Message boards : ATLAS application : No App config task limts
Message board moderation

To post messages, you must log in.

AuthorMessage
Toby Broom
Volunteer moderator

Send message
Joined: 27 Sep 08
Posts: 798
Credit: 644,822,440
RAC: 228,980
Message 29055 - Posted: 3 Mar 2017, 17:22:52 UTC

Hi,

I have the setting on the web to set the max number of cores to two, with 24 Tasks max, however this doesn't work well. BOINC doesn't respect or know the number of cores in the WU's so it tries to run 24 2 core WU on a 24 core machine so the performance so so.

Can this be addressed?
ID: 29055 · Report as offensive     Reply Quote
Profile Yeti
Volunteer moderator
Avatar

Send message
Joined: 2 Sep 04
Posts: 453
Credit: 193,369,412
RAC: 10,065
Message 29084 - Posted: 6 Mar 2017, 9:30:17 UTC - in response to Message 29055.  
Last modified: 6 Mar 2017, 9:30:39 UTC

BOINC doesn't respect or know the number of cores in the WU's so it tries to run 24 2 core WU on a 24 core machine so the performance so so.

Does it show "Running (2 CPUs)" in BOINC-Manager ?

Then it knows that these are 2-Core-WUs


Supporting BOINC, a great concept !
ID: 29084 · Report as offensive     Reply Quote
Toby Broom
Volunteer moderator

Send message
Joined: 27 Sep 08
Posts: 798
Credit: 644,822,440
RAC: 228,980
Message 29085 - Posted: 6 Mar 2017, 9:40:47 UTC - in response to Message 29084.  

Yes, it has the (2) after the task, it's strange
ID: 29085 · Report as offensive     Reply Quote
Profile Yeti
Volunteer moderator
Avatar

Send message
Joined: 2 Sep 04
Posts: 453
Credit: 193,369,412
RAC: 10,065
Message 29087 - Posted: 6 Mar 2017, 12:51:45 UTC - in response to Message 29085.  

Yes, it has the (2) after the task, it's strange

You could fix it for your machines with a small app_config.xml in the Project-Directory of LHCATHOME
<app_config>
 <app>
 <name>ATLAS</name>
 <max_concurrent>12</max_concurrent>
 </app>
 </app_config>



Supporting BOINC, a great concept !
ID: 29087 · Report as offensive     Reply Quote
Toby Broom
Volunteer moderator

Send message
Joined: 27 Sep 08
Posts: 798
Credit: 644,822,440
RAC: 228,980
Message 29094 - Posted: 7 Mar 2017, 0:01:18 UTC - in response to Message 29087.  

I'll give it ago
ID: 29094 · Report as offensive     Reply Quote
Toby Broom
Volunteer moderator

Send message
Joined: 27 Sep 08
Posts: 798
Credit: 644,822,440
RAC: 228,980
Message 29232 - Posted: 13 Mar 2017, 21:44:07 UTC

OK, I tried it and it works if there is all atlas, but in a mixed scenario it doesn't work.

The setting are 24 WU's if you have 10 Atlas and 14 others, this is 34 cores used on a 24core machine.

I guess it's the same as the other memory problems, BOINC doesn't actually know what the VM is doing so it doesn't allocate resources correctly.
ID: 29232 · Report as offensive     Reply Quote
Profile rbpeake

Send message
Joined: 17 Sep 04
Posts: 99
Credit: 30,619,757
RAC: 3,820
Message 34814 - Posted: 30 Mar 2018, 17:46:34 UTC - in response to Message 29087.  

[quote]
You could fix it for your machines with a small app_config.xml in the Project-Directory of LHCATHOME
<app_config>
 <app>
 <name>ATLAS</name>
 <max_concurrent>12</max_concurrent>
 </app>
 </app_config>

How do I combine this (or do I?) with the command to limit ATLAS to 2-cores per work unit?
Thanks!
Regards,
Bob P.
ID: 34814 · Report as offensive     Reply Quote
Erich56

Send message
Joined: 18 Dec 15
Posts: 1686
Credit: 100,440,738
RAC: 102,989
Message 34818 - Posted: 30 Mar 2018, 19:31:35 UTC - in response to Message 34814.  

How do I combine this (or do I?) with the command to limit ATLAS to 2-cores per work unit?
the number of cores per task you set in your web-settings ("preferences")
ID: 34818 · Report as offensive     Reply Quote
computezrmle
Volunteer moderator
Volunteer developer
Volunteer tester
Help desk expert
Avatar

Send message
Joined: 15 Jun 08
Posts: 2386
Credit: 222,996,280
RAC: 136,203
Message 34819 - Posted: 30 Mar 2018, 19:52:35 UTC - in response to Message 34814.  

... How do I combine this (or do I?) with the command to limit ATLAS to 2-cores per work unit?
Thanks!

See the following example that combines ATLAS 2core and CMS singlecore:
<app_config>
   <app>
      <name>ATLAS</name>
      <max_concurrent>1</max_concurrent>
   </app>
   <app_version>
      <app_name>ATLAS</app_name>
      <plan_class>vbox64_mt_mcore_atlas</plan_class>
      <avg_ncpus>2</avg_ncpus>
      <cmdline>--nthreads 2 --memory_size_mb 5000</cmdline>
   </app_version>
   <app>
      <name>CMS</name>
      <max_concurrent>1</max_concurrent>
   </app>
   <app_version>
      <app_name>CMS</app_name>
      <plan_class>vbox64</plan_class>
      <avg_ncpus>1</avg_ncpus>
      <cmdline>--nthreads 1</cmdline>
   </app_version>
   <project_max_concurrent>2</project_max_concurrent>
</app_config>

A complete explanation can be found in the BOINC documentation:
http://boinc.berkeley.edu/wiki/client_configuration#Application_configuration
ID: 34819 · Report as offensive     Reply Quote
Profile rbpeake

Send message
Joined: 17 Sep 04
Posts: 99
Credit: 30,619,757
RAC: 3,820
Message 34820 - Posted: 30 Mar 2018, 22:18:04 UTC - in response to Message 34819.  

Thanks!
The problem I have is that 11 instances of ATLAS 2-core download, but then only 7 of them run. With my allocating 22 cores to LHC, the 4 ATLAS 2-cores that are there but not running prevent other LHC work units from running. So i an stuck with just 14 of my 22 cores running.
Regards,
Bob P.
ID: 34820 · Report as offensive     Reply Quote

Message boards : ATLAS application : No App config task limts


©2024 CERN