Bug in server affecting older BOINC clients with NVIDIA GPUs.


log in

Advanced search

Message boards : News : Bug in server affecting older BOINC clients with NVIDIA GPUs.

Previous · 1 · 2 · 3 · 4 · 5 · 6 . . . 7 · Next
Author Message
rynoski
Send message
Joined: 12 Mar 06
Posts: 8
Credit: 159,559
RAC: 9
United States
Message 1273593 - Posted: 21 Aug 2012, 3:14:13 UTC - in response to Message 1272118.

I have a rig with an NVIDIA GPU and I noticed this problem like at least one BOINC version ago. In my case the GPU would overheat and shutdown the machine. I disabled the GPU computing on my settings as a result. It seems to be more manageable if you limit the time your BOINC software runs. During the day was the worst time and it got a little better duing the overnight hours. Eventually though it will overheat or crash the GPU. A little hit and miss. I just switch to CPU only until the bug was fixed. Guess I should have said something sooner. I've noticed this bug for at the very least 6-8 months.

-Ryno
____________

EldRick
Send message
Joined: 21 Jan 11
Posts: 4
Credit: 23,989
RAC: 0
United States
Message 1273594 - Posted: 21 Aug 2012, 3:19:23 UTC

Over the past ten years when I've run S@H, the uptime track record has simply been miserable, in my opinion.

It seems that every time I've tried to run the software for a while, some component or other has been down for one reason or another within a few days, ranging from SW bugs to power outages (third-world electric company?).

IMHO, you'd find the signals a LOT faster, and have more participants if the overall system had a higher level of availability.

Eric KorpelaProject donor
Volunteer moderator
Project administrator
Project developer
Project scientist
Avatar
Send message
Joined: 3 Apr 99
Posts: 1088
Credit: 8,975,826
RAC: 12,899
United States
Message 1273611 - Posted: 21 Aug 2012, 5:30:32 UTC - in response to Message 1273594.
Last modified: 21 Aug 2012, 5:33:57 UTC

That's probably true. But we'd need more than four part time employees. A budget of 1.5 million dollars a year would help a lot towards getting 97% uptime. We'd really like a time machine to allow us to go back and put more effort into convincing the State of California that saving money by forcing the people who understood the campus power system to take early retirement was a bad idea.

Unfortunately none of those things is going to happen, unless you happen to have a billionaire in your pocket. Uptime costs money, salaries cost money, network bandwidth costs money. We don't have money. It's amazing what you can do without money, but it's equally amazing what you can't do.
____________

Profile Willi Kretzschmar
Volunteer tester
Send message
Joined: 23 Mar 00
Posts: 5
Credit: 261,875
RAC: 73
Germany
Message 1273613 - Posted: 21 Aug 2012, 5:35:23 UTC

I think, I have the same problem?
But we have 4 days ago since this "news" about "older BOINC clients with NVIDIA GPUs.
What can I do to normal function of Seti@home?

Greatings from Germany
Willi Kretzschmar
____________

rob smithProject donor
Volunteer tester
Send message
Joined: 7 Mar 03
Posts: 8422
Credit: 57,465,436
RAC: 74,956
United Kingdom
Message 1273614 - Posted: 21 Aug 2012, 5:46:13 UTC

Morning Eric, Thank you (and the rest of the lab team) for your efforts.



Now how about you get some sleep or I'll set Angela onto you ;-)
____________
Bob Smith
Member of Seti PIPPS (Pluto is a Planet Protest Society)
Somewhere in the (un)known Universe?

EldRick
Send message
Joined: 21 Jan 11
Posts: 4
Credit: 23,989
RAC: 0
United States
Message 1273617 - Posted: 21 Aug 2012, 6:19:05 UTC

I understand some of the problems, and I do appreciate all your efforts.

I've been excited about the project for many years, and I'd love to be able to crank out the 50-60 WUs/day that my system could deliver, but when it's this flaky over and over again I'm out of here - the annoyance level is just too high for me.

Perhaps I'll give it a whirl in another few months. Meanwhile, my processor will be doing Einstein@ and Collatz@ instead.

Grant (SSSF)
Send message
Joined: 19 Aug 99
Posts: 5831
Credit: 59,408,427
RAC: 47,478
Australia
Message 1273619 - Posted: 21 Aug 2012, 6:23:48 UTC - in response to Message 1273594.


