Message boards : Theory Application : Theory native - Runtime Pitfall
Message board moderation

To post messages, you must log in.

AuthorMessage
computezrmle
Volunteer moderator
Project tester
Volunteer developer
Volunteer tester
Help desk expert
Avatar

Send message
Joined: 28 Jul 16
Posts: 467
Credit: 379,047
RAC: 0
Message 5856 - Posted: 11 Feb 2019, 20:49:23 UTC

The different runtimes of the scientific apps most likely cause a pitfall regarding the use in a production environment, especially if volunteers also participate in other projects.


Reason
Out of experience (e.g. SixTrack) we know that variable runtimes/CPU-times do not only result in credit variations (much too low <--> much too high).
They also result in a wrong runtime estimation. This is a minor problem if the estimated runtime is too long as the BOINC client would simply request fewer tasks.
It could become a major problem if the estimated runtime is far too low as the BOINC client would request lots of tasks that will hardly/never meet their due dates or block other tasks/projects.


Measure
As the <rsc_fpops_est> value is a major factor used to calculate the estimated runtime, this value must be set by the server as precise as possible.
This has to be done individually for each scientific app, e.g. Herwig, Pythia, Sherpa ...
ID: 5856 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Laurence
Project administrator
Project developer
Project tester
Avatar

Send message
Joined: 12 Sep 14
Posts: 1064
Credit: 319,055
RAC: 49
Message 5862 - Posted: 12 Feb 2019, 8:39:27 UTC - in response to Message 5856.  


Measure
As the value is a major factor used to calculate the estimated runtime, this value must be set by the server as precise as possible.
This has to be done individually for each scientific app, e.g. Herwig, Pythia, Sherpa ...


This should be possible.
ID: 5862 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Laurence
Project administrator
Project developer
Project tester
Avatar

Send message
Joined: 12 Sep 14
Posts: 1064
Credit: 319,055
RAC: 49
Message 5864 - Posted: 12 Feb 2019, 9:52:42 UTC - in response to Message 5856.  

Just for reference, here is a list of job types that I found.
epos
herwig++
herwig7
herwig++powheg
madgraph5amc
phojet
pythia6
pythia8
sherpa
vincia
ID: 5864 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
computezrmle
Volunteer moderator
Project tester
Volunteer developer
Volunteer tester
Help desk expert
Avatar

Send message
Joined: 28 Jul 16
Posts: 467
Credit: 379,047
RAC: 0
Message 5865 - Posted: 12 Feb 2019, 10:01:39 UTC - in response to Message 5864.  

Just for reference, here is a list of job types that I found.

agile-runmc??
It's currently running on one of my VMs.
ID: 5865 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
maeax

Send message
Joined: 22 Apr 16
Posts: 651
Credit: 1,678,202
RAC: 7
Message 5866 - Posted: 12 Feb 2019, 10:09:22 UTC - in response to Message 5865.  

ID: 5866 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Crystal Pellet
Volunteer tester

Send message
Joined: 13 Feb 15
Posts: 1167
Credit: 785,507
RAC: 664
Message 5867 - Posted: 12 Feb 2019, 10:57:53 UTC
Last modified: 12 Feb 2019, 11:05:20 UTC

What happens when one catch an endless running Sherpa job?

Will it be killed after xx hours runtime? -- without granting credits ;-(
ID: 5867 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote

Message boards : Theory Application : Theory native - Runtime Pitfall


©2024 CERN