Linux boinc version 7.6.31 can't run optimized applications

Message boards : Number crunching : Linux boinc version 7.6.31 can't run optimized applications
Message board moderation

To post messages, you must log in.

1 · 2 · 3 · Next

AuthorMessage
Zytra

Send message
Joined: 29 Aug 16
Posts: 36
Credit: 58,532,935
RAC: 0
United States
Message 1815814 - Posted: 8 Sep 2016, 23:37:19 UTC

Hi
I've installed the boinc client/manager on a Ubuntu 16.04 machine (i7-5930K) and it worked fine for a day. Like my other windows based systems I wanted to install optimized app.

I tried both of the following (AVX CPU):

1. 08-Linux 32bit AstroPulse_v7 v7.05 AVX (r2728), October 2014 from http://lunatics.kwsn.info/index.php?action=downloads;cat=34

2. 06 - Linux 64bit Multibeam v8 for AVX CPUs (r3345), January 2016 from http://lunatics.kwsn.info/index.php?action=downloads;cat=48

I followed the instructions, turned off Boinc, copied the files. Restarted Boinc and there, nothing. After a few seconds I would get a notice that says:

Your app_info.xml file doesn't have a usable version of SETI@Home v8


Looking online I found that maybe the app didn't have the right permissions, so I chmod the file and still the same thing. Though I am not sure if I saw the notice about the xml file each time, one thing for sure is no download or work was happening.

Like I said I tried both of these optimized app. I also tried reinstalling Boinc.

Now, I removed the files from the setiathome.berkely.edu folder and Boinc will not download any work.

So now I am wondering if there is actual work available or if the server is down or something. I figured once I removed those 2 files boinc should run like it was before attempting running those optimized app. Especially after reinstalling Boinc.

Anyway, if someone could confirm if those 2 optimized app are the correct ones for my CPU? I've got a similar machine running on W10 and it's crunching through an AVX optimized app just fine.

Thanks
ID: 1815814 · Report as offensive
Zytra

Send message
Joined: 29 Aug 16
Posts: 36
Credit: 58,532,935
RAC: 0
United States
Message 1815818 - Posted: 9 Sep 2016, 0:00:24 UTC - in response to Message 1815814.  

letting it hang a while made it start downloading work and processing them (without the optimized app).

so I redid it, copied the 2 files, chmod 777 the app to be safe, no notice... will let it hang a while and see what happens.
ID: 1815818 · Report as offensive
Urs Echternacht
Volunteer tester
Avatar

Send message
Joined: 15 May 99
Posts: 692
Credit: 135,197,781
RAC: 211
Germany
Message 1815825 - Posted: 9 Sep 2016, 1:16:03 UTC - in response to Message 1815818.  
Last modified: 9 Sep 2016, 1:17:32 UTC

Hi,

how does your app_info.xml file look like, maybe you have a typo somewhere.

hint: the downloaded lunatics packages contain examples of some app_info.xml files, if you used one of these make sure to doublecheck names, version numbers and plan classes if you change anything.

The MultiBeam v8 app is ok, but for AstroPulse you should also use a 64bit app.
_\|/_
U r s
ID: 1815825 · Report as offensive
Zytra

Send message
Joined: 29 Aug 16
Posts: 36
Credit: 58,532,935
RAC: 0
United States
Message 1815829 - Posted: 9 Sep 2016, 1:36:46 UTC

Thanks
My AP was 64 bit also I just didnt copy paste the right one to my post (used a different computer).
I didn't modify the xml file at all and just followed the instructions included in the archive.

I will check what they look like.


Also the lunatics are 7z files which are not natively compatible with ubuntu. I installed a package for it and thought maybe it didn't conserve the files permissions that's why I chmod'ed the app to be sure. I didnt touch permissions on the xml though.

Thanks
ID: 1815829 · Report as offensive
Zytra

Send message
Joined: 29 Aug 16
Posts: 36
Credit: 58,532,935
RAC: 0
United States
Message 1815849 - Posted: 9 Sep 2016, 4:51:23 UTC
Last modified: 9 Sep 2016, 4:58:40 UTC

here's the content of the app_info.xml for the 05-Linux 64bit AstroPulse_v7 7.05 AVX (r2728), October 2014:

<app_info>
    <app>
        <name>astropulse_v7</name>
    </app>
    <file_info>
        <name>ap_7.05r2728_avx_linux64</name>
        <executable/>
    </file_info>
    <app_version>
        <app_name>astropulse_v7</app_name>
        <version_num>705</version_num>
        <platform>x86_64-pc-linux-gnu</platform>
        <plan_class></plan_class>
        <cmdline></cmdline>
        <file_ref>
            <file_name>ap_7.05r2728_avx_linux64</file_name>
            <main_program/>
        </file_ref>
    </app_version>
</app_info>



and the app_info.xml for 06 - Linux 64bit Multibeam v8 for AVX CPUs (r3345), January 2016

<app_info>
    <app>
      <name>setiathome_v8</name>
    </app>
    <file_info>
      <name>MBv8_8.05r3345_avx_linux64</name>
      <executable/>
    </file_info>
    <app_version>
      <app_name>setiathome_v8</app_name>
      <version_num>804</version_num>
      <platform>x86_64-pc-linux-gnu</platform>
      <cmdline></cmdline>
      <file_ref>
        <file_name>MBv8_8.05r3345_avx_linux64</file_name>
        <main_program/>
      </file_ref>
    </app_version>
    <app_version>
      <app_name>setiathome_v8</app_name>
      <version_num>805</version_num>
      <platform>x86_64-pc-linux-gnu</platform>
      <cmdline></cmdline>
      <file_ref>
        <file_name>MBv8_8.05r3345_avx_linux64</file_name>
        <main_program/>
      </file_ref>
    </app_version>
</app_info>



does this look all right?

not sure why the multibeam code has two sets of <app_version> tags. hmm maybe that was the problem huh
ID: 1815849 · Report as offensive
Profile BilBg
Volunteer tester
Avatar

Send message
Joined: 27 May 07
Posts: 3720
Credit: 9,385,827
RAC: 0
Bulgaria
Message 1815917 - Posted: 9 Sep 2016, 13:47:30 UTC - in response to Message 1815849.  
Last modified: 9 Sep 2016, 14:09:44 UTC

not sure why the multibeam code has two sets of <app_version> tags.

It is OK, on Windows we use maybe ~10 "sets" (for the GPU apps)

They differ only by <version_num>804</version_num>
To preserve already Downloaded tasks marked 8.04 and 8.05

But the numbers seem wrong for that (to preserve), look:
http://setiathome.berkeley.edu/apps.php

Maybe 804 was for SETI@home Beta site, you may change it to <version_num>800</version_num>
("Linux/x86_64 8.00")


2) The 2 app_info.xml files look OK but you can't have 2 app_info.xml
You should merge them (e.g. if you put only the first app_info.xml for astropulse_v7 - NO setiathome_v8 work will be Downloaded)

Just "cat" the 2 files (I don't use Linux but think "cat" = merge)
Then in the resulting text delete in the middle the 2 lines:
</app_info>
<app_info>


P.S.
The errors are "execv: Permission denied" so be sure to chmod +x the executables:
http://setiathome.berkeley.edu/results.php?hostid=8089610&offset=0&show_names=0&state=6&appid=
http://setiathome.berkeley.edu/result.php?resultid=5145347844
 


- ALF - "Find out what you don't do well ..... then don't do it!" :)
 
ID: 1815917 · Report as offensive
Urs Echternacht
Volunteer tester
Avatar

Send message
Joined: 15 May 99
Posts: 692
Credit: 135,197,781
RAC: 211
Germany
Message 1816063 - Posted: 10 Sep 2016, 2:13:45 UTC - in response to Message 1815849.  
Last modified: 10 Sep 2016, 2:14:05 UTC

BilBg is right, only one app_info.xml per project is expected by BOINC.
So merging the two quoted files could look like this :

<app_info>
    <app>
        <name>astropulse_v7</name>
    </app>
    <file_info>
        <name>ap_7.05r2728_avx_linux64</name>
        <executable/>
    </file_info>
    <app_version>
        <app_name>astropulse_v7</app_name>
        <version_num>705</version_num>
        <platform>x86_64-pc-linux-gnu</platform>
        <plan_class></plan_class>
        <cmdline></cmdline>
        <file_ref>
            <file_name>ap_7.05r2728_avx_linux64</file_name>
            <main_program/>
        </file_ref>
    </app_version>
    <app>
      <name>setiathome_v8</name>
    </app>
    <file_info>
      <name>MBv8_8.05r3345_avx_linux64</name>
      <executable/>
    </file_info>
    <app_version>
      <app_name>setiathome_v8</app_name>
      <version_num>800</version_num>
      <platform>x86_64-pc-linux-gnu</platform>
      <cmdline></cmdline>
      <file_ref>
        <file_name>MBv8_8.05r3345_avx_linux64</file_name>
        <main_program/>
      </file_ref>
    </app_version>
    <app_version>
      <app_name>setiathome_v8</app_name>
      <version_num>805</version_num>
      <platform>x86_64-pc-linux-gnu</platform>
      <cmdline></cmdline>
      <file_ref>
        <file_name>MBv8_8.05r3345_avx_linux64</file_name>
        <main_program/>
      </file_ref>
    </app_version>
</app_info>


