SETI applications for NVIDIA GPU improvement - how you can help

Message boards : Number crunching : SETI applications for NVIDIA GPU improvement - how you can help
Message board moderation

To post messages, you must log in.

Previous · 1 · 2 · 3 · 4 · 5 . . . 13 · Next

AuthorMessage
Stephen "Heretic"Project Donor
Volunteer tester
Avatar

Send message
Joined: 20 Sep 12
Posts: 2628
Credit: 47,935,377
RAC: 129,508
Australia
Message 1796328 - Posted: 15 Jun 2016, 12:11:30 UTC - in response to Message 1796312.  


. . The new command line was : -use_sleep_ex 1 -sbs 384 -v

Option to try is -v 6 .
6 is important here.


. . Sorry but I was hoping that it would speed things up enough to get more info.

Nope. Debug/verbose build will speedup things ultimately but because of info and new optimizations I can gather from its output, not because it faster per se. Increased verbosity comes with performance degradation - rule of thumb (for human operation too btw ;) ).

BTW, most likely failures in task come from not correct -v option use.
Will make it more foolproof as result.


. . OK, I get it I am a chatterbox :)

. . I will try again with that change to the command line.

. . But I meant I thought that adding the -sbs 384 would speed it up, I thought I had caused the error crashes by using that. I never suspected the -v command because of a missing '6'. :(
ID: 1796328 · Report as offensive     Reply Quote
Stephen "Heretic"Project Donor
Volunteer tester
Avatar

Send message
Joined: 20 Sep 12
Posts: 2628
Credit: 47,935,377
RAC: 129,508
Australia
Message 1796333 - Posted: 15 Jun 2016, 12:29:55 UTC - in response to Message 1796312.  
Last modified: 15 Jun 2016, 13:23:47 UTC

. . I have not found a host for text files but someone pointed me to this site for images:

. . First image -

https://s31.postimg.org/aldaojnff/Change_CUDA50to_So_G_r3472.png

. . [edit]This is the changeover from nonVLAR to Guppie running SoG

. . Second Image -

https://s32.postimg.org/4nce3zuut/Test_Guppi1.png

. . [edit]This is the early stage of 1st Guppi

. . Third image -

https://s32.postimg.org/jk1m7dbc5/Test_Guppi1_full_flight.png

. . This is 1st Guppi still in early stages, seems OK

. . Fourth image -

https://s32.postimg.org/3nq82xyid/Test_Guppi1_75percent.png

. . At this point Guppi has been running for 2 to 3 hours and I expected it to finish within the next hour or so, but it took another 4 hours.

. . [edit]Notice that the GPU and Frame Buffer usage is lower than with nonVLAR task, and the temp is down nearly 20 deg as a result. But surprisingly the usage seems to be dropping even further at the 75% mark which seems to explain why it still took another 4 hours to complete.

.
ID: 1796333 · Report as offensive     Reply Quote
Profile Raistmer
Volunteer developer
Volunteer tester
Avatar

Send message
Joined: 16 Jun 01
Posts: 5804
Credit: 75,942,444
RAC: 50,427
Russia
Message 1796342 - Posted: 15 Jun 2016, 12:53:26 UTC - in response to Message 1796333.  

Could you archive stderr.txt from corresponding slot and send a link to it for download along with link to completed result on webpage.
SETI apps news
We're not gonna fight them. We're gonna transcend them.
ID: 1796342 · Report as offensive     Reply Quote
Stephen "Heretic"Project Donor
Volunteer tester
Avatar

Send message
Joined: 20 Sep 12
Posts: 2628
Credit: 47,935,377
RAC: 129,508
Australia
Message 1796347 - Posted: 15 Jun 2016, 13:15:54 UTC - in response to Message 1796342.  

Could you archive stderr.txt from corresponding slot and send a link to it for download along with link to completed result on webpage.


. . Here are some of the results of errored out tasks -

http://setiathome.berkeley.edu/result.php?resultid=4984162341

http://setiathome.berkeley.edu/result.php?resultid=4984162694

. . The one successful task -

http://setiathome.berkeley.edu/result.php?resultid=4984156421

. . I cannot identify the nonVLAR task that was running during the changeover.
ID: 1796347 · Report as offensive     Reply Quote
Stephen "Heretic"Project Donor
Volunteer tester
Avatar

Send message
Joined: 20 Sep 12
Posts: 2628
Credit: 47,935,377
RAC: 129,508
Australia
Message 1796349 - Posted: 15 Jun 2016, 13:17:24 UTC - in response to Message 1796342.  

Could you archive stderr.txt from corresponding slot and send a link to it for download along with link to completed result on webpage.