IMHO, you'd find the signals a LOT faster, and have more participants if the overall system had a higher level of availability.

Not really, most people run with a cache.
As long as the system comes back up before their cache runs dry, there is no impact on the work being done.
____________
Grant
Darwin NT.

OzzFan
Volunteer tester
Avatar
Send message
Joined: 9 Apr 02
Posts: 13619
Credit: 30,537,996
RAC: 21,191
United States
Message 1273697 - Posted: 21 Aug 2012, 11:47:33 UTC - in response to Message 1273617.

Why not let BOINC do what it was designed to do? Let it grab work from SETI when it has the ability, and let it grab work from other projects when SETI is down.

I don't see a reason to quite a project over a human emotion such as frustration when the BOINC software doesn't get frustrated. I've been with this project since 2002 and I've seen quite a few outages, but I'm still here.

Profile Marcus
Send message
Joined: 22 Apr 02
Posts: 2
Credit: 2,661,881
RAC: 729
United Kingdom
Message 1273907 - Posted: 22 Aug 2012, 8:59:22 UTC - in response to Message 1271877.

of course they do they use more resources depends how much RAM and how big your Gpu card and cpus r as this also is a factor i have multi core processors and Seti runs all the time i also play World of Warcraft with the gpu running on ultra with out any noticeable problems hope this helps
____________

Profile Interstel
Avatar
Send message
Joined: 29 Nov 01
Posts: 23
Credit: 2,222,625
RAC: 0
United States
Message 1274093 - Posted: 22 Aug 2012, 19:43:14 UTC

Have the new Nvidia APPs gone out? What would be the version number to tell if you had gotten them?

James

ps I feel you guys and the power outages every day here I have to worry about lightning and thunderstorms taking down either the power or the T1's.
____________

Joined SETI@Home in 2001
Online since ArpNET days
First activity on Honeywell 1648
Series Mainframe in 1975 at age 12.

ClaggyProject donor
Volunteer tester
Send message
Joined: 5 Jul 99
Posts: 4101
Credit: 33,140,511
RAC: 8,433
United Kingdom
Message 1274095 - Posted: 22 Aug 2012, 19:55:45 UTC - in response to Message 1274093.

Have the new Nvidia APPs gone out? What would be the version number to tell if you had gotten them?

Yes, see the Setiathome applications page.

Claggy

Eric KorpelaProject donor
Volunteer moderator
Project administrator
Project developer
Project scientist
Avatar
Send message
Joined: 3 Apr 99
Posts: 1088
Credit: 8,975,826
RAC: 12,899
United States
Message 1274108 - Posted: 22 Aug 2012, 20:45:30 UTC - in response to Message 1274095.

The NVIDIA version for BOINC v7+ is out. The v6 version will probably go out on Monday.
____________

Profile Len
Avatar
Send message
Joined: 15 Mar 10
Posts: 53
Credit: 1,950,827
RAC: 1,421
United Kingdom
Message 1274255 - Posted: 23 Aug 2012, 6:48:49 UTC

This explains the behaviour of my EEEBox starting to overheat. I shall await an update.

Len
____________
I think I am. Therefore I am. I think.

Profile Interstel
Avatar
Send message
Joined: 29 Nov 01
Posts: 23
Credit: 2,222,625
RAC: 0
United States
Message 1274268 - Posted: 23 Aug 2012, 7:38:22 UTC - in response to Message 1274108.

Well checking my directory with SETI definitions under BOINC. I find that the dates and times do not match the ones listed in the link. But I do find the following 2 entries are not present in any kind of wording....

6.10 (cuda_fermi)

6.04 (opencl_nvidia_100)

And I guess I had some stuff not used anymore like astropulse 5.00, 5.03. And at least twice applications that seemed to have 2 slightly different versions such as

setiathome_6.03_windows_intelx86.exe

setigraphics_6.03_windows_intelx86.exe

James
____________

Joined SETI@Home in 2001
Online since ArpNET days
First activity on Honeywell 1648
Series Mainframe in 1975 at age 12.