_\|/_
U r s
ID: 1816063 · Report as offensive
Richard Haselgrove Project Donor
Volunteer tester

Send message
Joined: 4 Jul 99
Posts: 14649
Credit: 200,643,578
RAC: 874
United Kingdom
Message 1816119 - Posted: 10 Sep 2016, 8:04:21 UTC

May I make two suggestions to developers who package sample app_info elements ('AIstub files') with their test builds?

1) Make it clear via the file names whether the AIstub has been prepared to preserve cached Main project work, or Beta project work.

2) When preserving multiple versions, sequence them in decreasing order of version_num. Replacement work after the cached task supply has been exhausted is always given the version number and plan class of the first <app_version> encountered in app_info (for the required application and device, of course). You'll find it quicker and easier to retire old, deprecated, versions if you don't keep re-using the numbers for new work.
ID: 1816119 · Report as offensive
Zytra

Send message
Joined: 29 Aug 16
Posts: 36
Credit: 58,532,935
RAC: 0
United States
Message 1816602 - Posted: 12 Sep 2016, 6:09:48 UTC

I was able to get a linux machine to use optimized app for the CPU.
while I was able to get boinc to recognize the GPU (it appears on the seti@home website computer page), it won't download and start crunching on the GPU.
nothing in the logs seems to be pointing at a problem.... just nothing.

here's my app_info.xml file:

the gpu on this machine is a R9 290X, and I'm running ubuntu 16.04 with the latest AMDGPU drivers from the AMD website.

maybe there is no AP work available right now?
I'll let it run overnight and check in the morning.



<app_info>
  <app>
    <name>astropulse_v7</name>
  </app>
  <file_info>
    <name>ap_7.05r2728_avx_linux64</name>
    <executable/>
  </file_info>
  <app_version>
    <app_name>astropulse_v7</app_name>
    <version_num>705</version_num>
    <platform>x86_64-pc-linux-gnu</platform>
    <file_ref>
      <file_name>ap_7.05r2728_avx_linux64</file_name>
      <main_program/>
    </file_ref>
  </app_version>
  <app>
    <name>astropulse_v7</name>
  </app>
  <file_info>
    <name>ap_7.08r2751_sse2_clATI_x86_64-pc-linux-gnu</name>
    <executable/>
  </file_info>
  <app_version>
    <app_name>astropulse_v7</app_name>
    <version_num>708</version_num>
    <platform>x86_64-pc-linux-gnu</platform>
    <coproc>
      <type>ATI</type>
      <count>1</count>
    </coproc>
    <plan_class>ati_opencl_100</plan_class>
    <avg_ncpus>0.05</avg_ncpus>
    <max_ncpus>0.2</max_ncpus>
    <cmdline></cmdline>
    <file_ref>
      <file_name>ap_7.08r2751_sse2_clATI_x86_64-pc-linux-gnu</file_name>
      <main_program/>
    </file_ref>
  </app_version>
  <app>
    <name>setiathome_v8</name>
  </app>
  <file_info>
    <name>MBv8_8.05r3345_avx_linux64</name>
    <executable/>
  </file_info>
  <app_version>
    <app_name>setiathome_v8</app_name>
    <version_num>805</version_num>
    <platform>x86_64-pc-linux-gnu</platform>
    <file_ref>
      <file_name>MBv8_8.05r3345_avx_linux64</file_name>
      <main_program/>
    </file_ref>
  </app_version>
</app_info>
ID: 1816602 · Report as offensive
Darth Beaver Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Avatar

Send message
Joined: 20 Aug 99
Posts: 6728
Credit: 21,443,075
RAC: 3
Australia
Message 1816603 - Posted: 12 Sep 2016, 6:18:17 UTC - in response to Message 1816602.  

Have you told the servers what pref to use .

Have you assigned that computer to Home , Work or School preferences

Check to see if the preferences are set to give you GPU units
ID: 1816603 · Report as offensive
Zytra

Send message
Joined: 29 Aug 16
Posts: 36
Credit: 58,532,935
RAC: 0
United States
Message 1816607 - Posted: 12 Sep 2016, 6:40:18 UTC

no I have not. but to be fair I have not had to do that on any of my other machines which get GPU work just fine (though, they are windows based machines).