. . I am not sure how to link to an archive I create locally.
ID: 1796349 · Report as offensive     Reply Quote
Profile Raistmer
Volunteer developer
Volunteer tester
Avatar

Send message
Joined: 16 Jun 01
Posts: 5804
Credit: 75,942,444
RAC: 50,427
Russia
Message 1796426 - Posted: 15 Jun 2016, 19:51:03 UTC - in response to Message 1796349.  

Could you archive stderr.txt from corresponding slot and send a link to it for download along with link to completed result on webpage.


. . I am not sure how to link to an archive I create locally.

To upload it to some cloud storage of course.
SETI apps news
We're not gonna fight them. We're gonna transcend them.
ID: 1796426 · Report as offensive     Reply Quote
Stephen "Heretic"Project Donor
Volunteer tester
Avatar

Send message
Joined: 20 Sep 12
Posts: 2628
Credit: 47,935,377
RAC: 129,508
Australia
Message 1796493 - Posted: 16 Jun 2016, 2:59:01 UTC - in response to Message 1796426.  

Could you archive stderr.txt from corresponding slot and send a link to it for download along with link to completed result on webpage.


. . I am not sure how to link to an archive I create locally.

To upload it to some cloud storage of course.


. . OK I thought that is what I had to do.

. . Can you suggest one? The only one I know of only supports image files not text or archives.
ID: 1796493 · Report as offensive     Reply Quote
Grant (SSSF)
Volunteer tester

Send message
Joined: 19 Aug 99
Posts: 8877
Credit: 114,939,115
RAC: 69,600
Australia
Message 1796499 - Posted: 16 Jun 2016, 3:23:52 UTC - in response to Message 1796493.  

. . Can you suggest one? The only one I know of only supports image files not text or archives.

If you've got an Apple account or Microsoft account both of them provide free online storage.
I think Google also have some offerings if you've got a Google account (eg Gmail).
Or there's Dropbox.
Grant
Darwin NT
ID: 1796499 · Report as offensive     Reply Quote
Stephen "Heretic"Project Donor
Volunteer tester
Avatar

Send message
Joined: 20 Sep 12
Posts: 2628
Credit: 47,935,377
RAC: 129,508
Australia
Message 1796526 - Posted: 16 Jun 2016, 5:20:20 UTC - in response to Message 1796499.  
Last modified: 16 Jun 2016, 5:21:03 UTC

. . Can you suggest one? The only one I know of only supports image files not text or archives.

If you've got an Apple account or Microsoft account both of them provide free online storage.
I think Google also have some offerings if you've got a Google account (eg Gmail).
Or there's Dropbox.


. . Thanks I will try dropbox.
.
ID: 1796526 · Report as offensive     Reply Quote
Stephen "Heretic"Project Donor
Volunteer tester
Avatar

Send message
Joined: 20 Sep 12
Posts: 2628
Credit: 47,935,377
RAC: 129,508
Australia
Message 1796529 - Posted: 16 Jun 2016, 5:30:17 UTC - in response to Message 1796342.  

Could you archive stderr.txt from corresponding slot and send a link to it for download along with link to completed result on webpage.



. . Second trial, they run OK for 140 mins (give or take one minute) then error out.


http://setiathome.berkeley.edu/result.php?resultid=4986875001

http://setiathome.berkeley.edu/result.php?resultid=4986873722

http://setiathome.berkeley.edu/result.php?resultid=4986874828

http://setiathome.berkeley.edu/result.php?resultid=4986874828

. . They get exit message of EXIT_DISK_LIMIT_EXCEEDED, but there is almost 2GB free on the SETI drive.
ID: 1796529 · Report as offensive     Reply Quote
Tutankhamon
Volunteer tester
Avatar

Send message
Joined: 1 Nov 08
Posts: 6704
Credit: 42,236,404
RAC: 13,448
Sweden
Message 1796535 - Posted: 16 Jun 2016, 6:06:05 UTC - in response to Message 1796529.  

Could you archive stderr.txt from corresponding slot and send a link to it for download along with link to completed result on webpage.



. . Second trial, they run OK for 140 mins (give or take one minute) then error out.


http://setiathome.berkeley.edu/result.php?resultid=4986875001

http://setiathome.berkeley.edu/result.php?resultid=4986873722

http://setiathome.berkeley.edu/result.php?resultid=4986874828

http://setiathome.berkeley.edu/result.php?resultid=4986874828

. . They get exit message of EXIT_DISK_LIMIT_EXCEEDED, but there is almost 2GB free on the SETI drive.


