Welcome to MilkyWay@home

Linux x64 CUDA app - HOWTO?..

Message boards : Number crunching : Linux x64 CUDA app - HOWTO?..
Message board moderation

To post messages, you must log in.

AuthorMessage
Rabinovitch

Send message
Joined: 3 Nov 09
Posts: 9
Credit: 74,895
RAC: 0
Message 34095 - Posted: 2 Dec 2009, 7:53:11 UTC

Hi all!

Just attached new host to the project (Debian Squeeze), it loading all the project's files but cannot process WUs:



<core_client_version>6.10.17</core_client_version>
<![CDATA[
<message>
process exited with code 127 (0x7f, -129)
</message>
<stderr_txt>
../../projects/milkyway.cs.rpi.edu_milkyway/milkyway_0.21_x86_64-pc-linux-gnu__cuda23: error while loading shared libraries: libcudart.so.2: cannot open shared object file: No such file or directory

</stderr_txt>
]]>

I can't find a solvation here, in the Message boards, and instaed of using a shaman's technics I want to ask gurus how to make it work. If I need to make any symbolic links via ln then please explain me what exactly should I do.
Previously on Xubuntu 9.10 x64 it started to work without any actions from my side.

Thanks in advance!

p.s. collatz is working on this host right after extracting an archive with application and corresponding files to the project's directory.
ID: 34095 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Rabinovitch

Send message
Joined: 3 Nov 09
Posts: 9
Credit: 74,895
RAC: 0
Message 34156 - Posted: 3 Dec 2009, 20:51:39 UTC

Just attached to the Einstein, and it's cuda-task started to work without my participation... :-)

So please anyone help me to crunh Milky!
ID: 34156 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile banditwolf
Avatar

Send message
Joined: 12 Nov 07
Posts: 2425
Credit: 524,164
RAC: 0
Message 34157 - Posted: 3 Dec 2009, 20:55:50 UTC

Did you read the read-me file with the app? It should explain anything you need to do to the app.
Doesn't expecting the unexpected make the unexpected the expected?
If it makes sense, DON'T do it.
ID: 34157 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Rabinovitch

Send message
Joined: 3 Nov 09
Posts: 9
Credit: 74,895
RAC: 0
Message 34158 - Posted: 3 Dec 2009, 21:12:01 UTC
Last modified: 3 Dec 2009, 21:12:16 UTC

Hm. Seeing into the /var/lib/boinc-client/ directory I can't find any readme.

Remind that all project's files was downloaded automatically by BOINC Manager.
ID: 34158 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile banditwolf
Avatar

Send message
Joined: 12 Nov 07
Posts: 2425
Credit: 524,164
RAC: 0
Message 34159 - Posted: 3 Dec 2009, 21:16:00 UTC - in response to Message 34158.  
Last modified: 3 Dec 2009, 21:17:32 UTC

Hm. Seeing into the /var/lib/boinc-client/ directory I can't find any readme.

Remind that all project's files was downloaded automatically by BOINC Manager.

Oh, then you wouldn't have a readme in that nor will it run cuda. It is only a basic stock app. You need to download an opti app and install it yourself. Those will have a readme with them.

*edit* looking for the site with cuda apps.
ID: 34159 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Rabinovitch

Send message
Joined: 3 Nov 09
Posts: 9
Credit: 74,895
RAC: 0
Message 34160 - Posted: 3 Dec 2009, 21:19:37 UTC - in response to Message 34159.  

Thanx, I'll try.

It's strange that the stock application was working on Xubuntu a couple weeks ago...
ID: 34160 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile banditwolf
Avatar

Send message
Joined: 12 Nov 07
Posts: 2425
Credit: 524,164
RAC: 0
Message 34161 - Posted: 3 Dec 2009, 21:22:11 UTC - in response to Message 34160.  
Last modified: 3 Dec 2009, 21:37:00 UTC

It won't run your cuda gpu though. (or atleast I don't think the stock linux runs gpu either) This thread might help; http://milkyway.cs.rpi.edu/milkyway/forum_thread.php?id=584#9123.

I can't find the cuda opti app only the ati app for linux. I know I am missing one site.
ID: 34161 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Rabinovitch

Send message
Joined: 3 Nov 09
Posts: 9
Credit: 74,895
RAC: 0
Message 34162 - Posted: 3 Dec 2009, 22:12:08 UTC

Useless thread, unfortunately.

