BOINC won't download more projects, and won't use 2nd GPU

Questions and Answers : GPU applications : BOINC won't download more projects, and won't use 2nd GPU
Message board moderation

To post messages, you must log in.

AuthorMessage
Aeon2058

Send message
Joined: 30 Jan 08
Posts: 1
Credit: 1,217
RAC: 0
United States
Message 947350 - Posted: 15 Nov 2009, 3:57:20 UTC
Last modified: 15 Nov 2009, 4:01:12 UTC

I have 2 9800GTX+s that I want to use to contribute, but for some reason BOINC is only using one (it recognizes both though). How do I get it to run on both? Also, when I attempted to use BOINC a while ago when I only had one GPU, it seemed to get projects done MUCH faster than it is now- something like <20 min per project; now it's estimating >50min, and my GPU temp is only ~5ºC above idle temp.

Also, in the messages tab in the BOINC manager, it keeps trying to fetch more work (I told it to keep 10 days' worth) for both CPU and GPU, but it keeps getting the error "SETI@home Message from server: SETI@home Enhanced needs 32.00MB more disk space. You currently have 0.00 MB available and it needs 32.00 MB." I have allocated 25GB of my HD to this, so what's wrong?

One other minor thing- I changed my name from Agni451 to Aeon2058 in my account page. Free-DC still thinks I'm Agni451; will that be updated sometime?
ID: 947350 · Report as offensive
Profile Gundolf Jahn

Send message
Joined: 19 Sep 00
Posts: 3184
Credit: 446,358
RAC: 0
Germany
Message 947419 - Posted: 15 Nov 2009, 8:44:30 UTC - in response to Message 947350.  

I have 2 9800GTX+s that I want to use to contribute, but for some reason BOINC is only using one (it recognizes both though). How do I get it to run on both?

Do a forum search (top left corner of this page) for the term use_all_gpus.

Also, when I attempted to use BOINC a while ago when I only had one GPU, it seemed to get projects done MUCH faster than it is now- something like <20 min per project; now it's estimating >50min, and my GPU temp is only ~5ºC above idle temp.

Perhaps your card suffers from memory congestion and the task falls back to the CPU, but since your computers are hidden, I can't tell.

Also, in the messages tab in the BOINC manager, it keeps trying to fetch more work (I told it to keep 10 days' worth) for both CPU and GPU, but it keeps getting the error "SETI@home Message from server: SETI@home Enhanced needs 32.00MB more disk space. You currently have 0.00 MB available and it needs 32.00 MB." I have allocated 25GB of my HD to this, so what's wrong?

Check your Computing preferences for the setting of
Use at most X GB disk space 
Leave at least Y GB disk space free 
Use at most N% of total disk space
and the Messages tab of BOINC manager after a reboot for
09/11/2009 11:26:42||Preferences limit disk usage to Z GB

One other minor thing- I changed my name from Agni451 to Aeon2058 in my account page. Free-DC still thinks I'm Agni451; will that be updated sometime?

I have no experience with Free-DC, but if it works as most statistic sites, it should adjust over time.

Gruß,
Gundolf
Computer sind nicht alles im Leben. (Kleiner Scherz)

SETI@home classic workunits 3,758
SETI@home classic CPU time 66,520 hours
ID: 947419 · Report as offensive
Aeon2058

Send message
Joined: 24 Jan 09
Posts: 9
Credit: 1,184,681
RAC: 0
United States
Message 947552 - Posted: 16 Nov 2009, 4:23:48 UTC

Okay, I found a copy of cc_config.xml, and I set the individual parts of it how I want, including using all GPUs. I put that xml file into C:\Program Data\BOINC\ while BOINC was shut down, but now BOINC won't connect at all.