Check your preferences for disk size settings (Disk: use at most _ GB; Disk: leave free at least; Disk: use at most _% of total)

Both on the web settings, and the Boinc manager settings (Depending upon if you use the local or the web settings)
Too much hormone treated meat.
Too much Monsanto veggies.
Too old and outdated constitution.
A crazy problem, as you Yanks use to say......

There is no God, and God never existed.
ID: 1796535 · Report as offensive     Reply Quote
Richard HaselgroveProject Donor
Volunteer tester

Send message
Joined: 4 Jul 99
Posts: 11516
Credit: 106,069,989
RAC: 70,652
United Kingdom
Message 1796550 - Posted: 16 Jun 2016, 6:53:38 UTC - in response to Message 1796535.  

. . They get exit message of EXIT_DISK_LIMIT_EXCEEDED, but there is almost 2GB free on the SETI drive.

Check your preferences for disk size settings (Disk: use at most _ GB; Disk: leave free at least; Disk: use at most _% of total)

It's also possible that you're breaking the disk resource limit set in each workunit:

<rsc_disk_bound>33554432.000000</rsc_disk_bound>

but that's 32 MB, so one heck of a long log.

Let one run for a couple of hours (so approaching the danger point), then cheack the file sizes in the slot directory.
ID: 1796550 · Report as offensive     Reply Quote
Tutankhamon
Volunteer tester
Avatar

Send message
Joined: 1 Nov 08
Posts: 6704
Credit: 42,236,404
RAC: 13,448
Sweden
Message 1796551 - Posted: 16 Jun 2016, 7:00:29 UTC - in response to Message 1796550.  

. . They get exit message of EXIT_DISK_LIMIT_EXCEEDED, but there is almost 2GB free on the SETI drive.

Check your preferences for disk size settings (Disk: use at most _ GB; Disk: leave free at least; Disk: use at most _% of total)

It's also possible that you're breaking the disk resource limit set in each workunit:

<rsc_disk_bound>33554432.000000</rsc_disk_bound>

but that's 32 MB, so one heck of a long log.

Let one run for a couple of hours (so approaching the danger point), then cheack the file sizes in the slot directory.

You're probably right, because the text of the stderr does not start in the usual way, with data about the GPU app and version numbers and so on, but goes directly to this:

<core_client_version>7.6.22</core_client_version>
<![CDATA[
<message>
Disk usage limit exceeded
</message>
<stderr_txt>
or completion
Partial PulseFind_3 (before buffer read): Awaited 1 iterations for completion
Partial PulseFind_3 (before buffer read): Awaited 1 iterations for completion
Partial PulseFind_3 (before buffer read): Awaited 1 iterations for completion
Too much hormone treated meat.
Too much Monsanto veggies.
Too old and outdated constitution.
A crazy problem, as you Yanks use to say......

There is no God, and God never existed.
ID: 1796551 · Report as offensive     Reply Quote
Richard HaselgroveProject Donor
Volunteer tester

Send message
Joined: 4 Jul 99
Posts: 11516
Credit: 106,069,989
RAC: 70,652
United Kingdom
Message 1796553 - Posted: 16 Jun 2016, 7:03:27 UTC - in response to Message 1796551.  

stderr transferred to the server is the final 64 KB of the file - but there's a big range between 64 KB and 32 MB. That's probably why Raistmer asked for it by dropbox - you can't paste 32 MB into a thread here, either.
ID: 1796553 · Report as offensive     Reply Quote
Stephen "Heretic"Project Donor
Volunteer tester
Avatar

Send message
Joined: 20 Sep 12
Posts: 2628
Credit: 47,935,377
RAC: 129,508
Australia
Message 1796579 - Posted: 16 Jun 2016, 12:42:11 UTC - in response to Message 1796535.  



Check your preferences for disk size settings (Disk: use at most _ GB; Disk: leave free at least; Disk: use at most _% of total)

Both on the web settings, and the Boinc manager settings (Depending upon if you use the local or the web settings)



. . They are on a 2GB flashdrive and have access to all of it, no limits.
ID: 1796579 · Report as offensive     Reply Quote
Stephen "Heretic"Project Donor
Volunteer tester
Avatar

Send message
Joined: 20 Sep 12
Posts: 2628
Credit: 47,935,377
RAC: 129,508
Australia
Message 1796580 - Posted: 16 Jun 2016, 12:45:12 UTC - in response to Message 1796550.  

. . They get exit message of EXIT_DISK_LIMIT_EXCEEDED, but there is almost 2GB free on the SETI drive.

Check your preferences for disk size settings (Disk: use at most _ GB; Disk: leave free at least; Disk: use at most _% of total)

It's also possible that you're breaking the disk resource limit set in each workunit:

<rsc_disk_bound>33554432.000000</rsc_disk_bound>

but that's 32 MB, so one heck of a long log.

