Rogue 560 Ti's

Message boards : Number crunching : Rogue 560 Ti's
Message board moderation

To post messages, you must log in.

1 · 2 · Next

AuthorMessage
Terror Australis
Volunteer tester

Send message
Joined: 14 Feb 04
Posts: 1817
Credit: 262,693,308
RAC: 44
Australia
Message 1297435 - Posted: 21 Oct 2012, 5:34:42 UTC

Now that we have a "Validation Inconclusive" tag on our results pages, it it making it easier to track down machines that are producing large numbers of invalid results or errors.

One thing I've noticed is that better than 80% of the "big" error generators are running GTX560 Ti's and the "stock" V10 CUDA app.

Those of us that haunt the forums know there were problems with the 560 Ti when it was first released but obviously a lot of people are still not aware.

What can be done ? Would it be possible for The Project to all 560 users with a general message that a problem has been discovered with these cards and advising to upgrade their app. This would also grab the "Anonymous" users that can't be reached by PM ?

T.A.
ID: 1297435 · Report as offensive
Profile Slavac
Volunteer tester
Avatar

Send message
Joined: 27 Apr 11
Posts: 1932
Credit: 17,952,639
RAC: 0
United States
Message 1297436 - Posted: 21 Oct 2012, 5:45:27 UTC - in response to Message 1297435.  

Not sure however several iterations ago the optimization app fixed this issue with 560ti's totally.

The issue is a lot of people probably don't want to install a 'third party' app on their machine for whatever reason which makes me wonder how we can solve the issue on a large scale vs reaching out to one cruncher at a time.


Executive Director GPU Users Group Inc. -
brad@gpuug.org
ID: 1297436 · 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 1297448 - Posted: 21 Oct 2012, 6:51:09 UTC - in response to Message 1297436.  

Not sure however several iterations ago the optimization app fixed this issue with 560ti's totally.

The issue is a lot of people probably don't want to install a 'third party' app on their machine for whatever reason which makes me wonder how we can solve the issue on a large scale vs reaching out to one cruncher at a time.

Have the copyright issues with the Lunatics Optimized Apps been resolved? Can't ask 560TI users to upgrade to Opt. Apps. when they are not available to download.
Donald
Infernal Optimist / Submariner, retired
ID: 1297448 · Report as offensive
Profile Slavac
Volunteer tester
Avatar

Send message
Joined: 27 Apr 11
Posts: 1932
Credit: 17,952,639
RAC: 0
United States
Message 1297449 - Posted: 21 Oct 2012, 6:52:49 UTC - in response to Message 1297448.  

Not sure however several iterations ago the optimization app fixed this issue with 560ti's totally.

The issue is a lot of people probably don't want to install a 'third party' app on their machine for whatever reason which makes me wonder how we can solve the issue on a large scale vs reaching out to one cruncher at a time.

Have the copyright issues with the Lunatics Optimized Apps been resolved? Can't ask 560TI users to upgrade to Opt. Apps. when they are not available to download.


We should have a resolution on that soon. (I know that's vague as heck, sorry)


Executive Director GPU Users Group Inc. -
brad@gpuug.org
ID: 1297449 · Report as offensive
Profile Wiggo
Avatar

Send message
Joined: 24 Jan 00
Posts: 34744
Credit: 261,360,520
RAC: 489
Australia
Message 1297456 - Posted: 21 Oct 2012, 7:04:11 UTC - in response to Message 1297435.  

Now that we have a "Validation Inconclusive" tag on our results pages, it it making it easier to track down machines that are producing large numbers of invalid results or errors.

One thing I've noticed is that better than 80% of the "big" error generators are running GTX560 Ti's and the "stock" V10 CUDA app.

Those of us that haunt the forums know there were problems with the 560 Ti when it was first released but obviously a lot of people are still not aware.

What can be done ? Would it be possible for The Project to all 560 users with a general message that a problem has been discovered with these cards and advising to upgrade their app. This would also grab the "Anonymous" users that can't be reached by PM ?

T.A.