I hope that someone of project developers will help...

Resetting project was also useless.
ID: 34162 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile ritterm
Avatar

Send message
Joined: 16 Jun 08
Posts: 93
Credit: 366,882,323
RAC: 0
Message 34175 - Posted: 4 Dec 2009, 3:45:37 UTC
Last modified: 4 Dec 2009, 3:46:49 UTC

I'm running Ubuntu 9.04 on a C2Q host with a GTX265 (driver 190.18) and get the following error when I try to download a WU:

Thu 03 Dec 2009 10:24:05 PM EST Milkyway@home Message from server: Can't use NVIDIA GPU app for MilkyWay@Home: NVIDIA driver version 19038 or later needed

I'm assuming that the stock app is the only way to go since there doesn't seem to be an optimized app for CUDA at Arkayn's site. I don't see at the NVIDIA website where 190.38 is available for any Linux OS. What am I missing?
ID: 34175 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile ritterm
Avatar

Send message
Joined: 16 Jun 08
Posts: 93
Credit: 366,882,323
RAC: 0
Message 34198 - Posted: 4 Dec 2009, 14:04:17 UTC - in response to Message 34175.  
Last modified: 4 Dec 2009, 14:05:02 UTC

I don't see at the NVIDIA website where 190.38 is available for any Linux OS

Well, I only looked in the "CUDA Zone" at NVIDIA. Had I looked harder, I might have found these:

http://www.nvidia.com/object/linux_display_amd64_190.42.html
http://www.nvidia.com/object/linux_display_ia32_190.42.html

Duh... <blushing with embarassment>
ID: 34198 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Anthony Waters

Send message
Joined: 16 Jun 09
Posts: 85
Credit: 172,476
RAC: 0
Message 34317 - Posted: 6 Dec 2009, 19:16:44 UTC
Last modified: 6 Dec 2009, 19:17:12 UTC

Please try the new version, 0.23, I included libcudart.so.2
ID: 34317 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Rabinovitch

Send message
Joined: 3 Nov 09
Posts: 9
Credit: 74,895
RAC: 0
Message 34359 - Posted: 7 Dec 2009, 14:46:08 UTC
Last modified: 7 Dec 2009, 14:47:15 UTC

Well, I've installed Ubuntu 9.10 x64, and got now 12 tasks with 0.24 app - working well till now, I'll watch how it crunch...
ID: 34359 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Rabinovitch

Send message
Joined: 3 Nov 09
Posts: 9
Credit: 74,895
RAC: 0
Message 34361 - Posted: 7 Dec 2009, 15:50:46 UTC
Last modified: 7 Dec 2009, 15:58:23 UTC

The only thing I don't like is that BOINC Manager can't see an usable GPUs under my account - it need to be started via sudo /etc/init.d/boinc-client start to make it find a GPU...

Earlier in Xubuntu there was a way to fix it - add boinc user to video group (usermod -a -G video boinc), but somewhy now it don't work.....

Trying now to folow this advice from Gpugrid forums. Will see what happen.
ID: 34361 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Rabinovitch

Send message
Joined: 3 Nov 09
Posts: 9
Credit: 74,895
RAC: 0
Message 34410 - Posted: 9 Dec 2009, 11:12:55 UTC

Well, everything is OK now - GPU is being recognized on system startup. There were applied two fixes (in my previous post and posted here. I don't know which of them is the main fix... :-) But it works.
ID: 34410 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile [AF>HFR>RR] Black Hole S...
Avatar

Send message
Joined: 2 Apr 08
Posts: 10
Credit: 8,126,465
RAC: 0
Message 34610 - Posted: 17 Dec 2009, 10:54:33 UTC
Last modified: 17 Dec 2009, 11:01:10 UTC

Hi

Having problems with my GTX275 using 190.42 nvidia drivers on Boinc 6.10.17 for Ubuntu 9.10 x64.

