client has a problem or the WU's are causing problems.

Every WU I checked showed memory leaks in them. this can only be caused by the core client or by the WU's. This would not be caused by the Boinc Client.


Good work, Dan. The fingerpointing stats ho's go nuts if their clients get knocked down by an erroring client. :-(

Crystallize is using her own BOINC skin, running Crunch3r's old optimized BOINC client that reports work immediately. She mistakenly thought that her client was TSWB's because of the wording TSWB-BOINC Client in the window title.

She has been advised to upgrade her BOINC client. IIRC, 5.10.1x reports immediately, but 5.10.2x does not. In any case, further advances in the BOINC client, whether it's credit issues, bug fixes, or added features certainly makes it worth upgrading to 5.10.28.
Team Starfire World BOINC
IRC- irc//irc.teamstarfire.net:6667/team_starfire
ID: 570 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Jenik

Send message
Joined: 7 Oct 07
Posts: 4
Credit: 9,887,161
RAC: 0
Message 571 - Posted: 25 Nov 2007, 14:19:47 UTC - in response to Message 569.  

The High Benchmarks could be causing a problem if that is how the Core client figures out scores. That is an Old version of BOINC and was known to cause weird results. They did not cheat but should update their Boinc client to a newer version.

Version 5.10.20 had no problems with correctly benchmarking a CPU.
The BOINC versions that possibly gave weird outcomes due to benchmark inconsistencies were 5.8.17, 5.10.1 and 5.10.6

None of the above versions gave integer benchmark claims in the way the two clients in this thread have them. So either someone took the core client code and adjusted how the benchmarks should be done before compiling it as a 5.10.20 client (easily done by changing the version number before compiling), or they've changed their values in client_state.xml



BINGO! client_state.xml
But "changing the version number" is OT.
Try edit (make higher) benchmark in this file.
For example <p_iops>500002814793331.347700</p_iops> instead of <p_iops>2814793331.347700</p_iops>
and "credit jumped from 1.xx to 1xxx.xx per result with no increase in crunch time."
Version number of the Boinc client is irrelevant.
To admin: ...Houston, ve have problem...
Best and simply is FIXED credit.
ID: 571 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Jord
Avatar

Send message
Joined: 30 Aug 07
Posts: 125
Credit: 207,206
RAC: 0
Message 572 - Posted: 25 Nov 2007, 14:23:02 UTC - in response to Message 570.  

IIRC, 5.10.1x reports immediately, but 5.10.2x does not.

All up to 5.10.13 report the same day when asking for new work.

5.10.14 and above report after 24 hours, or when requesting more work when the queue is empty, or when done manually and further following the normal rules of contact.

No client reports immediately. That was a command line option in the 4.xx version, long since deprecated. (i.e. no longer in the code)
Jord.

The BOINC FAQ Service.
ID: 572 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Travis
Volunteer moderator
Project administrator
Project developer
Project tester
Project scientist

Send message
Joined: 30 Aug 07
Posts: 2046
Credit: 26,480
RAC: 0
Message 582 - Posted: 25 Nov 2007, 22:28:22 UTC - in response to Message 572.  

I'll try and update our validator tonight to get a quorum of 2 going, so hopefully this will fix any cheating.
ID: 582 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile rebirther
Avatar

Send message
Joined: 28 Aug 07
Posts: 52
Credit: 8,353,747
RAC: 0
Message 584 - Posted: 25 Nov 2007, 22:30:28 UTC - in response to Message 582.  

I'll try and update our validator tonight to get a quorum of 2 going, so hopefully this will fix any cheating.


This is not a good solution but temporarily.
ID: 584 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Travis
Volunteer moderator
Project administrator
Project developer
Project tester
Project scientist

Send message
Joined: 30 Aug 07
Posts: 2046
Credit: 26,480
RAC: 0
Message 586 - Posted: 25 Nov 2007, 22:38:37 UTC - in response to Message 571.  

The High Benchmarks could be causing a problem if that is how the Core client figures out scores. That is an Old version of BOINC and was known to cause weird results. They did not cheat but should update their Boinc client to a newer version.

Version 5.10.20 had no problems with correctly benchmarking a CPU.
The BOINC versions that possibly gave weird outcomes due to benchmark inconsistencies were 5.8.17, 5.10.1 and 5.10.6

None of the above versions gave integer benchmark claims in the way the two clients in this thread have them. So either someone took the core client code and adjusted how the benchmarks should be done before compiling it as a 5.10.20 client (easily done by changing the version number before compiling), or they've changed their values in client_state.xml



BINGO! client_state.xml
But "changing the version number" is OT.
Try edit (make higher) benchmark in this file.
For example 500002814793331.347700 instead of 2814793331.347700
and "credit jumped from 1.xx to 1xxx.xx per result with no increase in crunch time."
Version number of the Boinc client is irrelevant.
To admin: ...Houston, ve have problem...
Best and simply is FIXED credit.


Thats probably a good idea. The work units are (for the most part) fixed size, so fixed credit might be the way to go. Currently, the amount of work done is based off two things: 1. the size of the volume, and 2. the number of stars.

Between a quorum of 2 and a way of calculating credit not based off boinc's benchmarks, maybe that will fix the problem?
ID: 586 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile rebirther
Avatar

Send message
Joined: 28 Aug 07
Posts: 52
Credit: 8,353,747
RAC: 0
Message 587 - Posted: 25 Nov 2007, 22:40:19 UTC - in response to Message 586.  

The High Benchmarks could be causing a problem if that is how the Core client figures out scores. That is an Old version of BOINC and was known to cause weird results. They did not cheat but should update their Boinc client to a newer version.

Version 5.10.20 had no problems with correctly benchmarking a CPU.
The BOINC versions that possibly gave weird outcomes due to benchmark inconsistencies were 5.8.17, 5.10.1 and 5.10.6

None of the above versions gave integer benchmark claims in the way the two clients in this thread have them. So either someone took the core client code and adjusted how the benchmarks should be done before compiling it as a 5.10.20 client (easily done by changing the version number before compiling), or they've changed their values in client_state.xml



BINGO! client_state.xml
But "changing the version number" is OT.
Try edit (make higher) benchmark in this file.
For example <p_iops>500002814793331.347700</p_iops> instead of <p_iops>2814793331.347700</p_iops>
and "credit jumped from 1.xx to 1xxx.xx per result with no increase in crunch time."
Version number of the Boinc client is irrelevant.
To admin: ...Houston, ve have problem...
Best and simply is FIXED credit.


Thats probably a good idea. The work units are (for the most part) fixed size, so fixed credit might be the way to go. Currently, the amount of work done is based off two things: 1. the size of the volume, and 2. the number of stars.

Between a quorum of 2 and a way of calculating credit not based off boinc's benchmarks, maybe that will fix the problem?


You can define credits in the wu template and validator.
ID: 587 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Jayargh
Avatar

Send message
Joined: 8 Oct 07
Posts: 289
Credit: 3,690,838
RAC: 0
Message 588 - Posted: 25 Nov 2007, 23:21:31 UTC

Travis please don't go to a quorum of 2 if it doesn't affect the science as it is a waste of cpu time...seems you have other tools to adjust for the cheaters :)
ID: 588 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
1 · 2 · Next

Message boards : Number crunching : Please check this host

©2024 Astroinformatics Group