Here is what my cc_config.xml file looks like:
<cc_config>
   <log_flags>
       <task>1</task>
       <file_xfer>1</file_xfer>
       <sched_ops>1</sched_ops>
       <state_debug>0</state_debug>
       <task_debug>0</task_debug>
       <file_xfer_debug>0</file_xfer_debug>
       <sched_op_debug>0</sched_op_debug>
       <http_debug>0</http_debug>
       <work_fetch_debug>0</work_fetch_debug>
       <unparsed_xml>0</unparsed_xml> 
       <proxy_debug>0</proxy_debug>
       <time_debug>0</time_debug>
       <http_xfer_debug>0</http_xfer_debug>
       <benchmark_debug>0</benchmark_debug>
       <poll_debug>0</poll_debug>
       <guirpc_debug>0</guirpc_debug>
       <scrsave_debug>0</scrsave_debug>
       <rr_simulation>0</rr_simulation>
       <cpu_sched>0</cpu_sched>
       <cpu_sched_debug>0</cpu_sched_debug>
       <app_msg_send>0</app_msg_send>
       <app_msg_receive>0</app_msg_receive>
       <mem_usage_debug>0</mem_usage_debug>
       <network_status_debug>0</network_status_debug>
       <checkpoint_debug>0</checkpoint_debug> 
       <coproc_debug>0</coproc_debug>
       <dcf_debug>0</dcf_debug>
       <debt_debug>0</debt_debug>
       <statefile_debug>0</statefile_debug>
       <slot_debug>0</slot_debug>
    </log_flags>
  <options>
        <alt_platform>platform_name</alt_platform>
        <data_dir>/projects/setiathome.berkeley.edu/</data_dir>
        <disallow_attach>0</disallow_attach>
        <dont_contact_ref_site>
        <dont_check_file_sizes>0</dont_check_file_sizes>
        <exclusive_app>important.exe</exclusive_app>
        <exclusive_gpu_app>important.exe</exclusive_gpu_app>
        <exit_after_finish>0</exit_after_finish>
        <force_auth>basic | digest | gss-negotiate | ntlm</force_auth>
        <http_1_0>0</http_1_0>
        <max_file_xfers>8</max_file_xfers>
        <max_file_xfers_per_project>2</max_file_xfers_per_project>
        <max_stderr_file_size>2097152</max_stderr_file_size>
        <max_stdout_file_size>2097152</max_stdout_file_size>
        <ncpus>-1</ncpus>
        <no_alt_platform>1</no_alt_platform>
        <no_gpus>0</no_gpus>
        <no_priority_change>0</no_priority_change>
        <os_random_only>0</os_random_only>
        <report_results_immediately>0</report_results_immediately>
        <run_apps_manually>0</run_apps_manually>
        <save_stats_days>30</save_stats_days>
        <simple_gui_only>1</simple_gui_only>
        <skip_cpu_benchmarks>0</skip_cpu_benchmarks>
        <start_apps_manually>0</start_apps_manually>
        <start_delay>0</start_delay>
        <suppress_net_info>1</suppress_net_info>
        <unsigned_apps_ok>0</unsigned_apps_ok>
        <use_all_gpus>1</use_all_gpus>
        <use_certs>1</use_certs>
        <use_certs_only>0</use_certs_only>
  </options>
</cc_config> 

I only changed things in the "</options>" area. I'm at home, so I don't have any special networking/proxy setup situation to deal with.

I set my preferences to
    Use at most 5 GB disk space
    Leave at least 1 GB disk space free
    Use at most 80% of total disk space


BOINC still thinks my preferences limit my disk space to 0.00GB in the messages tab. I have a 500GB hard drive with 35% (165GB) free.

My 9800GTX+s have 512MB of memory, so that shouldn't be a problem, and my entire system has 8GB DDR2 800. However, if you need to see more info, please tell me what you'd like to see.

This stuff was working fine back in January- I didn't change a thing between then and now, so I'm not seeing what the problem could be. Thanks a lot for the help!

ID: 947552 · Report as offensive
Aeon2058

Send message
Joined: 24 Jan 09
Posts: 9
Credit: 1,184,681
RAC: 0
United States
Message 947560 - Posted: 16 Nov 2009, 5:09:39 UTC

Okay, did a bit more googling and tweaking, and I have both GPUs working (on separate tasks), and I finally got it to see that I was giving it more than 0.00GB for extra tasks.

I do still have an issue- is there any way to get BOINC to do tasks in order of their deadline? I've got some listed as due 11/29/09, but it's currently working on 3 that are due 1/3/2010.

Oh, and is there a structure to the credit system? Are some tasks worth more than others, and how can I tell what is worth what?
ID: 947560 · Report as offensive
Profile Gundolf Jahn

Send message
Joined: 19 Sep 00
Posts: 3184
Credit: 446,358
RAC: 0
Germany
Message 947583 - Posted: 16 Nov 2009, 8:31:33 UTC - in response to Message 947560.  
Last modified: 16 Nov 2009, 9:08:26 UTC

You should delete all but the really needed entry (use_all_gpus) from your cc_config.xml file. For instance, why would you disallow the advanced view of BOINC manager? And the data_dir and alt_platform options are definitely wrong.

Okay, did a bit more googling and tweaking, and I have both GPUs working (on separate tasks), and I finally got it to see that I was giving it more than 0.00GB for extra tasks.

Glad that you worked it out :-)

I do still have an issue- is there any way to get BOINC to do tasks in order of their deadline? I've got some listed as due 11/29/09, but it's currently working on 3 that are due 1/3/2010.

There is no need for that! BOINC takes care of that on its own. Tasks are processed in First In First Out order, except if BOINC determines in a simulation that a task is in deadline trouble. Then that task gets priority over all others until the problem is solved.