All wus exit like this:
<core_client_version>6.10.17</core_client_version>
<![CDATA[
<message>
process exited with code 1 (0x1, -255)
</message>
<stderr_txt>
Device index specified on the command line was 0
Looking for a Double Precision capable NVIDIA GPU
The device GeForce GTX 275 specified on the command line can be used
Used 288540/916800 memory, 628259 remaining
Error executing gpu__zero_integrals error message: unspecified launch failure
Error executing gpu__zero_integrals error message: unspecified launch failure
Error executing gpu__zero_integrals error message: unspecified launch failure
Error executing gpu__zero_integrals error message: unspecified launch failure
Error executing gpu__zero_integrals error message: unspecified launch failure
Error executing gpu__zero_integrals error message: unspecified launch failure
Error executing gpu__zero_integrals error message: unspecified launch failure
Error executing gpu__zero_integrals error message: unspecified launch failure
Error executing gpu__zero_integrals error message: unspecified launch failure
Error executing gpu__zero_integrals error message: unspecified launch failure
Error executing gpu__zero_integrals error message: unspecified launch failure
Error executing gpu__zero_integrals error message: unspecified launch failure
Error executing gpu__zero_integrals error message: unspecified launch failure
Error executing gpu__zero_integrals error message: unspecified launch failure
Error executing gpu__zero_integrals error message: unspecified launch failure
Error executing gpu__zero_integrals error message: unspecified launch failure
Error executing gpu__zero_integrals error message: unspecified launch failure
Error executing gpu__zero_integrals error message: unspecified launch failure
Error executing gpu__zero_integrals error message: unspecified launch failure
Error executing gpu__zero_integrals error message: unspecified launch failure
Error executing gpu__zero_integrals error message: unspecified launch failure
Error executing gpu__zero_integrals error message: unspecified launch failure
Error executing gpu__zero_integrals error message: unspecified launch failure
Error executing gpu__zero_integrals error message: unspecified launch failure
Error executing gpu__zero_integrals error message: unspecified launch failure
Error executing gpu__zero_integrals error message: unspecified launch failure
Error executing gpu__zero_integrals error message: unspecified launch failure
Error executing gpu__zero_integrals error message: unspecified launch failure
Error executing gpu__zero_integrals error message: unspecified launch failure
Error executing gpu__zero_integrals error message: unspecified launch failure
Error executing gpu__zero_integrals error message: unspecified launch failure
Error executing gpu__zero_integrals error message: unspecified launch failure
Error executing gpu__zero_integrals error message: unspecified launch failure
Error executing gpu__zero_integrals error message: unspecified launch failure
Error executing gpu__zero_integrals error message: unspecified launch failure
Error executing gpu__zero_integrals error message: unspecified launch failure
Error executing gpu__zero_integrals error message: unspecified launch failure
Error executing gpu__zero_integrals error message: unspecified launch failure
Error executing gpu__integral_kernel3 error message: unspecified launch failure

</stderr_txt>
]]>


application version 0.24


Here's the log at Boinc Manager start:
jeu. 17 déc. 2009 11:44:09 CET Starting BOINC client version 6.10.17 for x86_64-pc-linux-gnu
jeu. 17 déc. 2009 11:44:09 CET Config: run apps at regular priority
jeu. 17 déc. 2009 11:44:09 CET Config: report completed tasks immediately
jeu. 17 déc. 2009 11:44:09 CET log flags: file_xfer, sched_ops, task
jeu. 17 déc. 2009 11:44:09 CET Libraries: libcurl/7.19.5 OpenSSL/0.9.8g zlib/1.2.3.3 libidn/1.15
jeu. 17 déc. 2009 11:44:09 CET Data directory: /var/lib/boinc-client
jeu. 17 déc. 2009 11:44:09 CET Processor: 4 GenuineIntel Intel(R) Core(TM)2 Quad CPU Q6600 @ 2.40GHz [Family 6 Model 15 Stepping 11]
jeu. 17 déc. 2009 11:44:09 CET Processor: 4.00 MB cache
jeu. 17 déc. 2009 11:44:09 CET Processor features: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe syscall nx lm constant_tsc arch_perfmon pebs bts rep_good pni dtes64 monitor ds_cpl vmx est tm2 ssse3 cx16 xtpr pdc
jeu. 17 déc. 2009 11:44:09 CET OS: Linux: 2.6.31-14-generic
jeu. 17 déc. 2009 11:44:09 CET Memory: 1.96 GB physical, 9.38 GB virtual
jeu. 17 déc. 2009 11:44:09 CET Disk: 9.17 GB total, 2.38 GB free
jeu. 17 déc. 2009 11:44:09 CET Local time is UTC +1 hours
jeu. 17 déc. 2009 11:44:09 CET NVIDIA GPU 0: GeForce GTX 275 (driver version unknown, CUDA version 2030, compute capability 1.3, 895MB, 674 GFLOPS peak)



Help please

Thanks
ID: 34610 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Anthony Waters

Send message
Joined: 16 Jun 09
Posts: 85
Credit: 172,476
RAC: 0
Message 34650 - Posted: 18 Dec 2009, 1:15:05 UTC - in response to Message 34610.  