From the log I can see "GPU computing is stopped because computer is in use".

I thought that was it, so instead of waiting 3 minutes I just went into the preferences and unchecked the box so that GPU computing is enabled all the time.

Then from the log I could see "Resuming GPU computing". but it didn't download any AP work (I don't have any MB app for my gpu) and it certainly didn't start crunching either.


I'm happy to have the CPU working optimized now but I won't give up until I understand what's going on...

thanks!!
ID: 1816607 · Report as offensive
Darth Beaver Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Avatar

Send message
Joined: 20 Aug 99
Posts: 6728
Credit: 21,443,075
RAC: 3
Australia
Message 1816608 - Posted: 12 Sep 2016, 6:51:02 UTC - in response to Message 1816607.  

I thought that was it, so instead of waiting 3 minutes I just went into the preferences and unchecked the box so that GPU computing is enabled all the time


Yeh that setting works from your mouse so stops it crunching if you move the mouse or use the keyboard mine is set like you now have it .

That's one lot of preferences there should be another set the web preferences ( at least there is for Windows)

look at the account page and you should see Seti home preferences and look there to see if you have check boxed the use ATI GPU
ID: 1816608 · Report as offensive
TBar
Volunteer tester

Send message
Joined: 22 May 99
Posts: 5204
Credit: 840,779,836
RAC: 2,768
United States
Message 1816609 - Posted: 12 Sep 2016, 6:52:14 UTC - in response to Message 1816602.  
Last modified: 12 Sep 2016, 7:07:19 UTC

...maybe there is no AP work available right now?
I'll let it run overnight and check in the morning.

Yes, there are rarely APs available. I don't see an AMD MBv8 App in your app_info file. I also can't find a Linux AMD GPU App at Lunatics. Strange. I posted an Older version some time ago at Crunchers Anonymous. It should work without any problems other than it still has the Sanity check that some times fails with noisy GUPPIs. I'm using a Newer version that doesn't have the Sanity check, it seems to be working fine. Maybe I should post it some day ;-)

If you want to use the older version it's here, http://www.arkayn.us/forum/index.php?topic=191.msg4368#msg4368
Basically you would just add the MBv8 GPU section to your app_info file making the file have four sections. I've been threatening to build a more updated version with r3528. I'll probably do that before posting any updated Apps.

You could just use the Stock GPU App and add your modified app_info section to your existing app_info. The Stock GPU App looks much newer than the one at Crunchers Anonymous.
ID: 1816609 · Report as offensive
Darth Beaver Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Avatar

Send message
Joined: 20 Aug 99
Posts: 6728
Credit: 21,443,075
RAC: 3
Australia
Message 1816610 - Posted: 12 Sep 2016, 7:02:30 UTC

I'll leave any more help to you Tbar your the Linex man I was only trying the most obvious things he should check .:)
ID: 1816610 · Report as offensive
TBar
Volunteer tester

Send message
Joined: 22 May 99
Posts: 5204
Credit: 840,779,836
RAC: 2,768
United States
Message 1816630 - Posted: 12 Sep 2016, 11:45:44 UTC - in response to Message 1816607.  

no I have not. but to be fair I have not had to do that on any of my other machines which get GPU work just fine (though, they are windows based machines).

From the log I can see "GPU computing is stopped because computer is in use".

I thought that was it, so instead of waiting 3 minutes I just went into the preferences and unchecked the box so that GPU computing is enabled all the time.

Then from the log I could see "Resuming GPU computing". but it didn't download any AP work (I don't have any MB app for my gpu) and it certainly didn't start crunching either.

I'm happy to have the CPU working optimized now but I won't give up until I understand what's going on...

thanks!!

I just posted the AMD App I've been running for over a month instead of building anything new. You'll have to integrate the GPU section to your existing app_info.xml. It's pretty straightforward, the section begins with <app> and ends with </app_version>. You can find it here, http://www.arkayn.us/forum/index.php?topic=197.msg4497#msg4497
They are the same Apps being run on this machine, http://setiathome.berkeley.edu/results.php?hostid=6906726&offset=200
ID: 1816630 · Report as offensive
Zytra

Send message
Joined: 29 Aug 16
Posts: 36
Credit: 58,532,935
RAC: 0
United States
Message 1816682 - Posted: 12 Sep 2016, 18:26:00 UTC

Thanks guys
I'm gonna give those a shot now and let you know!