It has improved though since the "Sleepy driver" problem, coupled with them was a real nightmare not long back.

Cheers.
ID: 1297456 · 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 1297469 - Posted: 21 Oct 2012, 7:31:03 UTC

Said in another thread.
Stop sending them tasks to F up.
The servers should be able to figure in no simple terms when a rig is soaking up tasks just to screw up the processing.
There is far too much of the project's bandwidth and other resources being wasted on rouge rigs.
"Freedom is just Chaos, with better lighting." Alan Dean Foster

ID: 1297469 · Report as offensive
Profile James Sotherden
Avatar

Send message
Joined: 16 May 99
Posts: 10436
Credit: 110,373,059
RAC: 54
United States
Message 1297636 - Posted: 21 Oct 2012, 16:28:08 UTC

I have to agree with Mark. Stop sending them tasks. It shows up you have what ever GPU card running with such and such driver. Seems it wouldnt be to hard to shut down the rogues turning in junk. I plan on using a 560Ti but not untill I get the new updates from lunatics.( When they resolve the legal issues ) And you can bet I sure will keep an eye on that card also. And as an open invitation if it starts messing up please feel free to PM and tell me.

This issue has been going on know for a while. Now that its been showing up in the new category Hopefully it will get resolved.
[/quote]

Old James
ID: 1297636 · Report as offensive
Profile Tron

Send message
Joined: 16 Aug 09
Posts: 180
Credit: 2,250,468
RAC: 0
United States
Message 1297658 - Posted: 21 Oct 2012, 17:30:08 UTC
Last modified: 21 Oct 2012, 17:35:28 UTC

Work should not be distributed to machines with a

Valid : Invalid ratio of 50/50 or worse
This would allow wiggle room for those of us who do pay attention to our rigs, to find and resolve problems without interrupting the work flow.
While at the same time, cutting off the jokers running on auto pilot before they have a chance to produce truck loads of bogus work.
ID: 1297658 · 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 1297679 - Posted: 21 Oct 2012, 18:37:31 UTC - in response to Message 1297658.  
Last modified: 21 Oct 2012, 18:48:50 UTC

Work should not be distributed to machines with a

Valid : Invalid ratio of 50/50 or worse
This would allow wiggle room for those of us who do pay attention to our rigs, to find and resolve problems without interrupting the work flow.
While at the same time, cutting off the jokers running on auto pilot before they have a chance to produce truck loads of bogus work.

I don't understand, I guess, why the throttling system supposedly in place does not stop them before they amass hundreds or thousands of invalid results.
Maybe it needs a bit of tweaking....
I mean, I would not get ticked off one bit if the servers cut one of my rigs off should a GPU go wonky.

Of course, if a good running rig with a big cache goes south, you are gonna have a ton of bad results as it plows through the cache. The kitties have seen that happen here once or twice. Main thing is that the servers pick up on that fact and limit the damage to work already issued.
"Freedom is just Chaos, with better lighting." Alan Dean Foster

ID: 1297679 · Report as offensive
Wembley
Volunteer tester
Avatar

Send message
Joined: 16 Sep 09
Posts: 429
Credit: 1,844,293
RAC: 0
United States
Message 1297769 - Posted: 22 Oct 2012, 0:42:48 UTC - in response to Message 1297679.  

Work should not be distributed to machines with a

Valid : Invalid ratio of 50/50 or worse
This would allow wiggle room for those of us who do pay attention to our rigs, to find and resolve problems without interrupting the work flow.
While at the same time, cutting off the jokers running on auto pilot before they have a chance to produce truck loads of bogus work.

I don't understand, I guess, why the throttling system supposedly in place does not stop them before they amass hundreds or thousands of invalid results.
Maybe it needs a bit of tweaking....
I mean, I would not get ticked off one bit if the servers cut one of my rigs off should a GPU go wonky.

Of course, if a good running rig with a big cache goes south, you are gonna have a ton of bad results as it plows through the cache. The kitties have seen that happen here once or twice. Main thing is that the servers pick up on that fact and limit the damage to work already issued.