Do other CUDA applications experience the same behavior?

Is the boinc user in the video group?
ID: 34650 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile [AF>HFR>RR] Black Hole S...
Avatar

Send message
Joined: 2 Apr 08
Posts: 10
Credit: 8,126,465
RAC: 0
Message 34662 - Posted: 18 Dec 2009, 13:35:53 UTC - in response to Message 34650.  
Last modified: 18 Dec 2009, 13:36:16 UTC

Thanks for your reply.

Do other CUDA applications experience the same behavior?


No. I can play TrackMania Nations with 3D acceleration through wine without any problem. Collatz used to work fine on my host but lately it made my xorg dramatically freeze so I came to Milkyway to test it.

Is the boinc user in the video group?

No it wasn't.
I added myself to the video group => same problem
I added the root account to the video group => same problem

These changes were done and the Boinc daemon was then stopped and restarted.


Here are the messages in BoincManager:
ven. 18 déc. 2009 14:26:27 CET Milkyway@home Starting de_s222_3s_best_1p_03r_43_5474221_1261046141_0
ven. 18 déc. 2009 14:26:27 CET Milkyway@home Starting task de_s222_3s_best_1p_03r_43_5474221_1261046141_0 using milkyway version 24
ven. 18 déc. 2009 14:26:29 CET Milkyway@home Computation for task de_s222_3s_best_1p_03r_43_5474221_1261046141_0 finished
ven. 18 déc. 2009 14:26:29 CET Milkyway@home Output file de_s222_3s_best_1p_03r_43_5474221_1261046141_0_0 for task de_s222_3s_best_1p_03r_43_5474221_1261046141_0 absent
ven. 18 déc. 2009 14:26:29 CET Milkyway@home Starting de_s222_3s_best_1p_03r_43_5474220_1261046141_0
ven. 18 déc. 2009 14:26:29 CET Milkyway@home Starting task de_s222_3s_best_1p_03r_43_5474220_1261046141_0 using milkyway version 24
ven. 18 déc. 2009 14:26:32 CET Milkyway@home Computation for task de_s222_3s_best_1p_03r_43_5474220_1261046141_0 finished
ven. 18 déc. 2009 14:26:32 CET Milkyway@home Output file de_s222_3s_best_1p_03r_43_5474220_1261046141_0_0 for task de_s222_3s_best_1p_03r_43_5474220_1261046141_0 absent


This is my cc_config in case something would be wrong:
<cc_config>
<options>
<report_results_immediately>1</report_results_immediately>
<no_priority_change>1</no_priority_change>
</options>
<log_flags>
<task>1</task>
<file_xfer>1</file_xfer>
<sched_ops>1</sched_ops>
<cpu_sched>0</cpu_sched>
<cpu_sched_debug>0</cpu_sched_debug>
<rr_simulation>0</rr_simulation>
<debt_debug>0</debt_debug>
<task_debug>0</task_debug>
<work_fetch_debug>0</work_fetch_debug>
<unparsed_xml>0</unparsed_xml>
<state_debug>0</state_debug>
<file_xfer_debug>0</file_xfer_debug>
<sched_op_debug>0</sched_op_debug>
<http_debug>0</http_debug>
<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>
<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>
</log_flags>
</cc_config>
ID: 34662 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Anthony Waters

Send message
Joined: 16 Jun 09
Posts: 85
Credit: 172,476
RAC: 0
Message 34683 - Posted: 19 Dec 2009, 4:38:11 UTC

Is it acceptable to assume that it worked prior to v0.24?
ID: 34683 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile [AF>HFR>RR] Black Hole S...
Avatar

Send message
Joined: 2 Apr 08
Posts: 10
Credit: 8,126,465
RAC: 0
Message 34689 - Posted: 19 Dec 2009, 11:27:04 UTC

Well, no idea since I missed Milkyway's first CUDA apps: I only tried with the current 0.24 version.
I believe my system is OK since Collatz used to work without any problem until I began to get freezes with their new app, and this summer, I ran Aqua's CUDA app without any problem either.
ID: 34689 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote

Message boards : Number crunching : Linux x64 CUDA app - HOWTO?..

©2024 Astroinformatics Group