Older Nvidia cards deprecated?


log in

Advanced search

Message boards : Number crunching : Older Nvidia cards deprecated?

Author Message
Wiyosaya
Send message
Joined: 19 May 99
Posts: 39
Credit: 804,526
RAC: 18
United States
Message 1385997 - Posted: 30 Jun 2013, 3:47:57 UTC
Last modified: 30 Jun 2013, 3:55:08 UTC

Since the latest version of S@H hit the streets, my 8800 GT has been unable to successfully complete any work units. For example -

http://setiathome.berkeley.edu/result.php?resultid=3044437557

Was this planned or is it an unintended consequence?

Thanks.

PS - I should mention that my other two hosts, one a GTX 460 and the other a GTX 580 have had no problems since the update. However, the last good one I returned on the 8800 GT was on May 11, 2013.
____________

Profile Donald L. JohnsonProject donor
Avatar
Send message
Joined: 5 Aug 02
Posts: 6263
Credit: 739,741
RAC: 1,175
United States
Message 1386003 - Posted: 30 Jun 2013, 3:57:31 UTC - in response to Message 1385997.
Last modified: 30 Jun 2013, 4:00:17 UTC

Since the latest version of S@H hit the streets, my 8800 GT has been unable to successfully complete any work units. For example -

http://setiathome.berkeley.edu/result.php?resultid=3044437557

Was this planned or is it an unintended consequence?

Thanks.

PS - I should mention that my other two hosts, one a GTX 460 and the other a GTX 580 have had no problems since the update. However, the last good one I returned on the 8800 GT was on May 11, 2013.

Made the link active.

Check out this thread .
____________
Donald
Infernal Optimist / Submariner, retired

Profile BilBg
Volunteer tester
Avatar
Send message
Joined: 27 May 07
Posts: 2806
Credit: 6,329,590
RAC: 7,466
Bulgaria
Message 1386006 - Posted: 30 Jun 2013, 4:05:35 UTC - in response to Message 1385997.


The reason for the error is not the GPU 'version'

If you want to read about the reason:
"v7 cuda23 WUs getting ERR_TOO_MANY_EXITS"
http://setiathome.berkeley.edu/forum_thread.php?id=71953

Some posts in the same thread:
http://setiathome.berkeley.edu/forum_thread.php?id=71953&postid=1379544#1379544
http://setiathome.berkeley.edu/forum_thread.php?id=71953&postid=1383240#1383240


____________



- ALF - "Find out what you don't do well ..... then don't do it!" :)

Profile Jeff Buck
Send message
Joined: 11 Feb 00
Posts: 360
Credit: 36,386,752
RAC: 98,950
United States
Message 1386020 - Posted: 30 Jun 2013, 4:37:24 UTC - in response to Message 1385997.

As noted by the two previous replies, this ongoing problem is discussed in:

http://setiathome.berkeley.edu/forum_thread.php?id=71953

The short version of all that is that you can temporarily get around the problem by manually deleting the cudart.dll and cufft.dll files from the S@H project directory. However, unless you're running 24/7, you'll have to delete them again every time you reboot, since each time BOINC restarts, it notices that the files are missing and quickly downloads them again!

Profile Alex Storey
Volunteer tester
Avatar
Send message
Joined: 14 Jun 04
Posts: 553
Credit: 1,667,638
RAC: 575
Greece
Message 1386087 - Posted: 30 Jun 2013, 11:32:37 UTC - in response to Message 1386020.

As noted by the two previous replies, this ongoing problem is discussed in:

http://setiathome.berkeley.edu/forum_thread.php?id=71953

The short version of all that is that you can temporarily get around the problem by manually deleting the cudart.dll and cufft.dll files from the S@H project directory. However, unless you're running 24/7, you'll have to delete them again every time you reboot, since each time BOINC restarts, it notices that the files are missing and quickly downloads them again!


Eh? Did you disable V6 in prefs?

Mine appears to be working fine (so far). Maybe I never restarted my notebook... I'll do that now just to make sure.

ClaggyProject donor
Volunteer tester
Send message
Joined: 5 Jul 99
Posts: 4141
Credit: 33,627,602
RAC: 28,097
United Kingdom
Message 1386092 - Posted: 30 Jun 2013, 11:44:34 UTC - in response to Message 1386087.
Last modified: 30 Jun 2013, 11:45:54 UTC

As noted by the two previous replies, this ongoing problem is discussed in:

http://setiathome.berkeley.edu/forum_thread.php?id=71953

The short version of all that is that you can temporarily get around the problem by manually deleting the cudart.dll and cufft.dll files from the S@H project directory. However, unless you're running 24/7, you'll have to delete them again every time you reboot, since each time BOINC restarts, it notices that the files are missing and quickly downloads them again!