The server code only recognizes tasks that error out as 'Errors', while tasks that fail validation at a later date (could be up to a month or two after the task is submitted) aren't counted as 'Errors'.
ID: 1297769 · Report as offensive
Profile Tron

Send message
Joined: 16 Aug 09
Posts: 180
Credit: 2,250,468
RAC: 0
United States
Message 1297771 - Posted: 22 Oct 2012, 1:07:35 UTC

ouch ... just noticed another wave of inconclusive ... a quick look at the wingmen ... gtx 560 ti ....

Is there a thread where you guys are keeping track of who has been PMed ?

one to add to your list : 9302233 (Murat Adas) PM sent 9pm oct 21

ID: 1297771 · Report as offensive
Profile Slavac
Volunteer tester
Avatar

Send message
Joined: 27 Apr 11
Posts: 1932
Credit: 17,952,639
RAC: 0
United States
Message 1297773 - Posted: 22 Oct 2012, 2:17:57 UTC - in response to Message 1297636.  

I have to agree with Mark. Stop sending them tasks. It shows up you have what ever GPU card running with such and such driver. Seems it wouldnt be to hard to shut down the rogues turning in junk. I plan on using a 560Ti but not untill I get the new updates from lunatics.( When they resolve the legal issues ) And you can bet I sure will keep an eye on that card also. And as an open invitation if it starts messing up please feel free to PM and tell me.

This issue has been going on know for a while. Now that its been showing up in the new category Hopefully it will get resolved.


It'll have to be a clever system to avoid involuntarily shutting down someone. It'll also need a mechanism to start sending tasks (like we already do w/ # of valids = # of tasks sent) once they get their machine back working properly.

One way it'll need to be clever is for people who do stupid stuff like me. I was testing an Op App that was in Beta and typo'd one component and auto errored out thousands of tasks. While that sucks for everyone involved it wasn't indicative of a broken machine, just a stupid user.


Executive Director GPU Users Group Inc. -
brad@gpuug.org
ID: 1297773 · Report as offensive
Profile Wiggo
Avatar

Send message
Joined: 24 Jan 00
Posts: 34744
Credit: 261,360,520
RAC: 489
Australia
Message 1297818 - Posted: 22 Oct 2012, 8:24:35 UTC - in response to Message 1297773.  

The thing with the server side fix, which has been pointed out before and I've seen in action, is that once a person is blocked/restricted then they can just detach and then re-attach which just resets their corruption again with a new PC ID.

Cheers.
ID: 1297818 · Report as offensive
Profile Tron

Send message
Joined: 16 Aug 09
Posts: 180
Credit: 2,250,468
RAC: 0
United States
Message 1297821 - Posted: 22 Oct 2012, 8:53:27 UTC - in response to Message 1297818.  

The thing with the server side fix, which has been pointed out before and I've seen in action, is that once a person is blocked/restricted then they can just detach and then re-attach which just resets their corruption again with a new PC ID.

Cheers.

Well thats just malicious and a reasonable justification for banning.
ID: 1297821 · Report as offensive
Profile Wiggo
Avatar

Send message
Joined: 24 Jan 00
Posts: 34744
Credit: 261,360,520
RAC: 489
Australia
Message 1297830 - Posted: 22 Oct 2012, 9:55:02 UTC - in response to Message 1297821.  

The thing with the server side fix, which has been pointed out before and I've seen in action, is that once a person is blocked/restricted then they can just detach and then re-attach which just resets their corruption again with a new PC ID.

Cheers.

Well thats just malicious and a reasonable justification for banning.

And how are you suppose to ban someone when I just pointed out the way that they are getting around it?

I don't want them banned, I want them fixed, and that is why I'll keep sending out PM's.

Cheers.
ID: 1297830 · Report as offensive
Terror Australis
Volunteer tester

Send message
Joined: 14 Feb 04
Posts: 1817
Credit: 262,693,308
RAC: 44
Australia
Message 1297871 - Posted: 22 Oct 2012, 13:09:44 UTC - in response to Message 1297771.  
Last modified: 22 Oct 2012, 13:10:10 UTC

ouch ... just noticed another wave of inconclusive ... a quick look at the wingmen ... gtx 560 ti ....

