Questions and Answers :
Unix/Linux :
Cuda70 tasks crash due to missing libraries [SOLVED]
Message board moderation
| Author | Message |
|---|---|
|
Send message Joined: 31 Oct 15 Posts: 2 Credit: 64,912 RAC: 0 |
Hello. I had a problem and I want to share how I solved it. The problem On my system (Linux mangan 4.2.0-1-amd64 #1 SMP Debian 4.2.6-1 (2015-11-10) x86_64 GNU/Linux) (nVidia GeForce 840M Driver Version: 340.96) The cuda libraries for cuda70 are not available (65 and 60 are). Boinc kept sending me cuda70 WUs, which crashed after 3 seconds. I have got like 90 cuda70 wus and only 5 cuda60 ones. I think boinc figured out that cuda70 tasks are fastest on my machine. The solution Naviate to boinc directory /var/lib/boinc-client/projects/moowrap.net/ and SwitchUser to boinc user if you run boinc-client as boinc. Now MoVe/rename dnetc520-linux-amd64-cuda70 binary away and LiNk dnetc520-linux-amd64-cuda60 as dnetc520-linux-amd64-cuda70. Here are the commands I used: sudo -u boinc bash (the forum removes newlines from code blocks, srsly?) cd /var/lib/boinc-client/projects/moowrap.net/ mv dnetc520-linux-amd64-cuda70 dnetc520-linux-amd64-cuda70-orig ln dnetc520-linux-amd64-cuda60 dnetc520-linux-amd64-cuda70 My first taskhttp://moowrap.net/result.php?resultid=41297620 is going to finish and I will post if it gets validated or not. The real solution will of course be checkboxes in the project preference, like Einstein or PrimeGrid has already.[/code] |
|
Send message Joined: 31 Oct 15 Posts: 2 Credit: 64,912 RAC: 0 |
NOPE, that hack did not work. Boinc decided to overwrite my link due to checksum error. So I finally read the Anonumous version articla about app_info.xml and it seems to work. Boinc refuse to download files specified in app_info, so I have to download and chmod them manually. Anyway, here is the app_info.xml http://brod.holes.sk/up/moowrap_app_info.xml <app_info>
<app>
<name>dnetc</name>
<user_friendly_name>Distributed.net Client BR+</user_friendly_name>
<non_cpu_intensive>0</non_cpu_intensive>
<fraction_done_exact>0</fraction_done_exact>
</app>
<file_info>
<name>dnetc_wrapper_1.4_x86_64-pc-linux-gnu__cuda</name>
<executable/>
</file_info>
<file_info>
<name>dnetc520-linux-amd64-cuda60</name>
<executable/>
</file_info>
<file_info>
<name>dnetc-gpu-1.3.ini</name>
</file_info>
<file_info>
<name>job-lin64-cuda60-520.1.xml</name>
</file_info>
<app_version>
<app_name>dnetc</app_name>
<version_num>104</version_num>
<api_version>7.5.0</api_version>
<file_ref>
<file_name>dnetc_wrapper_1.4_x86_64-pc-linux-gnu__cuda</file_name>
<main_program/>
</file_ref>
<file_ref>
<file_name>dnetc520-linux-amd64-cuda60</file_name>
<copy_file/>
</file_ref>
<file_ref>
<file_name>dnetc-gpu-1.3.ini</file_name>
<open_name>dnetc.ini</open_name>
<copy_file/>
</file_ref>
<file_ref>
<file_name>job-lin64-cuda60-520.1.xml</file_name>
<open_name>job.xml</open_name>
<copy_file/>
</file_ref>
<is_wrapper/>
<plan_class>cuda60</plan_class>
<avg_ncpus>0.200000</avg_ncpus>
<max_ncpus>0.200000</max_ncpus>
<flops>114433597115.420319</flops>
<coproc>
<type>NVIDIA</type>
<count>1.000000</count>
</coproc>
<gpu_ram>33554432.000000</gpu_ram>
</app_version>
</app_info>
You need the following files in the project directory, do not forget to chmod +x executables (bold).
[/url] |