Profile BilBg
Volunteer tester
Avatar
Send message
Joined: 27 May 07
Posts: 2711
Credit: 6,139,770
RAC: 5,485
Bulgaria
Message 1274273 - Posted: 23 Aug 2012, 7:56:20 UTC - in response to Message 1274255.
Last modified: 23 Aug 2012, 8:02:44 UTC

This explains the behaviour of my EEEBox starting to overheat. I shall await an update.

Len

Which "This"?
What "update"?

* If you talk about "Bug in server ..." it was already fixed (read the whole thread).
And all your CUDA tasks on "EEEBox" are from 2 Aug 2012 when this bug (17 Aug 2012) was not yet introduced.
http://setiathome.berkeley.edu/results.php?hostid=6203689&offset=0&show_names=0&state=1&appid=2

And you use BOINC 7.0.28 which was never affected by this bug.


* If you talk about "(AstroPulse) NVIDIA version for BOINC v7+ is out"
- you don't have any GPU AstroPulse tasks on "EEEBox" (http://en.wikipedia.org/wiki/Asus_EeeBox_PC)
http://setiathome.berkeley.edu/results.php?hostid=6203689&offset=0&show_names=0&state=0&appid=12


____________



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

Profile BilBg
Volunteer tester
Avatar
Send message
Joined: 27 May 07
Posts: 2711
Credit: 6,139,770
RAC: 5,485
Bulgaria
Message 1274309 - Posted: 23 Aug 2012, 9:25:22 UTC - in response to Message 1274268.

I find that the dates and times do not match the ones listed in the link.

Normal:
- Your time (file times) is not in UTC
- You finish their download later than server sends (assigns) them to you


But I do find the following 2 entries are not present in any kind of wording....
6.10 (cuda_fermi)

You have NVIDIA GeForce GTS 250 which is not fermi, you have tasks marked "SETI@home Enhanced v6.09 (cuda23)" and "SETI@home Enhanced v6.08 (cuda)":
http://setiathome.berkeley.edu/results.php?hostid=3691074&offset=0&show_names=0&state=2&appid=

They are processed by:
setiathome_6.08_windows_intelx86__cuda.exe
setiathome_6.09_windows_intelx86__cuda23.exe

(and several .dll files)


6.04 (opencl_nvidia_100)

You still do not have AstroPulse tasks allocated to the GPU, the app will be downloaded with the first AstroPulse task for the GPU:
http://setiathome.berkeley.edu/results.php?hostid=3691074&offset=0&show_names=0&state=0&appid=12


And I guess I had some stuff not used anymore like astropulse 5.00, 5.03.

Since all the apps (.exe .dll .cl and other types of files) consume very little HDD space - better do not delete things you are not sure what they are (e.g. .DLLs)


And at least twice applications that seemed to have 2 slightly different versions such as

setiathome_6.03_windows_intelx86.exe
setigraphics_6.03_windows_intelx86.exe

Do not delete them - both are standard current apps (setigraphics used for "Show graphics" and screensaver)


____________



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

robert spry
Send message
Joined: 24 May 99
Posts: 9
Credit: 5,587,820
RAC: 6,173
United States
Message 1274344 - Posted: 23 Aug 2012, 11:10:46 UTC

Not sure where to post this problem:

8/23/2012 7:02:57 AM | SETI@home | Sending scheduler request: Requested by user.
8/23/2012 7:02:57 AM | SETI@home | Requesting new tasks for CPU and ATI
8/23/2012 7:02:58 AM | | App version needs OpenCL but GPU doesn't support it
8/23/2012 7:02:58 AM | SETI@home | Scheduler request completed: got 1 new tasks
8/23/2012 7:02:58 AM | SETI@home | Resent lost task ap_24mr10ac_B6_P0_00269_20120701_03541.wu_3
8/23/2012 7:02:58 AM | SETI@home | Project has no tasks available
8/23/2012 7:02:58 AM | SETI@home | [error] App version uses non-existent ATI GPU
8/23/2012 7:02:58 AM | SETI@home | [error] Missing coprocessor for task ap_24mr10ac_B6_P0_00269_20120701_03541.wu_3; aborting

I'm not receiving tasks?
I'm running 7.0.28 (x86) Am I doing something wrong?