Is there a thread where you guys are keeping track of who has been PMed ?

one to add to your list : 9302233 (Murat Adas) PM sent 9pm oct 21


It's "kind of" being tracked in the "Validation Inconclusive" thread. Maybe you could post this in there.

T.A.
ID: 1297871 · Report as offensive
Terror Australis
Volunteer tester

Send message
Joined: 14 Feb 04
Posts: 1817
Credit: 262,693,308
RAC: 44
Australia
Message 1297873 - Posted: 22 Oct 2012, 13:12:43 UTC - in response to Message 1297818.  

The thing with the server side fix, which has been pointed out before and I've seen in action, is that once a person is blocked/restricted then they can just detach and then re-attach which just resets their corruption again with a new PC ID.

Cheers.

Hopefully, if they are savvy enough to do that they are probably savvy enough to sort out their computer before they re-attach.

T.A.
ID: 1297873 · Report as offensive
Profile Slavac
Volunteer tester
Avatar

Send message
Joined: 27 Apr 11
Posts: 1932
Credit: 17,952,639
RAC: 0
United States
Message 1297880 - Posted: 22 Oct 2012, 14:24:07 UTC - in response to Message 1297871.  

ouch ... just noticed another wave of inconclusive ... a quick look at the wingmen ... gtx 560 ti ....

Is there a thread where you guys are keeping track of who has been PMed ?

one to add to your list : 9302233 (Murat Adas) PM sent 9pm oct 21


It's "kind of" being tracked in the "Validation Inconclusive" thread. Maybe you could post this in there.

T.A.


Fancy starting up a new thread where we can track folks and try to get them some PC help?


Executive Director GPU Users Group Inc. -
brad@gpuug.org
ID: 1297880 · Report as offensive
Profile David Anderson (not *that* DA) Project Donor
Avatar

Send message
Joined: 5 Dec 09
Posts: 215
Credit: 74,008,558
RAC: 74
United States
Message 1297904 - Posted: 22 Oct 2012, 16:07:47 UTC

One of my wingmen, DG71, has
valid 69
invalid 319
inconclusive 1138.

That does not seem too good.

NVIDIA but NOT a 560, a 580.
I sent a PM to DG71 just now.
ID: 1297904 · 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 1297913 - Posted: 22 Oct 2012, 16:57:50 UTC - in response to Message 1297769.  
Last modified: 22 Oct 2012, 17:02:37 UTC

Work should not be distributed to machines with a

Valid : Invalid ratio of 50/50 or worse
This would allow wiggle room for those of us who do pay attention to our rigs, to find and resolve problems without interrupting the work flow.
While at the same time, cutting off the jokers running on auto pilot before they have a chance to produce truck loads of bogus work.

I don't understand, I guess, why the throttling system supposedly in place does not stop them before they amass hundreds or thousands of invalid results.
Maybe it needs a bit of tweaking....
I mean, I would not get ticked off one bit if the servers cut one of my rigs off should a GPU go wonky.

Of course, if a good running rig with a big cache goes south, you are gonna have a ton of bad results as it plows through the cache. The kitties have seen that happen here once or twice. Main thing is that the servers pick up on that fact and limit the damage to work already issued.


The server code only recognizes tasks that error out as 'Errors', while tasks that fail validation at a later date (could be up to a month or two after the task is submitted) aren't counted as 'Errors'.

Well, perhaps a large percentage of invalids SHOULD trigger some throttling too.
Not the validation inconclusives, as they may not be the fault of the host rig, as we have seen. But if a rig starts to accumulate a large percentage of invalid results, I agree that work sent to them should be slowed to a trickle.
"Freedom is just Chaos, with better lighting." Alan Dean Foster

ID: 1297913 · Report as offensive
1 · 2 · Next

Message boards : Number crunching : Rogue 560 Ti's


 
©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.