Oh, and is there a structure to the credit system? Are some tasks worth more than others, and how can I tell what is worth what?

The amount of credits "paid" for each task is (in theory) proportional to the number of FLoating-point OPerations needed to complete it. So, longer tasks (like AstroPulse) get more credits than shorter ones (MultiBeam). You could divide granted credits by needed time to get an estimate. Those numbers you can see online under your account in the section "Computing and credit" (view tasks).

Gruß,
Gundolf
[edit]Did you see the chapter Computation credit in the BOINK Wiki?[/edit]
Computer sind nicht alles im Leben. (Kleiner Scherz)

SETI@home classic workunits 3,758
SETI@home classic CPU time 66,520 hours
ID: 947583 · Report as offensive
Aeon2058

Send message
Joined: 24 Jan 09
Posts: 9
Credit: 1,184,681
RAC: 0
United States
Message 947689 - Posted: 16 Nov 2009, 21:15:40 UTC

Thanks a lot for your help.

I readjusted my cc_config file; the data_dir is fixed, and I got rid of a large number of the extraneous flags. Everything's working great!

2 final questions:

1. do you need a special add-on/plug-in to get Astropulse tasks? All 4 of my CPUs have been working only on Multibeam tasks.

2. BOINC can't run one task using all 4 CPU cores at once? It can only run one task per core?

Thanks again for the help.
ID: 947689 · Report as offensive
Profile Gundolf Jahn

Send message
Joined: 19 Sep 00
Posts: 3184
Credit: 446,358
RAC: 0
Germany
Message 947719 - Posted: 16 Nov 2009, 22:29:28 UTC - in response to Message 947689.  

1. do you need a special add-on/plug-in to get Astropulse tasks? All 4 of my CPUs have been working only on Multibeam tasks.

If you don't use optimised applications (app_info.xml), the only thing you have to check is "If no work for selected applications is available, accept work from other applications?" in your SETI@home preferences. However, AstroPulse tasks are few and far between, so you have to be quite lucky to get one.

2. BOINC can't run one task using all 4 CPU cores at once? It can only run one task per core?

The newest BOINC versions can run one task on multiple cores, but as far as I know, only the AQUA project has applications for that feature. SETI only has tasks for one core each.

Gruß,
Gundolf
ID: 947719 · Report as offensive
Profile MusicGod
Avatar

Send message
Joined: 7 Dec 02
Posts: 97
Credit: 24,782,870
RAC: 0
United States
Message 953301 - Posted: 9 Dec 2009, 3:40:33 UTC
Last modified: 9 Dec 2009, 3:41:44 UTC

My cuda has stopped computing and I don`t know why. It has been chgging away clipping about 3,000 credits a day and there have always been 8 cores working. now none of the cores are computing and there is nothing that says it is waiting to run.
I am using Windows vista 64. Intel I7 920 with12 gigs of memory . any ideas?
My cache is full with waiting projects all seti
ID: 953301 · Report as offensive
Profile Gundolf Jahn

Send message
Joined: 19 Sep 00
Posts: 3184
Credit: 446,358
RAC: 0
Germany
Message 953338 - Posted: 9 Dec 2009, 8:27:32 UTC - in response to Message 953301.  

What does the Status column in the Projects tab tell about the SETI project?

Did I get it right that there are SETI tasks "Ready to start" but none "Running" or "Waiting to run" in Tasks tab?

Gruß,
Gundolf
Computer sind nicht alles im Leben. (Kleiner Scherz)

SETI@home classic workunits 3,758
SETI@home classic CPU time 66,520 hours
ID: 953338 · Report as offensive
Profile perryjay
Volunteer tester
Avatar

Send message
Joined: 20 Aug 02
Posts: 3377
Credit: 20,676,751
RAC: 0
United States
Message 953391 - Posted: 9 Dec 2009, 16:44:26 UTC - in response to Message 953301.  

It took a little digging but your Cuda does seem to be working. It is turning in completed work that you received on the 5th. I have no idea why it is so far behind. It is turning in good times. It must have got a herd of work before then that it is taking time to get through. This is the last one I see completed. It is a VLAR which could explain the time it is taking to complete if you hit a bunch of them... 1439659170 538705373 5 Dec 2009 12:22:06 UTC 9 Dec 2009 13:52:17 UTC Completed and validated 7,514.97 241.79 172.69 104.75 SETI@home Enhanced v6.08 (cuda)

This was only about 3 hours ago it reported.

Semper Fi, happy crunching.


PROUD MEMBER OF Team Starfire World BOINC
ID: 953391 · Report as offensive

Questions and Answers : GPU applications : BOINC won't download more projects, and won't use 2nd GPU


 
©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.