← Back to team overview

yade-users team mailing list archive

Re: [Question #240608]: How to increase the computing scale in YADE

 

Question #240608 on Yade changed:
https://answers.launchpad.net/yade/+question/240608

    Status: Open => Answered

Alexander Eulitz [Eugen] proposed the following answer:
Hello Fu zuoguang,
hope I can help you a bit.
I think that you miss one thing. The number of jobs that will be done is definied in your batch-table [1].
Look at [1] for examples. Basically each line in the table corresponds to one job.
Moreover your simulation script has to be changed in comparison to running it not-in-batch (stand-alone).
You'll find all inforamtions about that in [1].

according (1) of your question:
If you want 1 job with n threads, you don’t need batch mode. this can be done with 
yade -jN script.py

(2) yes, you'r right
(3) I hope I understood your question. If all of your jobs shall use the same number of threads each then it is sufficient doing as klaus desribed with --job-threads=2. That means that each job will get 2 threads. The number of jobs that run simulataneously is defined by the total number of threads that you allow by -j option.
E.g. j=10 and you set --job-threads=2 using a simulation script.py and a batch table myBatch.table
yade-batch -j10 --job-threads=2 myBatch.table script.py

in your table file you define 20 jobs (each for instance with different
parameters in a biaxial compression). Consequently 5 jobs will run
simulataneously (2 threads per job using 10 threads in total). After one
job finished the next job starts using again 2 threads.

Hope it helps,
Alex

[1] https://yade-dem.org/doc/user.html?highlight=batch#batch-queuing-
and-execution-yade-batch

-----Ursprüngliche Nachricht-----
Von: bounces@xxxxxxxxxxxxx [mailto:bounces@xxxxxxxxxxxxx] Im Auftrag von Fu zuoguang
Gesendet: Montag, 24. Februar 2014 08:36
An: Eulitz, Alexander
Betreff: Re: [Question #240608]: How to increase the computing scale in YADE

Question #240608 on Yade changed:
https://answers.launchpad.net/yade/+question/240608

    Status: Answered => Open

Fu zuoguang is still having a problem:
Dear Bruno Chareyre and all users:
    Thanks for the help! YADE batch mode, as I know, is just for running n jobs at the same time using N threads( n<=N ). Am I right? If yes, I then have some noticeable questions to ask for your advice, which are:

(1). If I just have only one job to run, just like that day, I just need
to run biaxial compression test simulation, my question can be described
as only 1 job using N threads. Can YADE batch mode help it to get
sucessfully a high speed running? If no, is this mean that YADE batch
mode can bring us a competitive running mode when the running tasks are
more than one?

(2). If the answer of the above is yes and I start my running just in
YADE, not in yade batch, the default option of thread used by this
running is just one, am I right?

(3). The method of assigning determined threads for jobs. I have gotten
information from Klaus Thoeni (klaus.thoeni) said on 2014-02-20:

yade-batch -j1 --job-threads=2 file.table file.py: should still use one
core, but I never used it yade-batch -j2 --job-threads=2 file.table
file.py: one simulation at a time with 2 cores yade-batch -j4 --job-
threads=2 file.table file.py: two simulations at a time with 2 cores
each yade-batch -j16 --job-threads=1 file.table file.py: 16 simulations
at a time with 1 core each

In above description, j means the number of jobs and job-threads means
the total number of threads for all the jobs. Is this right? But I can
not understand why 'two simulations at a time with 2 cores each' in the
third row.

(4).I have tried this way for my own simulation. It is as :
yadedaily-batch -j1 --job-threads=7 file.py

But yade gave me an error warning like that:

Running 1 stand-alone simulation(s) in batch mode.
WARNING:root:WARNING: job #0 will use 7 cores but only 1 are available WARNING:root:Job #0 will be uselessly run with 7 threads (compiled without OpenMP support).
Master process pid 14997

How to get these problems ? Seeking your help!

--
You received this question notification because you are a direct subscriber of the question.

-- 
You received this question notification because you are a member of
yade-users, which is an answer contact for Yade.