Thanks
ID: 1816682 · Report as offensive
Zytra

Send message
Joined: 29 Aug 16
Posts: 36
Credit: 58,532,935
RAC: 0
United States
Message 1816683 - Posted: 12 Sep 2016, 18:41:21 UTC

well that worked out just fine!
your app_info.xml is more complicated than the other ones with additional files and such but it worked nonetheless.
R9 is official crunching under linux, yay!

thank you very much guys
ID: 1816683 · Report as offensive
TBar
Volunteer tester

Send message
Joined: 22 May 99
Posts: 5204
Credit: 840,779,836
RAC: 2,768
United States
Message 1816848 - Posted: 13 Sep 2016, 3:15:36 UTC - in response to Message 1816683.  
Last modified: 13 Sep 2016, 3:41:34 UTC

It looks to be working well. You might want to change a few settings for your card. In the docs folder open the file ReadMe_MultiBeam_OpenCL.txt and scroll to near the bottom. There is a section named command line switches which has different settings. Some of them are already in the mb_cmdline-opencl_ati5_nocal.txt file but can be raised a little for your card. I'm not sure which of those settings in the readme are best for Linux, but you can change the existing settings without any trouble, try changing the settings to;
-sbs 256 -oclfft_tune_gr 256 -oclfft_tune_wg 256 -high_perf -period_iterations_num 16

The settings take effect when the next task is started, you can suspend a running task and force another to start. If there are any problems with the settings they will usually happen when the task is started. If you try any of the additional settings from the readme, it might be best to suspend all the other non-running tasks in case the new setting fails and then add new settings one at a time. If there is a problem with a particular setting just remove it and try the next one.
ID: 1816848 · Report as offensive
Zytra

Send message
Joined: 29 Aug 16
Posts: 36
Credit: 58,532,935
RAC: 0
United States
Message 1818583 - Posted: 21 Sep 2016, 6:51:29 UTC

Thanks Tbar, I will try those settings and see how much they help.
Which kind of leads me to the question of performance evaluation in general.

I took a long break off seti@home but back in the days prior to Boinc we had a bunch of tools and logs to evaluate how good a rig was doing.

Right now, at least with the tools in hand on this website, I feel like it is extremely difficult to get a feel of how good a machine is doing. Example: it takes several weeks for the RAC to start making sense compared to similar computers in the database which have been crunching for months or more.

Then, right now, 3 weeks in I look at any of my computers and it looks like I have about as much validated WU than I have pending. Which makes me wonder how those WU are validated. If in 3 weeks of work, out of all the WU I've computed only half has been validated then it means the whole validation system is backing up real bad.

Or maybe I just don't understand how the system works just yet?


Anyway getting sidetracked a bit here. hehe
thanks again for the precious help getting those GPU apps to work!
ID: 1818583 · Report as offensive
Grant (SSSF)
Volunteer tester

Send message
Joined: 19 Aug 99
Posts: 13720
Credit: 208,696,464
RAC: 304
Australia
Message 1818586 - Posted: 21 Sep 2016, 7:21:21 UTC - in response to Message 1818583.  

Right now, at least with the tools in hand on this website, I feel like it is extremely difficult to get a feel of how good a machine is doing. Example: it takes several weeks for the RAC to start making sense compared to similar computers in the database which have been crunching for months or more.

The best way to see how a system is doing is to compare run times of WUs- however you need to take in to account the different Angle Ranges (AR) of the different WUs, and the different types of WU (Arecibo & the Guppies from the Green Bank telescope).
As long as you compare like with like, you can see how well your system is doing with different applications & configuration settings.
However comparing Arecibo shorty run times with Guppie run times won't give you any useful information.


Then, right now, 3 weeks in I look at any of my computers and it looks like I have about as much validated WU than I have pending.

The smaller your cache, and the faster you process WUs, the greater the number of Pending WUs will be.


Which makes me wonder how those WU are validated. If in 3 weeks of work, out of all the WU I've computed only half has been validated then it means the whole validation system is backing up real bad.

Nope, it just means you're crunching a lot of work in a short period of time.
I've got a fast video card, and a few slower ones. My Pendings almost equal my work In Progress.
For those with monster crunchers, their Pendings can be from 5 to as much as 7 times their In Progress number.
People with very slow systems & large caches or running multiple projects will have hardly any Pendings at all.
Grant
Darwin NT
ID: 1818586 · Report as offensive
1 · 2 · 3 · Next

Message boards : Number crunching : Linux boinc version 7.6.31 can't run optimized applications


 
©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.