____________

Profile BilBg
Volunteer tester
Avatar
Send message
Joined: 27 May 07
Posts: 2711
Credit: 6,139,770
RAC: 5,485
Bulgaria
Message 1274352 - Posted: 23 Aug 2012, 11:51:17 UTC - in response to Message 1274344.
Last modified: 23 Aug 2012, 12:05:03 UTC

8/23/2012 7:02:58 AM | | App version needs OpenCL but GPU doesn't support it

Where did you get the drivers?
Use AMD/ATI drivers, Not Microsoft/Windows drivers.

As reference use this list (mind the red warning about XP, for XP get 11.12 or 12.1):
http://www.hal6000.com/seti/boinc_ati_gpu_cheat_sheet.htm

(you now use on the 2 computers: 11.10 and 11.11 - both have "high CPU usage bug" even if you manage to enable (install) OpenCL on these Catalyst versions)

GPU-Z shows is OpenCL driver installed/enabled
http://www.techpowerup.com/gpuz/

***

Another possibility is you use Remote Desktop (which of course replaces video drivers with generic and stops GPU computing).
Use something like TeamViewer instead.


____________



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

ClaggyProject donor
Volunteer tester
Send message
Joined: 5 Jul 99
Posts: 4101
Credit: 33,140,511
RAC: 8,433
United Kingdom
Message 1274454 - Posted: 23 Aug 2012, 16:41:50 UTC - in response to Message 1274344.
Last modified: 23 Aug 2012, 16:42:42 UTC

Not sure where to post this problem:

8/23/2012 7:02:57 AM | SETI@home | Sending scheduler request: Requested by user.
8/23/2012 7:02:57 AM | SETI@home | Requesting new tasks for CPU and ATI
8/23/2012 7:02:58 AM | | App version needs OpenCL but GPU doesn't support it
8/23/2012 7:02:58 AM | SETI@home | Scheduler request completed: got 1 new tasks
8/23/2012 7:02:58 AM | SETI@home | Resent lost task ap_24mr10ac_B6_P0_00269_20120701_03541.wu_3
8/23/2012 7:02:58 AM | SETI@home | Project has no tasks available
8/23/2012 7:02:58 AM | SETI@home | [error] App version uses non-existent ATI GPU
8/23/2012 7:02:58 AM | SETI@home | [error] Missing coprocessor for task ap_24mr10ac_B6_P0_00269_20120701_03541.wu_3; aborting

I'm not receiving tasks?
I'm running 7.0.28 (x86) Am I doing something wrong?

That is a Server Bug, If the ATI GPU doesn't have OpenCL support, the Server shouldn't send OpenCL work for it, (I had the same at Seti Beta with a Nvidia GPU that was running pre-OpenCL drivers, but since fixed)

Claggy

Eric KorpelaProject donor
Volunteer moderator
Project administrator
Project developer
Project scientist
Avatar
Send message
Joined: 3 Apr 99
Posts: 1088
Credit: 8,975,826
RAC: 12,899
United States
Message 1274461 - Posted: 23 Aug 2012, 17:02:28 UTC - in response to Message 1274344.

I'm looking for this in the logs now. This is probably an artifact of the way are BOINC 6 OpenCL kludge is implemented...

Not sure where to post this problem:

8/23/2012 7:02:57 AM | SETI@home | Sending scheduler request: Requested by user.
8/23/2012 7:02:57 AM | SETI@home | Requesting new tasks for CPU and ATI
8/23/2012 7:02:58 AM | | App version needs OpenCL but GPU doesn't support it
8/23/2012 7:02:58 AM | SETI@home | Scheduler request completed: got 1 new tasks
8/23/2012 7:02:58 AM | SETI@home | Resent lost task ap_24mr10ac_B6_P0_00269_20120701_03541.wu_3
8/23/2012 7:02:58 AM | SETI@home | Project has no tasks available
8/23/2012 7:02:58 AM | SETI@home | [error] App version uses non-existent ATI GPU
8/23/2012 7:02:58 AM | SETI@home | [error] Missing coprocessor for task ap_24mr10ac_B6_P0_00269_20120701_03541.wu_3; aborting

I'm not receiving tasks?
I'm running 7.0.28 (x86) Am I doing something wrong?


____________

Previous · 1 · 2 · 3 · 4 · 5 · 6 . . . 7 · Next

Message boards : News : Bug in server affecting older BOINC clients with NVIDIA GPUs.

Copyright © 2014 University of California