Eh? Did you disable V6 in prefs?

Of cause, if you disable MB v6 (SETI@home Enhanced) in the preferences you then can't get the 6.11 and now 6.12 (cuda)/(cuda22)/(cuda23) apps Eric is sending to try and get around the dll problem.

Setiathome applications

Claggy

Profile Alex Storey
Volunteer tester
Avatar
Send message
Joined: 14 Jun 04
Posts: 553
Credit: 1,667,638
RAC: 575
Greece
Message 1386095 - Posted: 30 Jun 2013, 12:04:29 UTC - in response to Message 1386092.

Not sure if I'm understanding you Claggy...

Do you want me to re-enable V6 for troubleshooting?

Juha
Volunteer tester
Send message
Joined: 7 Mar 04
Posts: 184
Credit: 143,295
RAC: 42
Finland
Message 1386239 - Posted: 30 Jun 2013, 23:09:20 UTC - in response to Message 1386095.

Not sure if I'm understanding you Claggy...

Do you want me to re-enable V6 for troubleshooting?

If you have never received tasks for 6.08 or
If you have since then reseted the project

You don't need to do anything else, your system doesn't have the problem causing DLLs anymore.

For those who have received tasks for 6.08 and have not reseted the project the best course of action is to set No new tasks, complete, upload & report any work on-board, reset project and Allow new tasks.

The server isn't sending the problem DLLs anymore so in your case (you reseted the project, right?) enabling or disabling v6 doesn't make any difference.

However, if you are running stock apps and have NVIDIA GPU you might want to not select v6 work. Both v6.11 and v6.12 are trashing work. v6.11 came with wrong app_version block in scheduler reply which caused BOINC to reject the tasks immediately. v6.12 comes with good app_version block. But v6.12 is really 6.03 CPU app in disguise (v6.11 was too), and for GPU apps BOINC adds --device N parameter to apps command line to select used GPU, and being a CPU app v6.03 doesn't know what that parameter means so it exits with error.

Wiyosaya
Send message
Joined: 19 May 99
Posts: 39
Credit: 804,526
RAC: 18
United States
Message 1386655 - Posted: 2 Jul 2013, 0:01:21 UTC
Last modified: 2 Jul 2013, 0:06:58 UTC

Thanks for the replies, all.

I've got an E@H Persus Arm Survey WU that has about 3-hours to go on that GPU right now, so I hit "reset project" on S@H since that sounds like part of the answer to the problem.

OK - so changing preference to exclued V6 work no longer seems to be an option as the only options that are available for me are the following:

SETI@home Enhanced
SETI@home v7
AstroPulse v6

Just in case SETI@home Enhanced is the "V6" app, I shut that off.

I'll post the results either way.
____________

Profile HAL9000
Volunteer tester
Avatar
Send message
Joined: 11 Sep 99
Posts: 4443
Credit: 119,158,636
RAC: 139,853
United States
Message 1386819 - Posted: 2 Jul 2013, 14:09:37 UTC - in response to Message 1386655.

Thanks for the replies, all.

I've got an E@H Persus Arm Survey WU that has about 3-hours to go on that GPU right now, so I hit "reset project" on S@H since that sounds like part of the answer to the problem.

OK - so changing preference to exclued V6 work no longer seems to be an option as the only options that are available for me are the following:

SETI@home Enhanced
SETI@home v7
AstroPulse v6

Just in case SETI@home Enhanced is the "V6" app, I shut that off.

I'll post the results either way.

SETI@home Enhanced contains the 6.xx app version. So that was the correct one.
____________
SETI@home classic workunits: 93,865 CPU time: 863,447 hours

Join the BP6/VP6 User Group today!

Wiyosaya
Send message
Joined: 19 May 99
Posts: 39
Credit: 804,526
RAC: 18
United States
Message 1386971 - Posted: 3 Jul 2013, 1:34:36 UTC

Thanks for the help, all.

My 8800 GT has successfully returned 4 WUs now. Back in business. :)


____________

Message boards : Number crunching : Older Nvidia cards deprecated?

Copyright © 2014 University of California