Let one run for a couple of hours (so approaching the danger point), then cheack the file sizes in the slot directory.


. . I think that may be it. When I checked the stderr file on one WU shortly before it crashed out the log was already up to 25MB. Hmm Raistmer wanted verbosity set to 6, maybe I will need to run a lower value.
ID: 1796580 · Report as offensive     Reply Quote
Stephen "Heretic"Project Donor
Volunteer tester
Avatar

Send message
Joined: 20 Sep 12
Posts: 2628
Credit: 47,935,377
RAC: 129,508
Australia
Message 1796581 - Posted: 16 Jun 2016, 12:47:56 UTC - in response to Message 1796551.  


It's also possible that you're breaking the disk resource limit set in each workunit:

<rsc_disk_bound>33554432.000000</rsc_disk_bound>

but that's 32 MB, so one heck of a long log.

Let one run for a couple of hours (so approaching the danger point), then cheack the file sizes in the slot directory.

You're probably right, because the text of the stderr does not start in the usual way, with data about the GPU app and version numbers and so on, but goes directly to this:

<core_client_version>7.6.22</core_client_version>
<![CDATA[
<message>
Disk usage limit exceeded
</message>
<stderr_txt>
or completion
Partial PulseFind_3 (before buffer read): Awaited 1 iterations for completion
Partial PulseFind_3 (before buffer read): Awaited 1 iterations for completion
Partial PulseFind_3 (before buffer read): Awaited 1 iterations for completion


. . The question is can that limit be increased?

.
ID: 1796581 · Report as offensive     Reply Quote
Stephen "Heretic"Project Donor
Volunteer tester
Avatar

Send message
Joined: 20 Sep 12
Posts: 2628
Credit: 47,935,377
RAC: 129,508
Australia
Message 1796582 - Posted: 16 Jun 2016, 12:49:37 UTC - in response to Message 1796553.  

stderr transferred to the server is the final 64 KB of the file - but there's a big range between 64 KB and 32 MB. That's probably why Raistmer asked for it by dropbox - you can't paste 32 MB into a thread here, either.



. . That's for sure. I need to set up dropbox again, something went wrong with the first try.
ID: 1796582 · Report as offensive     Reply Quote
Richard HaselgroveProject Donor
Volunteer tester

Send message
Joined: 4 Jul 99
Posts: 11516
Credit: 106,069,989
RAC: 70,652
United Kingdom
Message 1796590 - Posted: 16 Jun 2016, 13:24:54 UTC - in response to Message 1796581.  

It's also possible that you're breaking the disk resource limit set in each workunit:

<rsc_disk_bound>33554432.000000</rsc_disk_bound>

but that's 32 MB, so one heck of a long log.

The question is can that limit be increased?

Yes, it can, but you're venturing further into advanced territory. Be careful, and follow the instructions exactly. If anything doesn't immediately feel comfortable, back off and revert any changes.

0) Be running a short cache when testing!
1) Quit BOINC entirely - that's "Shut down connected client", not just the Manager.
2) Locate the file client_state.xml in the BOINC data directory.
3) Right-click and 'Edit', or 'Open with --- Notepad'. Don't use an XML editor.
4) Open Search & Replace with ctrl+H
5) Paste the entire code line above into the find box.
6) Paste it again into the Replace with box.
7) Edit the Replace version with an extra digit before the decimal point, to make the number ten times bigger.
8) 'Replace all'
9) Save and close
10) Restart BOINC

If the tasks run for more than a day, they'll probably still fail. Rinse and repeat, this time multiplying the bound by 100. And so on.
ID: 1796590 · Report as offensive     Reply Quote
Profile BilBg
Volunteer tester
Avatar

Send message
Joined: 27 May 07
Posts: 3717
Credit: 8,884,789
RAC: 603
Bulgaria
Message 1796839 - Posted: 17 Jun 2016, 13:50:34 UTC - in response to Message 1796582.  

I need to set up dropbox again, something went wrong with the first try.

You may also use:
http://www.zippyshare.com/

Note:
Since zippyshare uses some misleading [Download] buttons (ads) on the resulting pages you (all) may want to first add uBlock Origin to your browser:
https://chrome.google.com/webstore/detail/ublock-origin/cjpalhdlnbpafiamejdnhcphjbkeiagm

https://addons.mozilla.org/en-US/firefox/addon/ublock-origin/



- ALF - "Find out what you don't do well ..... then don't do it!" :)
ID: 1796839 · Report as offensive     Reply Quote
Previous · 1 · 2 · 3 · 4 · 5 . . . 13 · Next

Message boards : Number crunching : SETI applications for NVIDIA GPU improvement - how you can help


 
©2017 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.